tag:old.nabble.com,2006:forum-1502Nabble - Gnu - Bison2009-12-21T00:53:52ZBison is a general-purpose parser generator that converts a grammar description for an LALR context-free grammar into a C program to parse that grammar. Once you are proficient with Bison, you can use it to develop a wide range of language parsers, from those used in simple desk calculators to complex programming languages. Gnu - Bison home is <a href="http://www.gnu.org/software/bison/bison.html" target="_top" rel="nofollow">here</a>.tag:old.nabble.com,2006:post-26871564Re: remove YYFAIL from lalr1.java and yacc.c2009-12-21T00:53:52Z2009-12-21T00:53:52ZPaolo Bonzini-2<br>>> , and I think that
<br>>> breaking YYFAIL now for %error-verbose is fine too.
<br>>
<br>> Google code search turns up an instance where both are used:
<br>>
<br>> <a href="http://www.google.com/codesearch?hl=en&lr=&q=%25error-verbose+YYFAIL+file%3A" target="_top" rel="nofollow">http://www.google.com/codesearch?hl=en&lr=&q=%25error-verbose+YYFAIL+file%3A</a>\.y&sbtn=Search
<br><br>Hmm, right. They're not relying on the semantics, though and they can
<br>simply fix it by removing the "|| defined (YYFAIL)". So it's
<br>effectively a false positive of the warning. Well, let's live with it.
<br><br>> I agree we may have to keep it for a long time, but hopefully the warning
<br>> will prompt users to explore at least why YYFAIL is deprecated. Maybe I
<br>> should make the warning even stronger by saying that YYFAIL will
<br>> eventually be removed.
<br><br>Fine, even if both of us know how likely that is to happen. :-)
<br><br>Paolo
<br><br><br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Patches-f1505.html" embed="fixTarget[1505]" target="_top" >Gnu - Bison - Patches</a></p>tag:old.nabble.com,2006:post-26871525Re: remove YYFAIL from lalr1.java and yacc.c2009-12-21T00:42:24Z2009-12-21T00:42:24ZJoel E. Denny-4On Mon, 21 Dec 2009, Paolo Bonzini wrote:
<br><br>> > I'd like to push the first patch to master, branch-2.5, and branch-2.4.2
<br>> > in order to formally deprecate YYFAIL.
<br>>
<br>> I'm fine with removing it at once from the Java parser
<br><br>OK, I'll make that adjustment.
<br><br>> , and I think that
<br>> breaking YYFAIL now for %error-verbose is fine too.
<br><br>Google code search turns up an instance where both are used:
<br><br> <a href="http://www.google.com/codesearch?hl=en&lr=&q=%25error-verbose+YYFAIL+file%3A" target="_top" rel="nofollow">http://www.google.com/codesearch?hl=en&lr=&q=%25error-verbose+YYFAIL+file%3A</a>\.y&sbtn=Search
<br><br>I'm afraid there may be more instances that are not online.
<br><br>> However, I'm afraid that
<br>> without verbose errors we should keep it indefinitely though undocumented, so
<br>> there's probably no point in warning.
<br><br>I agree we may have to keep it for a long time, but hopefully the warning
<br>will prompt users to explore at least why YYFAIL is deprecated. Maybe I
<br>should make the warning even stronger by saying that YYFAIL will
<br>eventually be removed.
<br><br><br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Patches-f1505.html" embed="fixTarget[1505]" target="_top" >Gnu - Bison - Patches</a></p>tag:old.nabble.com,2006:post-26870641Re: remove YYFAIL from lalr1.java and yacc.c2009-12-20T23:19:20Z2009-12-20T23:19:20ZPaolo Bonzini-2> I'd like to push the first patch to master, branch-2.5, and branch-2.4.2
<br>> in order to formally deprecate YYFAIL.
<br><br>I'm fine with removing it at once from the Java parser, and I think that
<br>breaking YYFAIL now for %error-verbose is fine too. However, I'm afraid
<br>that without verbose errors we should keep it indefinitely though
<br>undocumented, so there's probably no point in warning.
<br><br>Paolo
<br><br><br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Patches-f1505.html" embed="fixTarget[1505]" target="_top" >Gnu - Bison - Patches</a></p>tag:old.nabble.com,2006:post-26869711Re: What is wrong with this simple example?2009-12-20T19:48:59Z2009-12-20T19:48:59ZJoel E. Denny-4On Sun, 20 Dec 2009, Peng Yu wrote:
<br><div class='shrinkable-quote'><br>> I try to run the example on
<br>> <a href="http://tldp.org/HOWTO/Lex-YACC-HOWTO-4.html" target="_top" rel="nofollow">http://tldp.org/HOWTO/Lex-YACC-HOWTO-4.html</a>. But I get the following
<br>> error. Could somebody let me know what I am wrong?
<br>>
<br>> $ bison -d example4.y
<br>> example4.y:24.1-8: syntax error, unexpected identifier:
<br>> $ cat example4.y
<br>> %{
<br>> #include <stdio.h>
<br>> #include <string.h>
<br>>
<br>> void yyerror(const char *str)
<br>> {
<br>> fprintf(stderr,"error: %s\n",str);
<br>> }
<br>>
<br>> int yywrap()
<br>> {
<br>> return 1;
<br>> }
<br>>
<br>> main()
<br>> {
<br>> yyparse();
<br>> }
<br>>
<br>> %}
<br>>
<br>> %token NUMBER TOKHEAT STATE TOKTARGET TOKTEMPERATURE
</div><br>You need this here:
<br><br> %%
<br><div class='shrinkable-quote'><br>> commands: /* empty */
<br>> | commands command
<br>> ;
<br>>
<br>> command:
<br>> heat_switch
<br>> |
<br>> target_set
<br>> ;
<br>>
<br>> heat_switch:
<br>> TOKHEAT STATE
<br>> {
<br>> printf("\tHeat turned on or off\n");
<br>> }
<br>> ;
<br>>
<br>> target_set:
<br>> TOKTARGET TOKTEMPERATURE NUMBER
<br>> {
<br>> printf("\tTemperature set\n");
<br>> }
<br>> ;
</div><br><br>_______________________________________________
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26869711&i=0" target="_top" rel="nofollow">help-bison@...</a> <a href="http://lists.gnu.org/mailman/listinfo/help-bison" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/help-bison</a><br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Help-f1504.html" embed="fixTarget[1504]" target="_top" >Gnu - Bison - Help</a></p>tag:old.nabble.com,2006:post-26869689What is wrong with this simple example?2009-12-20T19:42:58Z2009-12-20T19:42:58ZPeng YuI try to run the example on
<br><a href="http://tldp.org/HOWTO/Lex-YACC-HOWTO-4.html" target="_top" rel="nofollow">http://tldp.org/HOWTO/Lex-YACC-HOWTO-4.html</a>. But I get the following
<br>error. Could somebody let me know what I am wrong?
<br><br>$ bison -d example4.y
<br>example4.y:24.1-8: syntax error, unexpected identifier:
<br>$ cat example4.y
<br>%{
<br>#include <stdio.h>
<br>#include <string.h>
<br><br>void yyerror(const char *str)
<br>{
<br> fprintf(stderr,"error: %s\n",str);
<br>}
<br><br>int yywrap()
<br>{
<br> return 1;
<br>}
<br><br>main()
<br>{
<br> yyparse();
<br>}
<br><br>%}
<br><br>%token NUMBER TOKHEAT STATE TOKTARGET TOKTEMPERATURE
<br><br>commands: /* empty */
<br> | commands command
<br> ;
<br><br>command:
<br> heat_switch
<br> |
<br> target_set
<br> ;
<br><br>heat_switch:
<br> TOKHEAT STATE
<br> {
<br> printf("\tHeat turned on or off\n");
<br> }
<br> ;
<br><br>target_set:
<br> TOKTARGET TOKTEMPERATURE NUMBER
<br> {
<br> printf("\tTemperature set\n");
<br> }
<br> ;
<br><br><br>_______________________________________________
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26869689&i=0" target="_top" rel="nofollow">help-bison@...</a> <a href="http://lists.gnu.org/mailman/listinfo/help-bison" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/help-bison</a><br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Help-f1504.html" embed="fixTarget[1504]" target="_top" >Gnu - Bison - Help</a></p>tag:old.nabble.com,2006:post-26867313Re: [PATCH] lalr1.cc: don't discard non-existent lookahead on syntax error.2009-12-20T13:07:45Z2009-12-20T13:07:45ZJoel E. Denny-4On Sat, 19 Dec 2009, Joel E. Denny wrote:
<br><div class='shrinkable-quote'><br>> I pushed this to master. The bug does not exist on branch-2.5. The test
<br>> case makes this bug look pretty obscure, and maybe that's right, but I
<br>> really didn't spend much time searching for a simpler test case.
<br>>
<br>> >From d59beda068aef97834ea84ce4f5cc99b487e4b68 Mon Sep 17 00:00:00 2001
<br>> From: Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26867313&i=0" target="_top" rel="nofollow">jdenny@...</a>>
<br>> Date: Fri, 18 Dec 2009 23:57:18 -0500
<br>> Subject: [PATCH] lalr1.cc: don't discard non-existent lookahead on syntax error.
<br>>
<br>> * data/lalr1.cc (parser::parse): Check yyempty first.
<br>> * tests/c++.at (Syntax error discarding no lookahead): New test
<br>> group.
</div><br>I pushed the following to master.
<br><br>From 5335b65a79189afbbd27ecb55d76e74a4fe31863 Mon Sep 17 00:00:00 2001
<br>From: Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26867313&i=1" target="_top" rel="nofollow">jdenny@...</a>>
<br>Date: Sun, 20 Dec 2009 16:03:07 -0500
<br>Subject: [PATCH] tests: cleanup.
<br><br>* tests/c++.at (Syntax error discarding no lookahead): Don't
<br>ignore stderr. Instead, eliminate remaining warnings.
<br>---
<br> ChangeLog | 6 ++++++
<br> tests/c++.at | 6 +++++-
<br> 2 files changed, 11 insertions(+), 1 deletions(-)
<br><br>diff --git a/ChangeLog b/ChangeLog
<br>index 1c15cc5..e984569 100644
<br>--- a/ChangeLog
<br>+++ b/ChangeLog
<br>@@ -1,3 +1,9 @@
<br>+2009-12-20 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26867313&i=2" target="_top" rel="nofollow">jdenny@...</a>>
<br>+
<br>+ tests: cleanup.
<br>+ * tests/c++.at (Syntax error discarding no lookahead): Don't
<br>+ ignore stderr. Instead, eliminate remaining warnings.
<br>+
<br> 2009-12-18 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26867313&i=3" target="_top" rel="nofollow">jdenny@...</a>>
<br>
<br> lalr1.cc: don't discard non-existent lookahead on syntax error.
<br>diff --git a/tests/c++.at b/tests/c++.at
<br>index ce64d6d..6a8d2c5 100644
<br>--- a/tests/c++.at
<br>+++ b/tests/c++.at
<br>@@ -412,6 +412,10 @@ consistent-error:
<br> | /*empty*/ %prec 'a'
<br> ;
<br>
<br>+// Provide another context in which all rules are useful so that this
<br>+// test case looks a little more realistic.
<br>+start: 'b' consistent-error ;
<br>+
<br> %%
<br>
<br> int
<br>@@ -434,7 +438,7 @@ main (void)
<br> return parser.parse ();
<br> }
<br> ]])
<br>-AT_BISON_CHECK([[-o input.cc input.yy]], [[0]], [[]], [[ignore-nolog]])
<br>+AT_BISON_CHECK([[-o input.cc input.yy]])
<br> AT_COMPILE_CXX([[input]])
<br> # This used to print "Discarding 'a'." again at the end.
<br> AT_PARSER_CHECK([[./input]], [[1]], [[]],
<br>--
<br>1.5.4.3
<br><br><br><br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Patches-f1505.html" embed="fixTarget[1505]" target="_top" >Gnu - Bison - Patches</a></p>tag:old.nabble.com,2006:post-26866413Re: remove YYFAIL from lalr1.java and yacc.c2009-12-20T11:18:38Z2009-12-20T11:18:38ZJoel E. Denny-4Hi Paolo,
<br><br>On Sun, 20 Dec 2009, Paolo Bonzini wrote:
<br><br>> No problems for me for lalr1.java; for yacc.c however we have some uses:
<br>>
<br>> <a href="http://www.google.com/codesearch?hl=en&lr=&q=YYFAIL%3B&sbtn=Search" target="_top" rel="nofollow">http://www.google.com/codesearch?hl=en&lr=&q=YYFAIL%3B&sbtn=Search</a><br>>
<br>> Some of them are my own GNU Smalltalk (no problem, I now switched to recursive
<br>> descent) and of course GCC 1. However there are still about 5-6 other users.
<br>> I don't think it's reasonably possible for YYFAIL to die anytime soon.
<br><br>Thanks very much for that info.
<br><br>Here are a couple of patches to begin to phase out YYFAIL. Based on your
<br>research, the process will have to be very slow.
<br><br>I'd like to push the first patch to master, branch-2.5, and branch-2.4.2
<br>in order to formally deprecate YYFAIL.
<br><br>I'd like to push the second patch to master and branch-2.5 in order to
<br>generate warnings about uses of YYFAIL. I always worry that attempts to
<br>decipher literal code from users is error-prone. I'd appreciate any
<br>corrections or other comments.
<br><br>From 2ffce5412c9436eefc18b7198b31dce6e1c20e08 Mon Sep 17 00:00:00 2001
<br>From: Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26866413&i=0" target="_top" rel="nofollow">jdenny@...</a>>
<br>Date: Sun, 20 Dec 2009 13:38:26 -0500
<br>Subject: [PATCH] YYFAIL: deprecate.
<br><br>* NEWS (2.4.2): Document deprecation.
<br>* data/lalr1.java (YYStack::YYFAIL): Add comment about
<br>deprecation.
<br>* data/yacc.c (YYFAIL): Likewise.
<br>* doc/bison.texinfo (Java Action Features): Remove YYFAIL
<br>documentation.
<br>(LocalWords): Remove YYFAIL.
<br>---
<br> ChangeLog | 11 +++++++++++
<br> NEWS | 27 +++++++++++++++++++++++++++
<br> data/lalr1.java | 3 ++-
<br> data/yacc.c | 5 ++++-
<br> doc/bison.texinfo | 7 +------
<br> 5 files changed, 45 insertions(+), 8 deletions(-)
<br><br>diff --git a/ChangeLog b/ChangeLog
<br>index 1c15cc5..aed5998 100644
<br>--- a/ChangeLog
<br>+++ b/ChangeLog
<br>@@ -1,3 +1,14 @@
<br>+2009-12-20 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26866413&i=1" target="_top" rel="nofollow">jdenny@...</a>>
<br>+
<br>+ YYFAIL: deprecate.
<br>+ * NEWS (2.4.2): Document deprecation.
<br>+ * data/lalr1.java (YYStack::YYFAIL): Add comment about
<br>+ deprecation.
<br>+ * data/yacc.c (YYFAIL): Likewise.
<br>+ * doc/bison.texinfo (Java Action Features): Remove YYFAIL
<br>+ documentation.
<br>+ (LocalWords): Remove YYFAIL.
<br>+
<br> 2009-12-18 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26866413&i=2" target="_top" rel="nofollow">jdenny@...</a>>
<br>
<br> lalr1.cc: don't discard non-existent lookahead on syntax error.
<br>diff --git a/NEWS b/NEWS
<br>index 92567dd..7e8bd9f 100644
<br>--- a/NEWS
<br>+++ b/NEWS
<br>@@ -243,6 +243,33 @@ Bison News
<br> Bison's Java feature as a whole including its current usage of %code
<br> is still considered experimental.
<br>
<br>+** YYFAIL is deprecated.
<br>+
<br>+ YYFAIL has existed for many years as an undocumented feature of
<br>+ deterministic parsers in C generated by Bison. Previously, it was
<br>+ documented for Bison's experimental Java parsers. YYFAIL is no longer
<br>+ documented for Java parsers and is formally deprecated in both cases.
<br>+ Users are strongly encouraged to migrate to YYERROR, which has been
<br>+ documented for many years.
<br>+
<br>+ Like YYERROR, you can invoke YYFAIL from a semantic action in order to
<br>+ induce a syntax error. The most obvious difference from YYERROR is
<br>+ that YYFAIL will automatically invoke yyerror to report the syntax
<br>+ error so that you don't have to. However, there are several other
<br>+ subtle differences between YYERROR and YYFAIL, and YYFAIL suffers from
<br>+ inherent flaws when %error-verbose is used. For a more detailed
<br>+ discussion, see:
<br>+
<br>+ <a href="http://lists.gnu.org/archive/html/bison-patches/2009-12/msg00024.html" target="_top" rel="nofollow">http://lists.gnu.org/archive/html/bison-patches/2009-12/msg00024.html</a><br>+
<br>+ YYFAIL will continue to be maintained for deterministic parsers in C
<br>+ and Java for the immediate future. However, because YYFAIL is already
<br>+ flawed, it would be futile to try to make new Bison features
<br>+ compatible with it. During parser generation, some future release of
<br>+ Bison will begin to produce warnings whenever it discovers YYFAIL in a
<br>+ semantic action. In a later release, YYFAIL will be disabled for
<br>+ %error-verbose. Eventually, it will be removed altogether.
<br>+
<br> ** Internationalization.
<br>
<br> Fix a regression introduced in Bison 2.4: Under some circumstances,
<br>diff --git a/data/lalr1.java b/data/lalr1.java
<br>index 6637c3a..a9ae873 100644
<br>--- a/data/lalr1.java
<br>+++ b/data/lalr1.java
<br>@@ -352,7 +352,8 @@ b4_lexer_if([[
<br>
<br> /**
<br> * Returned by a Bison action in order to print an error message and start
<br>- * error recovery. */
<br>+ * error recovery. Formally deprecated in Bison 2.4.2's NEWS entry, where
<br>+ * a plan to phase it out is discussed. */
<br> public static final int YYFAIL = 3;
<br>
<br> private static final int YYNEWSTATE = 4;
<br>diff --git a/data/yacc.c b/data/yacc.c
<br>index 11ebbf1..6e74620 100644
<br>--- a/data/yacc.c
<br>+++ b/data/yacc.c
<br>@@ -609,7 +609,10 @@ static const ]b4_int_type_for([b4_toknum])[ yytoknum[] =
<br>
<br> /* Like YYERROR except do call yyerror. This remains here temporarily
<br> to ease the transition to the new meaning of YYERROR, for GCC.
<br>- Once GCC version 2 has supplanted version 1, this can go. */
<br>+ Once GCC version 2 has supplanted version 1, this can go. However,
<br>+ YYFAIL appears to be in use. Nevertheless, it is formally deprecated
<br>+ in Bison 2.4.2's NEWS entry, where a plan to phase it out is
<br>+ discussed. */
<br>
<br> #define YYFAIL goto yyerrlab
<br>
<br>diff --git a/doc/bison.texinfo b/doc/bison.texinfo
<br>index 2af8917..ab2cd6c 100644
<br>--- a/doc/bison.texinfo
<br>+++ b/doc/bison.texinfo
<br>@@ -9925,11 +9925,6 @@ Start error recovery without printing an error message.
<br> @xref{Error Recovery}.
<br> @end deffn
<br>
<br>-@deffn {Statement} {return YYFAIL;}
<br>-Print an error message and start error recovery.
<br>-@xref{Error Recovery}.
<br>-@end deffn
<br>-
<br> @deftypefn {Function} {boolean} recovering ()
<br> Return whether error recovery is being done. In this state, the parser
<br> reads token until it reaches a known state, and then restarts normal
<br>@@ -11330,5 +11325,5 @@ grammatically indivisible. The piece of text it represents is a token.
<br> @c LocalWords: superclasses boolean getErrorVerbose setErrorVerbose deftypecv
<br> @c LocalWords: getDebugStream setDebugStream getDebugLevel setDebugLevel url
<br> @c LocalWords: bisonVersion deftypecvx bisonSkeleton getStartPos getEndPos
<br>-@c LocalWords: getLVal defvar YYFAIL deftypefn deftypefnx gotos msgfmt
<br>+@c LocalWords: getLVal defvar deftypefn deftypefnx gotos msgfmt
<br> @c LocalWords: subdirectory Solaris nonassociativity
<br>--
<br>1.5.4.3
<br><br><br>From ceba3986c7b4538001cbfa133a311b5662da158d Mon Sep 17 00:00:00 2001
<br>From: Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26866413&i=3" target="_top" rel="nofollow">jdenny@...</a>>
<br>Date: Sun, 20 Dec 2009 13:47:51 -0500
<br>Subject: [PATCH] YYFAIL: warn about uses.
<br><br>* NEWS (2.5): Document.
<br>* src/scan-gram.l (SC_BRACED_CODE): Implement.
<br>---
<br> ChangeLog | 6 ++++++
<br> NEWS | 8 ++++++++
<br> src/scan-gram.l | 8 ++++++++
<br> 3 files changed, 22 insertions(+), 0 deletions(-)
<br><br>diff --git a/ChangeLog b/ChangeLog
<br>index aed5998..53dc2c4 100644
<br>--- a/ChangeLog
<br>+++ b/ChangeLog
<br>@@ -1,5 +1,11 @@
<br> 2009-12-20 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26866413&i=4" target="_top" rel="nofollow">jdenny@...</a>>
<br>
<br>+ YYFAIL: warn about uses.
<br>+ * NEWS (2.5): Document.
<br>+ * src/scan-gram.l (SC_BRACED_CODE): Implement.
<br>+
<br>+2009-12-20 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26866413&i=5" target="_top" rel="nofollow">jdenny@...</a>>
<br>+
<br> YYFAIL: deprecate.
<br> * NEWS (2.4.2): Document deprecation.
<br> * data/lalr1.java (YYStack::YYFAIL): Add comment about
<br>diff --git a/NEWS b/NEWS
<br>index 7e8bd9f..6b11958 100644
<br>--- a/NEWS
<br>+++ b/NEWS
<br>@@ -169,6 +169,14 @@ Bison News
<br> POSIX Yacc whose use is reported by -Wyacc, and rejected in Yacc
<br> mode (--yacc).
<br>
<br>+** YYFAIL is deprecated and produces warnings.
<br>+
<br>+ YYFAIL has existed for many years as an undocumented feature of
<br>+ deterministic parsers in C generated by Bison. Previously, it was
<br>+ documented for Bison's experimental Java parsers. As discussed in
<br>+ Bison 2.4.2's NEWS entry, YYFAIL has been formally deprecated. Now,
<br>+ any appearance of YYFAIL in a semantic action also produces a warning.
<br>+
<br> ** Temporary hack for adding a semicolon to the user action.
<br>
<br> Previously, Bison appended a semicolon to every user action for
<br>diff --git a/src/scan-gram.l b/src/scan-gram.l
<br>index aff0769..0971e3f 100644
<br>--- a/src/scan-gram.l
<br>+++ b/src/scan-gram.l
<br>@@ -717,6 +717,14 @@ splice (\\[ \f\t\v]*\n)*
<br> (as `<' `<%'). */
<br> "<"{splice}"<" STRING_GROW;
<br>
<br>+ /* YYFAIL is undocumented and was formally deprecated in Bison
<br>+ 2.4.2. */
<br>+ "YYFAIL" STRING_GROW; warn_at (*loc, "YYFAIL is deprecated");
<br>+
<br>+ /* The sole purpose of this is to make sure identifiers that merely
<br>+ contain YYFAIL don't produce the above warning. */
<br>+ [A-Za-z_][0-9A-Za-z_]* STRING_GROW;
<br>+
<br> <<EOF>> {
<br> unexpected_eof (code_start, "}");
<br> STRING_FINISH;
<br>--
<br>1.5.4.3
<br><br><br><br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Patches-f1505.html" embed="fixTarget[1505]" target="_top" >Gnu - Bison - Patches</a></p>tag:old.nabble.com,2006:post-26862272Re: remove YYFAIL from lalr1.java and yacc.c2009-12-20T02:27:08Z2009-12-20T02:27:08ZPaolo Bonzini-2On 12/19/2009 10:41 PM, Joel E. Denny wrote:
<div class='shrinkable-quote'><br>> a. YYFAIL appears in yacc.c but is not documented for yacc.c. Comments in
<br>> yacc.c describe it as follows:
<br>>
<br>> /* Like YYERROR except do call yyerror. This remains here temporarily
<br>> to ease the transition to the new meaning of YYERROR, for GCC.
<br>> Once GCC version 2 has supplanted version 1, this can go. */
<br>>
<br>> In other words, we've been planning to get rid of it for a long while.
<br>> (However, is there any evidence that users actually use it? Do we need to
<br>> poll them or warn them somehow?)
</div><br>No problems for me for lalr1.java; for yacc.c however we have some uses:
<br><br><a href="http://www.google.com/codesearch?hl=en&lr=&q=YYFAIL%3B&sbtn=Search" target="_top" rel="nofollow">http://www.google.com/codesearch?hl=en&lr=&q=YYFAIL%3B&sbtn=Search</a><br><br>Some of them are my own GNU Smalltalk (no problem, I now switched to
<br>recursive descent) and of course GCC 1. However there are still about
<br>5-6 other users. I don't think it's reasonably possible for YYFAIL to
<br>die anytime soon.
<br><br>Paolo
<br><br><br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Patches-f1505.html" embed="fixTarget[1505]" target="_top" >Gnu - Bison - Patches</a></p>tag:old.nabble.com,2006:post-26861692Re: [PATCH] Fix handling of yychar manipulation in user semantic actions.2009-12-20T00:17:03Z2009-12-20T00:17:03ZJoel E. Denny-4On Wed, 16 Dec 2009, Joel E. Denny wrote:
<br><div class='shrinkable-quote'><br>> A few months ago, I started talking about a lookahead correction mechanism
<br>> I was working on for Bison. Its most obvious benefit is to correct the
<br>> list of expected tokens in verbose syntax error messages. I'm finally
<br>> getting back to pushing that patch series through.
<br>>
<br>> The following patch is next. The problem it addresses is that users
<br>> sometimes write semantic actions that manipulate yychar. Of course, they
<br>> can do this with YYBACKUP or yyclearin. However, my impression is that
<br>> many users alter yychar directly. In this case, the translation in
<br>> yytoken isn't always updated, and bad destructor calls or syntax error
<br>> messages can result. This patch fixes that.
<br>>
<br>> I'd like to push this patch to master and branch-2.5. Any feedback would
<br>> be appreciated. I'd especially like to hear comments on my analysis of
<br>> glr.c, lalr1.cc, and lalr1.java.
<br>>
<br>> Don't worry about the unfixed bug I mention in the test suite. The next
<br>> patch will more thoroughly explain it and fix it.
<br>>
<br>> >From cfdf80e9e688706a449dc68662e5e4ad0674ab01 Mon Sep 17 00:00:00 2001
<br>> From: Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26861692&i=0" target="_top" rel="nofollow">jdenny@...</a>>
<br>> Date: Wed, 16 Dec 2009 20:55:40 -0500
<br>> Subject: [PATCH] Fix handling of yychar manipulation in user semantic actions.
</div><br>I botched the last test group. It exercised the lookahead destructor at
<br>parser return rather than the one during syntax error recovery. Here's a
<br>revised patch. I also extended the ChangeLog's discussion of lalr1.cc a
<br>little.
<br><br>From 575e4443e4993046e0b812ea1ef5ef3cbd4480a0 Mon Sep 17 00:00:00 2001
<br>From: Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26861692&i=1" target="_top" rel="nofollow">jdenny@...</a>>
<br>Date: Sun, 20 Dec 2009 02:22:53 -0500
<br>Subject: [PATCH] Fix handling of yychar manipulation in user semantic actions.
<br><br>The problem was that yacc.c didn't always update the yychar
<br>translation afterwards. However, other skeletons appear to be
<br>fine. glr.c appears to already translate yychar before every
<br>use. lalr1.cc does not define yychar and does not document its
<br>replacement, yyla, for users. It does provide yyclearin, but
<br>that does not manipulate yyla and thus requires no translation
<br>update. In lalr1.java, yychar is out of scope during semantic
<br>actions.
<br>* NEWS (2.5): Document.
<br>* data/yacc.c (YYBACKUP): Don't bother translating yychar into
<br>yytoken here.
<br>(yyparse, yypush_parse): Instead, translate before every use of
<br>yytoken, and add comments explaining this approach.
<br>* tests/actions.at (Destroying lookahead assigned by semantic
<br>action): New test group checking that translation happens before
<br>lookahead destructor calls at parser return. Previously,
<br>incorrect destructors were called.
<br>* tests/conflicts.at (parse.error=verbose and consistent
<br>errors): New test group checking that translation happens at
<br>syntax error detection before the associated verbose error
<br>message and the associated lookahead destructor calls. While
<br>the destructor call is fixed by this patch, the verbose error
<br>message is currently incorrect due to another bug (see
<br>comments in test group), so this is an expected failure for now.
<br>---
<br> ChangeLog | 28 +++
<br> NEWS | 7 +
<br> data/yacc.c | 25 +++-
<br> src/parse-gram.c | 522 +++++++++++++++++++++++-----------------------------
<br> src/parse-gram.h | 68 +------
<br> tests/actions.at | 71 +++++++
<br> tests/conflicts.at | 124 +++++++++++++
<br> 7 files changed, 491 insertions(+), 354 deletions(-)
<br><br>diff --git a/ChangeLog b/ChangeLog
<br>index 1c15cc5..c53baca 100644
<br>--- a/ChangeLog
<br>+++ b/ChangeLog
<br>@@ -1,3 +1,31 @@
<br>+2009-12-20 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26861692&i=2" target="_top" rel="nofollow">jdenny@...</a>>
<br>+
<br>+ Fix handling of yychar manipulation in user semantic actions.
<br>+ The problem was that yacc.c didn't always update the yychar
<br>+ translation afterwards. However, other skeletons appear to be
<br>+ fine. glr.c appears to already translate yychar before every
<br>+ use. lalr1.cc does not define yychar and does not document its
<br>+ replacement, yyla, for users. It does provide yyclearin, but
<br>+ that does not manipulate yyla and thus requires no translation
<br>+ update. In lalr1.java, yychar is out of scope during semantic
<br>+ actions.
<br>+ * NEWS (2.5): Document.
<br>+ * data/yacc.c (YYBACKUP): Don't bother translating yychar into
<br>+ yytoken here.
<br>+ (yyparse, yypush_parse): Instead, translate before every use of
<br>+ yytoken, and add comments explaining this approach.
<br>+ * tests/actions.at (Destroying lookahead assigned by semantic
<br>+ action): New test group checking that translation happens before
<br>+ lookahead destructor calls at parser return. Previously,
<br>+ incorrect destructors were called.
<br>+ * tests/conflicts.at (parse.error=verbose and consistent
<br>+ errors): New test group checking that translation happens at
<br>+ syntax error detection before the associated verbose error
<br>+ message and the associated lookahead destructor calls. While
<br>+ the destructor call is fixed by this patch, the verbose error
<br>+ message is currently incorrect due to another bug (see
<br>+ comments in test group), so this is an expected failure for now.
<br>+
<br> 2009-12-18 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26861692&i=3" target="_top" rel="nofollow">jdenny@...</a>>
<br>
<br> lalr1.cc: don't discard non-existent lookahead on syntax error.
<br>diff --git a/NEWS b/NEWS
<br>index 92567dd..8aaf9e0 100644
<br>--- a/NEWS
<br>+++ b/NEWS
<br>@@ -212,6 +212,13 @@ Bison News
<br> were resolved with %nonassoc. Such tokens are now properly omitted
<br> from the list.
<br>
<br>+** Destructor calls fixed for lookaheads altered in semantic actions.
<br>+
<br>+ Previously for deterministic parsers in C, if a user semantic action
<br>+ altered yychar, the parser in some cases used the old yychar value to
<br>+ determine which destructor to call for the lookahead upon a syntax
<br>+ error or upon parser return. This bug has been fixed.
<br>+
<br> * Changes in version 2.4.2 (????-??-??):
<br>
<br> ** Detection of GNU M4 1.4.6 or newer during configure is improved.
<br>diff --git a/data/yacc.c b/data/yacc.c
<br>index 11ebbf1..dce9a6a 100644
<br>--- a/data/yacc.c
<br>+++ b/data/yacc.c
<br>@@ -621,7 +621,6 @@ do \
<br> { \
<br> yychar = (Token); \
<br> yylval = (Value); \
<br>- yytoken = YYTRANSLATE (yychar); \
<br> YYPOPSTACK (1); \
<br> goto yybackup; \
<br> } \
<br>@@ -1420,6 +1419,17 @@ yyreduce:
<br> ]b4_user_actions[
<br> default: break;
<br> }
<br>+ /* User semantic actions sometimes alter yychar, and that requires
<br>+ that yytoken be updated with the new translation. We take the
<br>+ approach of translating immediately before every use of yytoken.
<br>+ One alternative is translating here after every semantic action,
<br>+ but that translation would be missed if the semantic action invokes
<br>+ YYABORT, YYACCEPT, or YYERROR immediately after altering yychar or
<br>+ if it invokes YYBACKUP. In the case of YYABORT or YYACCEPT, an
<br>+ incorrect destructor might then be invoked immediately. In the
<br>+ case of YYERROR or YYBACKUP, subsequent parser actions might lead
<br>+ to an incorrect destructor call or verbose syntax error message
<br>+ before the lookahead is translated. */
<br> YY_SYMBOL_PRINT ("-> $$ =", yyr1[yyn], &yyval, &yyloc);
<br>
<br> YYPOPSTACK (yylen);
<br>@@ -1448,6 +1458,10 @@ yyreduce:
<br> | yyerrlab -- here on detecting error |
<br> `------------------------------------*/
<br> yyerrlab:
<br>+ /* Make sure we have latest lookahead translation. See comments at
<br>+ user semantic actions for why this is necessary. */
<br>+ yytoken = YYTRANSLATE (yychar);
<br>+
<br> /* If not already recovering from an error, report this error. */
<br> if (!yyerrstatus)
<br> {
<br>@@ -1600,8 +1614,13 @@ yyexhaustedlab:
<br>
<br> yyreturn:
<br> if (yychar != YYEMPTY)
<br>- yydestruct ("Cleanup: discarding lookahead",
<br>- yytoken, &yylval]b4_locations_if([, &yylloc])[]b4_user_args[);
<br>+ {
<br>+ /* Make sure we have latest lookahead translation. See comments at
<br>+ user semantic actions for why this is necessary. */
<br>+ yytoken = YYTRANSLATE (yychar);
<br>+ yydestruct ("Cleanup: discarding lookahead",
<br>+ yytoken, &yylval]b4_locations_if([, &yylloc])[]b4_user_args[);
<br>+ }
<br> /* Do not reclaim the symbols of the rule which action triggered
<br> this YYABORT or YYACCEPT. */
<br> YYPOPSTACK (yylen);
<br>diff --git a/tests/actions.at b/tests/actions.at
<br>index d317920..144791c 100644
<br>--- a/tests/actions.at
<br>+++ b/tests/actions.at
<br>@@ -1408,3 +1408,74 @@ AT_MATCHES_CHECK([input.c], [[// TEST:Y:1 [;{}]*\n;\}$]], [[12]])
<br> AT_MATCHES_CHECK([input.c], [[#define TEST_MACRO_N \\\n\[\]"broken\\" \$ \@ \$\$ \@\$ \[\];\\\nstring;"\}]], [[2]])
<br>
<br> AT_CLEANUP
<br>+
<br>+
<br>+## -------------------------------------------------- ##
<br>+## Destroying lookahead assigned by semantic action. ##
<br>+## -------------------------------------------------- ##
<br>+
<br>+AT_SETUP([[Destroying lookahead assigned by semantic action]])
<br>+
<br>+AT_DATA_GRAMMAR([input.y],
<br>+[[
<br>+%code {
<br>+ #include <assert.h>
<br>+ #include <stdio.h>
<br>+ static void yyerror (char const *);
<br>+ static int yylex (void);
<br>+ #define USE(Var)
<br>+}
<br>+
<br>+%destructor { fprintf (stderr, "'a' destructor\n"); } 'a'
<br>+%destructor { fprintf (stderr, "'b' destructor\n"); } 'b'
<br>+
<br>+%%
<br>+
<br>+// In a previous version of Bison, yychar assigned by the semantic
<br>+// action below was not translated into yytoken before the lookahead was
<br>+// discarded and thus before its destructor (selected according to
<br>+// yytoken) was called in order to return from yyparse. This would
<br>+// happen even if YYACCEPT was performed in a later semantic action as
<br>+// long as only consistent states with default reductions were visited
<br>+// in between. However, we leave YYACCEPT in the same semantic action
<br>+// for this test in order to show that skeletons cannot simply translate
<br>+// immediately after every semantic action because a semantic action
<br>+// that has set yychar might not always return normally. Instead,
<br>+// skeletons must translate before every use of yytoken.
<br>+start: 'a' accept { USE($1); } ;
<br>+accept: /*empty*/ {
<br>+ assert (yychar == YYEMPTY);
<br>+ yychar = 'b';
<br>+ YYACCEPT;
<br>+} ;
<br>+
<br>+%%
<br>+
<br>+static void
<br>+yyerror (char const *msg)
<br>+{
<br>+ fprintf (stderr, "%s\n", msg);
<br>+}
<br>+
<br>+static int
<br>+yylex (void)
<br>+{
<br>+ static char const *input = "a";
<br>+ return *input++;
<br>+}
<br>+
<br>+int
<br>+main (void)
<br>+{
<br>+ return yyparse ();
<br>+}
<br>+]])
<br>+
<br>+AT_BISON_CHECK([[-o input.c input.y]])
<br>+AT_COMPILE([[input]])
<br>+AT_PARSER_CHECK([[./input]], [[0]], [],
<br>+[['b' destructor
<br>+'a' destructor
<br>+]])
<br>+
<br>+AT_CLEANUP
<br>diff --git a/tests/conflicts.at b/tests/conflicts.at
<br>index 26ec08d..faaa3b9 100644
<br>--- a/tests/conflicts.at
<br>+++ b/tests/conflicts.at
<br>@@ -139,6 +139,130 @@ AT_CLEANUP
<br>
<br>
<br>
<br>+## ------------------------------------------- ##
<br>+## parse.error=verbose and consistent errors. ##
<br>+## ------------------------------------------- ##
<br>+
<br>+AT_SETUP([[parse.error=verbose and consistent errors]])
<br>+
<br>+m4_pushdef([AT_CONSISTENT_ERRORS_CHECK], [
<br>+
<br>+AT_BISON_CHECK([$1[ -o input.c input.y]])
<br>+AT_COMPILE([[input]])
<br>+
<br>+m4_pushdef([AT_EXPECTING], [m4_if($3, [ab], [[, expecting 'a' or 'b']],
<br>+ $3, [a], [[, expecting 'a']],
<br>+ $3, [b], [[, expecting 'b']])])
<br>+
<br>+AT_PARSER_CHECK([[./input]], [[1]], [],
<br>+[[syntax error, unexpected ]$2[]AT_EXPECTING[
<br>+]])
<br>+
<br>+m4_popdef([AT_EXPECTING])
<br>+
<br>+])
<br>+
<br>+AT_DATA_GRAMMAR([input.y],
<br>+[[%code {
<br>+ #include <assert.h>
<br>+ #include <stdio.h>
<br>+ int yylex (void);
<br>+ void yyerror (char const *);
<br>+ #define USE(Var)
<br>+}
<br>+
<br>+%define parse.error verbose
<br>+
<br>+// The point isn't to test IELR here, but state merging happens to
<br>+// complicate the example.
<br>+%define lr.type ielr
<br>+
<br>+%nonassoc 'a'
<br>+
<br>+// If yylval=0 here, then we know that the 'a' destructor is being
<br>+// invoked incorrectly for the 'b' set in the semantic action below.
<br>+// All 'a' tokens are returned by yylex, which sets yylval=1.
<br>+%destructor {
<br>+ if (!$$)
<br>+ fprintf (stderr, "Wrong destructor.\n");
<br>+} 'a'
<br>+
<br>+%%
<br>+
<br>+// The lookahead assigned by the semantic action isn't needed before
<br>+// either error action is encountered. In a previous version of Bison,
<br>+// this was a problem as it meant yychar was not translated into yytoken
<br>+// before either error action. The second error action thus invoked a
<br>+// destructor that it selected according to the incorrect yytoken. The
<br>+// first error action would have reported an incorrect unexpected token
<br>+// except that, due to another bug, the unexpected token is not reported
<br>+// at all because the error action is the default action in a consistent
<br>+// state. That bug still needs to be fixed.
<br>+start: error-reduce consistent-error 'a' { USE ($3); } ;
<br>+
<br>+error-reduce:
<br>+ 'a' 'a' consistent-reduction consistent-error 'a'
<br>+ { USE (($1, $2, $5)); }
<br>+| 'a' error
<br>+ { USE ($1); }
<br>+;
<br>+
<br>+consistent-reduction: /*empty*/ {
<br>+ assert (yychar == YYEMPTY);
<br>+ yylval = 0;
<br>+ yychar = 'b';
<br>+} ;
<br>+
<br>+consistent-error:
<br>+ 'a' { USE ($1); }
<br>+| /*empty*/ %prec 'a'
<br>+;
<br>+
<br>+// Provide another context in which all rules are useful so that this
<br>+// test case looks a little more realistic.
<br>+start: 'b' consistent-error 'b' ;
<br>+
<br>+%%
<br>+
<br>+int
<br>+yylex (void)
<br>+{
<br>+ static char const *input = "aa";
<br>+ yylval = 1;
<br>+ return *input++;
<br>+}
<br>+
<br>+void
<br>+yyerror (char const *msg)
<br>+{
<br>+ fprintf (stderr, "%s\n", msg);
<br>+}
<br>+
<br>+int
<br>+main (void)
<br>+{
<br>+ return yyparse ();
<br>+}
<br>+]])
<br>+
<br>+# See comments in grammar for why this test doesn't succeed.
<br>+AT_XFAIL_IF([[:]])
<br>+
<br>+AT_CONSISTENT_ERRORS_CHECK([], [['b']], [[none]])
<br>+AT_CONSISTENT_ERRORS_CHECK([[-Dlr.default-reductions=consistent]],
<br>+ [['b']], [[none]])
<br>+
<br>+# Canonical LR doesn't foresee the error for 'a'!
<br>+AT_CONSISTENT_ERRORS_CHECK([[-Dlr.default-reductions=accepting]],
<br>+ [[$end]], [[a]])
<br>+AT_CONSISTENT_ERRORS_CHECK([[-Flr.type=canonical-lr]], [[$end]], [[a]])
<br>+
<br>+m4_popdef([AT_CONSISTENT_ERRORS_CHECK])
<br>+
<br>+AT_CLEANUP
<br>+
<br>+
<br>+
<br> ## ------------------------- ##
<br> ## Unresolved SR Conflicts. ##
<br> ## ------------------------- ##
<br>--
<br>1.5.4.3
<br><br><br><br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Patches-f1505.html" embed="fixTarget[1505]" target="_top" >Gnu - Bison - Patches</a></p>tag:old.nabble.com,2006:post-26858812remove YYFAIL from lalr1.java and yacc.c2009-12-19T13:40:33Z2009-12-19T13:40:33ZJoel E. Denny-4I propose we remove the YYFAIL feature from all skeletons as of release
<br>2.5. My rationale is as follows:
<br><br>1. YYFAIL only appears in two skeletons, and removing it from them should
<br>not create a legitimate backward incompatibility:
<br><br>a. YYFAIL appears in yacc.c but is not documented for yacc.c. Comments in
<br>yacc.c describe it as follows:
<br><br> /* Like YYERROR except do call yyerror. This remains here temporarily
<br> to ease the transition to the new meaning of YYERROR, for GCC.
<br> Once GCC version 2 has supplanted version 1, this can go. */
<br><br>In other words, we've been planning to get rid of it for a long while.
<br>(However, is there any evidence that users actually use it? Do we need to
<br>poll them or warn them somehow?)
<br><br>b. YYFAIL is documented for lalr1.java, but that skeleton is still labeled
<br>as experimental, so we're free to change its interface.
<br><br>2. YYFAIL's invocation of yyerror is not useful:
<br><br>a. If verbose error messages are not enabled, then it's easy for the user
<br>to just report the simple "syntax error" himself before invoking YYERROR.
<br>"semantic error" or a more specific message would really be more
<br>appropriate anyway because any syntactic analysis (at least in one sense)
<br>cannot predict YYERROR and YYFAIL invocations. That includes the
<br>syntactic analysis done for verbose error messages....
<br><br>b. If verbose error messages are enabled, the expected token list it
<br>reports is almost certainly wrong because YYFAIL rejects some tokens from
<br>that list. The expected token list might even include the token reported
<br>as unexpected! For example, try the input "b" for this grammar:
<br><br> start: { } 'a'
<br> | { YYFAIL; } 'b'
<br> ;
<br><br>The output is:
<br><br> syntax error, unexpected 'b', expecting 'b'
<br><br>Of course, expected token lists are broken anyway by state merging and
<br>default reductions (that's why 'a' is missing in this example), but my
<br>upcoming lookahead correction technique will address that issue for LALR
<br>and IELR, and canonical LR already addresses it:
<br><br> syntax error, unexpected 'b', expecting 'a' or 'b'
<br><br>3. Unlike YYERROR, YYFAIL does not pop the RHS of the production. Not
<br>popping could be dangerous if the semantic action already freed or
<br>otherwise used some of the RHS values. Moreover, it's inconsistent with
<br>the documented way in which Bison normally handles the RHS. That is,
<br>YYERROR, YYABORT, and YYACCEPT all pop the RHS so that their values won't
<br>be freed or reused later.
<br><br>4. There's one other difference between YYFAIL and YYERROR: if the parser
<br>has not shifted a token since the error token, YYFAIL will discard the
<br>lookahead before popping the stack again. In other words, YYFAIL acts
<br>more like a subsequent syntax error, but YYERROR restarts error recovery.
<br>It's not clear to me yet which approach is better, but the inconsistency
<br>is just another undocumented point of confusion. If the user really needs
<br>to remove the lookahead before invoking YYERROR, yyclearin should be
<br>provided.
<br><br>5. Bison development and user documentation would be easier if all
<br>skeletons made the same decision about whether to support YYFAIL.
<br><br><br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Patches-f1505.html" embed="fixTarget[1505]" target="_top" >Gnu - Bison - Patches</a></p>tag:old.nabble.com,2006:post-26858262Re: please help me with this2009-12-19T12:26:36Z2009-12-19T12:26:36ZPhilip HerronOn Fri, 2009-12-18 at 01:58 -0500, Tejas Kajarekar wrote:
<div class='shrinkable-quote'><div class='shrinkable-quote'><br>> Hello,
<br>> My name is Tejas. I am trying to use Flex and Bison to parse (and convert)
<br>> one file to another format. Currently, I am not bothered about conversion,
<br>> just the parsing.
<br>> I have never used either tools, and what my problem is, may be trivial.
<br>> (Worth laughing)
<br>> Here is my problem:
<br>> I am going to attach my .l and .y files and .txt file which is my input.
<br>> Here I will just include the tokenIZING part of .l and grammar of .y.
<br>> This is part of my flex which returns tokens to yyparse:-
<br>> "graph" { return GRAPH; }
<br>> "{" { return '{'; }
<br>> "}" { return '}'; }
<br>> "init" { yylval= yytext;
<br>> return INIT; }
<br>> {id} { yylval = yytext;
<br>> return ID; }
<br>> where id = [a-zA-Z_][a-zA-z_0-9]*
<br>>
<br>> This is my bison grammar:-
<br>> %%
<br>> input: /* empty */
<br>> | GRAPH '{' graphitemset '}' { printf("GRAPH
<br>> FOUND"); }
<br>> | error
<br>>
<br>> graphitemset: responsivestate { printf("RESPONSIVE
<br>> STATE"); }
<br>>
<br>> responsivestate: INIT statename '{' '}' { printf("INIT %s",$2);
<br>> }
<br>>
<br>> statename: /* empty */
<br>> | ID
<br>>
<br>> %%
<br>>
<br>> After running it,
<br>> C:\>gnu\bin\flex -olex.yy.c -i s.l
<br>> C:\>gnu\bin\bison -d s.y
<br>> C:\>gcc -ose s.tab.c lex.yy.c -ly -lfl
<br>> C:\>se < s.txt
<br>> Now if my s.txt = *graph{init{}}*
<br>> then I get output *INIT (null)RESPONSIVE STATEGRAPH FOUND.*
<br>>
<br>> However, if I change s.txt to *graph{initnabc{}}*
<br>> then I get *se: syntax error line 0*
<br>> i.e., it only recognizes state names which are empty, otherwise, there is
<br>> some problem with ID.
<br>>
<br>> I want it to recognoze graph, init as well as the state name.
<br>> I think ID definition conflicts with "graph", or "init".
<br>> It also generates a warning: No new line at end of file for .l file.
<br>> Please help me with this problem.
<br>> __________________________________________________________________________
<br>> Also, simlar kind of problem, but with very simple files:
<br>> "graph" { return GRAPH; }
<br>> {id} { yylval = yytext;
<br>> return ID; }
<br>> for tokens
<br>>
<br>>
<br>> input: GRAPH { printf("GRAPH FOUND"); }
<br>> | GRAPH ID { printf("GRAPH + ID"); }
<br>> the grammar
<br>>
<br>> with input graph, its executes to* GRAPH FOUND;*
<br>> with input graphgraph, it gives syntax error; and not GRAPH FOUND twice.,
<br>> nor GRAPH + ID for something like graphbhg.
<br>>
<br>> Do we have to run the tokenizing routing in some loop like while not EOF,
<br>> etc.?
<br>>
<br>> I have no idea what to do to get the desired output.
<br>> This may sound very funny, but as I said, I have no experience with these
<br>> two tools.
<br>> I will really appreciate your help.
<br>> The actual input will be quite large, I am just experimenting with
<br>> graph{init{}}, and having problems with that
<br>> Also, sorry for the long email.
<br>>
<br>> thanking you in advance,
<br>> tejasSK
<br>> _______________________________________________
<br>> <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26858262&i=0" target="_top" rel="nofollow">help-bison@...</a> <a href="http://lists.gnu.org/mailman/listinfo/help-bison" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/help-bison</a></div></div>Hey
<br><br>I think i see your problem, but not quite sure which attachments are
<br>relevant.
<br><br>Is the s.txt the input data i take it: ``graph { init { } }''
<br><br>Anyway's not totaly sure on your syntax. But if you want to build a
<br>lexer and parser for this you need to think about your syntax a little
<br>more as in know what is syntax/delimiters/operators and what are
<br>keywords and identifiers.
<br><br>So from here you only have 3 things to worry about 'lexing' which are
<br>'{', '}' keywords 'graph' 'init' and IDENTIFIERS
<br><br>So i take it from your s.y grammar you want the syntax to be something
<br>like:
<br><br>'graph { init <statename> { <parameters> } }'
<br><br>Well first i see you wanted it to look like 'graph{ initnabc { } }',
<br>your lexer cant split out the 'init' and 'abc' since the regex for that
<br>token will simply return an IDENTIFIER, since it see init<more chars> so
<br>it can't be the INIT keyword it has to be an identifier. So you language
<br>MUST be something like ``'graph { init abcdefg { } }' unless you want to
<br>treat the 'initabc' as an IDENTIFIER and do the lexing your self.
<br><br>So this leaves making your lexer look like this:
<br><br>"graph" { return GRAPH; }
<br>"{" { return '{'; }
<br>"}" { return '}'; }
<br>"init" { return INIT; }
<br>{id} { yylval.string= strdup(yytext);
<br> return ID; }
<br><br>Note i have started using 'yylval.<>' you should get in the habit of
<br>using a <%union> in bison since it makes type handling much easier since
<br>YYSTYPE is default int i IIRC and you can do #define YYSTYPE const char*
<br>or somthing but its simply not dynamic enough.
<br><br>So your lexer if it is the 's.l' needs a simply edit and make sure you
<br>include the header generated by bison in a sec.
<br><br>In your parser you need to add in your:
<br><br>%union{ const char* string; <other types you wish to return> }
<br><br>Then your grammar:
<br><br>%token GRAPH
<br>%token INIT
<br>%token<string> ID
<br><br>/* You can also add in %type<union ident type> rule for semantics which
<br>return values using $$=<foo>; */
<br><br>%start graph
<br><br>%%
<br><br>graph: /* empty */
<br> | GRAPH '{' graphparameters '}'
<br> | error
<br> ;
<br><br>graphparameters:
<br> INIT stateid '{' '}'
<br> ;
<br><br>stateid: /* empty */
<br> | ID
<br> ;
<br><br>%%
<br><br>And then you can add in what you want so your grammar was right. But
<br>with a little cleanup your better ready to handle more semantics easier
<br>using %union :). Remember think about what your syntax means is really
<br>important. I learnt that the hard way ;).
<br><br>Then you could start handling n length parameters for your state with
<br>grammar like:
<br><br>%type<string> stateid
<br>%type<paramlist> parameters
<br>%type<symbol> item
<br><br>%%
<br><br>graph: /* empty */
<br> | GRAPH '{' graphparameters '}'
<br> | error
<br> ;
<br><br>graphparameters:
<br> INIT stateid '{' parameters '}'
<br> ;
<br><br>stateid: /* empty */
<br> | ID { $$= $1; }
<br> ;
<br><br>parameters:
<br> parameters item
<br> | item
<br> ;
<br><br>item: INTEGER
<br> | ...
<br> ;
<br><br>Hope this helps.
<br><br>--Phil
<br><br /> <br />_______________________________________________
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26858262&i=1" target="_top" rel="nofollow">help-bison@...</a> <a href="http://lists.gnu.org/mailman/listinfo/help-bison" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/help-bison</a><div class="small"><br/><img src="http://old.nabble.com/images/icon_attachment.gif" > <strong>signature.asc</strong> (204 bytes) <a href="http://old.nabble.com/attachment/26858262/0/signature.asc" target="_top">Download Attachment</a></div><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Help-f1504.html" embed="fixTarget[1504]" target="_top" >Gnu - Bison - Help</a></p>tag:old.nabble.com,2006:post-26852864[PATCH] lalr1.cc: don't discard non-existent lookahead on syntax error.2009-12-18T21:17:39Z2009-12-18T21:17:39ZJoel E. Denny-4I pushed this to master. The bug does not exist on branch-2.5. The test
<br>case makes this bug look pretty obscure, and maybe that's right, but I
<br>really didn't spend much time searching for a simpler test case.
<br><br>>From d59beda068aef97834ea84ce4f5cc99b487e4b68 Mon Sep 17 00:00:00 2001
<br>From: Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26852864&i=0" target="_top" rel="nofollow">jdenny@...</a>>
<br>Date: Fri, 18 Dec 2009 23:57:18 -0500
<br>Subject: [PATCH] lalr1.cc: don't discard non-existent lookahead on syntax error.
<br><br>* data/lalr1.cc (parser::parse): Check yyempty first.
<br>* tests/c++.at (Syntax error discarding no lookahead): New test
<br>group.
<br>---
<br> ChangeLog | 7 +++++
<br> data/lalr1.cc | 2 +-
<br> tests/c++.at | 79 +++++++++++++++++++++++++++++++++++++++++++++++++++++++-
<br> 3 files changed, 85 insertions(+), 3 deletions(-)
<br><br>diff --git a/ChangeLog b/ChangeLog
<br>index 2926da7..1c15cc5 100644
<br>--- a/ChangeLog
<br>+++ b/ChangeLog
<br>@@ -1,3 +1,10 @@
<br>+2009-12-18 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26852864&i=1" target="_top" rel="nofollow">jdenny@...</a>>
<br>+
<br>+ lalr1.cc: don't discard non-existent lookahead on syntax error.
<br>+ * data/lalr1.cc (parser::parse): Check yyempty first.
<br>+ * tests/c++.at (Syntax error discarding no lookahead): New test
<br>+ group.
<br>+
<br> 2009-12-17 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26852864&i=2" target="_top" rel="nofollow">jdenny@...</a>>
<br>
<br> Code cleanup.
<br>diff --git a/data/lalr1.cc b/data/lalr1.cc
<br>index 9e0d522..6174e82 100644
<br>--- a/data/lalr1.cc
<br>+++ b/data/lalr1.cc
<br>@@ -876,7 +876,7 @@ m4_ifdef([b4_lex_param], [, ]b4_lex_param)));])[
<br> /* Return failure if at end of input. */
<br> if (yyla.type == yyeof_)
<br> YYABORT;
<br>- else
<br>+ else if (!yyempty)
<br> {
<br> yy_destroy_ ("Error: discarding", yyla);
<br> yyempty = true;
<br>diff --git a/tests/c++.at b/tests/c++.at
<br>index beffb1c..ce64d6d 100644
<br>--- a/tests/c++.at
<br>+++ b/tests/c++.at
<br>@@ -286,8 +286,6 @@ AT_CHECK_DOXYGEN([Public])
<br> AT_CHECK_DOXYGEN([Private])
<br>
<br>
<br>-
<br>-
<br> ## ------------ ##
<br> ## Namespaces. ##
<br> ## ------------ ##
<br>@@ -369,3 +367,80 @@ AT_CHECK_NAMESPACE([[foo[3]::bar::baz]], [[-]])
<br> AT_CHECK_NAMESPACE([[foo::bar,baz]], [[-]])
<br> AT_CHECK_NAMESPACE([[foo::bar::(baz /* Pacify Emacs ) */]], [[-]])
<br> AT_CLEANUP
<br>+
<br>+
<br>+## -------------------------------------- ##
<br>+## Syntax error discarding no lookahead. ##
<br>+## -------------------------------------- ##
<br>+
<br>+# After a syntax error, lalr1.cc used to not check whether there
<br>+# actually is a lookahead before discarding the lookahead. As a result,
<br>+# it mistakenly invoked the destructor for the previous lookahead.
<br>+
<br>+AT_SETUP([[Syntax error discarding no lookahead]])
<br>+
<br>+AT_DATA_GRAMMAR([[input.yy]],
<br>+[[%skeleton "lalr1.cc"
<br>+
<br>+%code {
<br>+ #include <string>
<br>+ int yylex (yy::parser::semantic_type *, yy::location *);
<br>+ #define USE(Args)
<br>+}
<br>+
<br>+%defines
<br>+%locations
<br>+%define parse.error verbose
<br>+
<br>+%nonassoc 'a' ;
<br>+
<br>+%destructor {
<br>+ std::cerr << "Discarding 'a'." << std::endl;
<br>+} 'a'
<br>+
<br>+%%
<br>+
<br>+start: error-reduce consistent-error 'a' { USE ($3); };
<br>+
<br>+error-reduce:
<br>+ 'a' 'a' consistent-error 'a' { USE (($1, $2, $4)); }
<br>+| 'a' error { std::cerr << "Reducing 'a'." << std::endl; USE ($1); }
<br>+;
<br>+
<br>+consistent-error:
<br>+ 'a'
<br>+| /*empty*/ %prec 'a'
<br>+;
<br>+
<br>+%%
<br>+
<br>+int
<br>+yylex (yy::parser::semantic_type *, yy::location *)
<br>+{
<br>+ static char const *input = "aa";
<br>+ return *input++;
<br>+}
<br>+
<br>+void
<br>+yy::parser::error (const location_type &, const std::string &m)
<br>+{
<br>+ std::cerr << m << std::endl;
<br>+}
<br>+
<br>+int
<br>+main (void)
<br>+{
<br>+ yy::parser parser;
<br>+ return parser.parse ();
<br>+}
<br>+]])
<br>+AT_BISON_CHECK([[-o input.cc input.yy]], [[0]], [[]], [[ignore-nolog]])
<br>+AT_COMPILE_CXX([[input]])
<br>+# This used to print "Discarding 'a'." again at the end.
<br>+AT_PARSER_CHECK([[./input]], [[1]], [[]],
<br>+[[syntax error
<br>+Discarding 'a'.
<br>+Reducing 'a'.
<br>+]])
<br>+
<br>+AT_CLEANUP
<br>--
<br>1.5.4.3
<br><br><br><br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Patches-f1505.html" embed="fixTarget[1505]" target="_top" >Gnu - Bison - Patches</a></p>tag:old.nabble.com,2006:post-26853636please help me with this2009-12-17T22:58:09Z2009-12-17T22:58:09ZTejas KajarekarHello,
<br>My name is Tejas. I am trying to use Flex and Bison to parse (and convert)
<br>one file to another format. Currently, I am not bothered about conversion,
<br>just the parsing.
<br>I have never used either tools, and what my problem is, may be trivial.
<br>(Worth laughing)
<br>Here is my problem:
<br>I am going to attach my .l and .y files and .txt file which is my input.
<br>Here I will just include the tokenIZING part of .l and grammar of .y.
<br>This is part of my flex which returns tokens to yyparse:-
<br>"graph" { return GRAPH; }
<br>"{" { return '{'; }
<br>"}" { return '}'; }
<br>"init" { yylval= yytext;
<br> return INIT; }
<br>{id} { yylval = yytext;
<br> return ID; }
<br> where id = [a-zA-Z_][a-zA-z_0-9]*
<br><br>This is my bison grammar:-
<br>%%
<br>input: /* empty */
<br> | GRAPH '{' graphitemset '}' { printf("GRAPH
<br>FOUND"); }
<br> | error
<br><br>graphitemset: responsivestate { printf("RESPONSIVE
<br>STATE"); }
<br><br>responsivestate: INIT statename '{' '}' { printf("INIT %s",$2);
<br> }
<br><br>statename: /* empty */
<br> | ID
<br><br>%%
<br><br>After running it,
<br>C:\>gnu\bin\flex -olex.yy.c -i s.l
<br>C:\>gnu\bin\bison -d s.y
<br>C:\>gcc -ose s.tab.c lex.yy.c -ly -lfl
<br>C:\>se < s.txt
<br>Now if my s.txt = *graph{init{}}*
<br>then I get output *INIT (null)RESPONSIVE STATEGRAPH FOUND.*
<br><br>However, if I change s.txt to *graph{initnabc{}}*
<br>then I get *se: syntax error line 0*
<br>i.e., it only recognizes state names which are empty, otherwise, there is
<br>some problem with ID.
<br><br>I want it to recognoze graph, init as well as the state name.
<br>I think ID definition conflicts with "graph", or "init".
<br>It also generates a warning: No new line at end of file for .l file.
<br>Please help me with this problem.
<br>__________________________________________________________________________
<br>Also, simlar kind of problem, but with very simple files:
<br>"graph" { return GRAPH; }
<br>{id} { yylval = yytext;
<br> return ID; }
<br>for tokens
<br><br><br>input: GRAPH { printf("GRAPH FOUND"); }
<br> | GRAPH ID { printf("GRAPH + ID"); }
<br>the grammar
<br><br>with input graph, its executes to* GRAPH FOUND;*
<br>with input graphgraph, it gives syntax error; and not GRAPH FOUND twice.,
<br>nor GRAPH + ID for something like graphbhg.
<br><br>Do we have to run the tokenizing routing in some loop like while not EOF,
<br>etc.?
<br><br>I have no idea what to do to get the desired output.
<br>This may sound very funny, but as I said, I have no experience with these
<br>two tools.
<br>I will really appreciate your help.
<br>The actual input will be quite large, I am just experimenting with
<br>graph{init{}}, and having problems with that
<br>Also, sorry for the long email.
<br><br>thanking you in advance,
<br>tejasSK
<br><br />graph{init{}}<br /> <br /> <br /> <br /> <br />_______________________________________________
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26853636&i=0" target="_top" rel="nofollow">help-bison@...</a> <a href="http://lists.gnu.org/mailman/listinfo/help-bison" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/help-bison</a><div class="small"><br/><img src="http://old.nabble.com/images/icon_attachment.gif" > <strong>s.y</strong> (976 bytes) <a href="http://old.nabble.com/attachment/26853636/0/s.y" target="_top">Download Attachment</a><br/><img src="http://old.nabble.com/images/icon_attachment.gif" > <strong>s.l</strong> (1K) <a href="http://old.nabble.com/attachment/26853636/1/s.l" target="_top">Download Attachment</a><br/><img src="http://old.nabble.com/images/icon_attachment.gif" > <strong>ex.l</strong> (340 bytes) <a href="http://old.nabble.com/attachment/26853636/2/ex.l" target="_top">Download Attachment</a><br/><img src="http://old.nabble.com/images/icon_attachment.gif" > <strong>ex.y</strong> (722 bytes) <a href="http://old.nabble.com/attachment/26853636/3/ex.y" target="_top">Download Attachment</a></div><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Help-f1504.html" embed="fixTarget[1504]" target="_top" >Gnu - Bison - Help</a></p>tag:old.nabble.com,2006:post-26837817Can't find yyreset in bison 2.4.1 for linux2009-12-17T01:43:52Z2009-12-17T01:43:52ZLi, Li-WeiHi
<br>
<br>Could someone please tell me where I can find the yyreset definition ?
<br>
<br>I downloaded and build bison 2.4.1 to my redhat 3.6 box, but yyreset is
<br>not there.
<br>
<br>Many thanks in advance.
<br>
<br>/li-wei
<br><br><br>This e-mail is confidential and is intended only for the person to whom it is
<br>addressed. It may be privileged and should not be read, copied or used by
<br>anyone other than the intended recipient. If you are not that person, you are
<br>not permitted to make use of the information and you are requested to notify
<br>the sender immediately that you have received it and then destroy the copy in
<br>your possession without disclosing its contents to any person.
<br><br>Commerzbank AG may monitor outgoing and incoming e-mails. By replying to this
<br>e-mail you consent to such monitoring. We believe, but do not warrant, that
<br>this e-mail and any attachments are virus free, but you must take full
<br>responsibility for virus checking.
<br><br>This e-mail was sent either by Commerzbank AG, London Branch, or by Commerzbank
<br>Corporates & Markets, a division of Commerzbank AG. Commerzbank AG is a
<br>company incorporated in the Federal Republic of Germany and registered in
<br>England (registered number FC008139, place of business 30 Gresham Street,
<br>London EC2V 7PG) and is authorised by Bundesanstalt fuer
<br>Finanzdienstleistungsaufsicht (BaFin) and authorised and subject to limited
<br>regulation by the Financial Services Authority (FSA).
<br>_______________________________________________
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26837817&i=0" target="_top" rel="nofollow">help-bison@...</a> <a href="http://lists.gnu.org/mailman/listinfo/help-bison" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/help-bison</a><br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Help-f1504.html" embed="fixTarget[1504]" target="_top" >Gnu - Bison - Help</a></p>tag:old.nabble.com,2006:post-26823719[PATCH] Code cleanup.2009-12-16T22:18:41Z2009-12-16T22:18:41ZJoel E. Denny-4I pushed this to master and branch-2.5.
<br><br>From 387b4d504728391605f751e74a36fb7d0d441b19 Mon Sep 17 00:00:00 2001
<br>From: Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26823719&i=0" target="_top" rel="nofollow">jdenny@...</a>>
<br>Date: Thu, 17 Dec 2009 01:09:09 -0500
<br>Subject: [PATCH] Code cleanup.
<br><br>* src/symtab.c, src/symtab.h (symbol_class_get_string): Remove
<br>function, which is no longer used.
<br>---
<br> ChangeLog | 6 ++++++
<br> src/symtab.c | 18 ------------------
<br> src/symtab.h | 3 ---
<br> 3 files changed, 6 insertions(+), 21 deletions(-)
<br><br>diff --git a/ChangeLog b/ChangeLog
<br>index 2817f11..2926da7 100644
<br>--- a/ChangeLog
<br>+++ b/ChangeLog
<br>@@ -1,3 +1,9 @@
<br>+2009-12-17 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26823719&i=1" target="_top" rel="nofollow">jdenny@...</a>>
<br>+
<br>+ Code cleanup.
<br>+ * src/symtab.c, src/symtab.h (symbol_class_get_string): Remove
<br>+ function, which is no longer used.
<br>+
<br> 2009-12-16 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26823719&i=2" target="_top" rel="nofollow">jdenny@...</a>>
<br>
<br> Add gcc's -Wundef to test suite and fix another warning from it.
<br>diff --git a/src/symtab.c b/src/symtab.c
<br>index ce29dfc..ebdf04c 100644
<br>--- a/src/symtab.c
<br>+++ b/src/symtab.c
<br>@@ -219,24 +219,6 @@ symbol_type_set (symbol *sym, uniqstr type_name, location loc)
<br> }
<br> }
<br>
<br>-/*-----------------------------------.
<br>-| Get the CLASS associated with SYM. |
<br>-`-----------------------------------*/
<br>-
<br>-const char *
<br>-symbol_class_get_string (symbol *sym)
<br>-{
<br>- if (sym->class)
<br>- {
<br>- if (sym->class == token_sym)
<br>- return "terminal";
<br>- else if (sym->class == nterm_sym)
<br>- return "nonterminal";
<br>- }
<br>- return "unknown";
<br>-}
<br>-
<br>-
<br> /*-----------------------------------------.
<br> | Set the DESTRUCTOR associated with SYM. |
<br> `-----------------------------------------*/
<br>diff --git a/src/symtab.h b/src/symtab.h
<br>index b093af3..6204d90 100644
<br>--- a/src/symtab.h
<br>+++ b/src/symtab.h
<br>@@ -144,9 +144,6 @@ void symbol_make_alias (symbol *sym, symbol *str, location loc);
<br> Do nothing if passed 0 as \c type_name. */
<br> void symbol_type_set (symbol *sym, uniqstr type_name, location loc);
<br>
<br>-/** Get the \c class string associated with \c sym. */
<br>-const char *symbol_class_get_string (symbol *sym);
<br>-
<br> /** Set the \c destructor associated with \c sym. */
<br> void symbol_destructor_set (symbol *sym, code_props const *destructor);
<br>
<br>--
<br>1.5.4.3
<br><br><br><br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Patches-f1505.html" embed="fixTarget[1505]" target="_top" >Gnu - Bison - Patches</a></p>tag:old.nabble.com,2006:post-26822157[PATCH] Fix handling of yychar manipulation in user semantic actions.2009-12-16T18:28:42Z2009-12-16T18:28:42ZJoel E. Denny-4A few months ago, I started talking about a lookahead correction mechanism
<br>I was working on for Bison. Its most obvious benefit is to correct the
<br>list of expected tokens in verbose syntax error messages. I'm finally
<br>getting back to pushing that patch series through.
<br><br>The following patch is next. The problem it addresses is that users
<br>sometimes write semantic actions that manipulate yychar. Of course, they
<br>can do this with YYBACKUP or yyclearin. However, my impression is that
<br>many users alter yychar directly. In this case, the translation in
<br>yytoken isn't always updated, and bad destructor calls or syntax error
<br>messages can result. This patch fixes that.
<br><br>I'd like to push this patch to master and branch-2.5. Any feedback would
<br>be appreciated. I'd especially like to hear comments on my analysis of
<br>glr.c, lalr1.cc, and lalr1.java.
<br><br>Don't worry about the unfixed bug I mention in the test suite. The next
<br>patch will more thoroughly explain it and fix it.
<br><br>From cfdf80e9e688706a449dc68662e5e4ad0674ab01 Mon Sep 17 00:00:00 2001
<br>From: Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26822157&i=0" target="_top" rel="nofollow">jdenny@...</a>>
<br>Date: Wed, 16 Dec 2009 20:55:40 -0500
<br>Subject: [PATCH] Fix handling of yychar manipulation in user semantic actions.
<br><br>The problem was that yacc.c didn't always update the yychar
<br>translation afterwards. However, other skeletons appear to be
<br>fine. glr.c appears to already translate yychar before every
<br>use. lalr1.cc does not define yychar and does not document its
<br>replacement, yyla, for users. In lalr1.java, yychar is out of
<br>scope during semantic actions.
<br>* NEWS (2.5): Document.
<br>* data/yacc.c (YYBACKUP): Don't bother translating yychar into
<br>yytoken here.
<br>(yyparse, yypush_parse): Instead, translate before every use of
<br>yytoken, and add comments explaining this approach.
<br>* tests/actions.at (Destroying lookahead assigned by semantic
<br>action): New test group checking that translation happens before
<br>lookahead destructor calls at parser return. Previously,
<br>incorrect destructors were called.
<br>* tests/conflicts.at (parse.error=verbose and consistent
<br>errors): New test group checking that translation happens at
<br>syntax error detection before the associated verbose error
<br>message and the associated lookahead destructor calls. While
<br>the destructor call is fixed by this patch, the verbose error
<br>message is currently incorrect due to another bug (see
<br>comments in test group), so this is an expected failure for now.
<br>---
<br> ChangeLog | 26 +++
<br> NEWS | 7 +
<br> data/yacc.c | 25 +++-
<br> src/parse-gram.c | 522 +++++++++++++++++++++++-----------------------------
<br> src/parse-gram.h | 68 +------
<br> tests/actions.at | 71 +++++++
<br> tests/conflicts.at | 115 ++++++++++++
<br> 7 files changed, 480 insertions(+), 354 deletions(-)
<br><br>diff --git a/ChangeLog b/ChangeLog
<br>index 2817f11..cd8016a 100644
<br>--- a/ChangeLog
<br>+++ b/ChangeLog
<br>@@ -1,5 +1,31 @@
<br> 2009-12-16 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26822157&i=1" target="_top" rel="nofollow">jdenny@...</a>>
<br>
<br>+ Fix handling of yychar manipulation in user semantic actions.
<br>+ The problem was that yacc.c didn't always update the yychar
<br>+ translation afterwards. However, other skeletons appear to be
<br>+ fine. glr.c appears to already translate yychar before every
<br>+ use. lalr1.cc does not define yychar and does not document its
<br>+ replacement, yyla, for users. In lalr1.java, yychar is out of
<br>+ scope during semantic actions.
<br>+ * NEWS (2.5): Document.
<br>+ * data/yacc.c (YYBACKUP): Don't bother translating yychar into
<br>+ yytoken here.
<br>+ (yyparse, yypush_parse): Instead, translate before every use of
<br>+ yytoken, and add comments explaining this approach.
<br>+ * tests/actions.at (Destroying lookahead assigned by semantic
<br>+ action): New test group checking that translation happens before
<br>+ lookahead destructor calls at parser return. Previously,
<br>+ incorrect destructors were called.
<br>+ * tests/conflicts.at (parse.error=verbose and consistent
<br>+ errors): New test group checking that translation happens at
<br>+ syntax error detection before the associated verbose error
<br>+ message and the associated lookahead destructor calls. While
<br>+ the destructor call is fixed by this patch, the verbose error
<br>+ message is currently incorrect due to another bug (see
<br>+ comments in test group), so this is an expected failure for now.
<br>+
<br>+2009-12-16 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26822157&i=2" target="_top" rel="nofollow">jdenny@...</a>>
<br>+
<br> Add gcc's -Wundef to test suite and fix another warning from it.
<br> * NEWS (2.4.2): Update description of -Wundef fix.
<br> * configure.ac (WARN_CXXFLAGS_TEST): New substitution.
<br>diff --git a/NEWS b/NEWS
<br>index 92567dd..8aaf9e0 100644
<br>--- a/NEWS
<br>+++ b/NEWS
<br>@@ -212,6 +212,13 @@ Bison News
<br> were resolved with %nonassoc. Such tokens are now properly omitted
<br> from the list.
<br>
<br>+** Destructor calls fixed for lookaheads altered in semantic actions.
<br>+
<br>+ Previously for deterministic parsers in C, if a user semantic action
<br>+ altered yychar, the parser in some cases used the old yychar value to
<br>+ determine which destructor to call for the lookahead upon a syntax
<br>+ error or upon parser return. This bug has been fixed.
<br>+
<br> * Changes in version 2.4.2 (????-??-??):
<br>
<br> ** Detection of GNU M4 1.4.6 or newer during configure is improved.
<br>diff --git a/data/yacc.c b/data/yacc.c
<br>index 11ebbf1..dce9a6a 100644
<br>--- a/data/yacc.c
<br>+++ b/data/yacc.c
<br>@@ -621,7 +621,6 @@ do \
<br> { \
<br> yychar = (Token); \
<br> yylval = (Value); \
<br>- yytoken = YYTRANSLATE (yychar); \
<br> YYPOPSTACK (1); \
<br> goto yybackup; \
<br> } \
<br>@@ -1420,6 +1419,17 @@ yyreduce:
<br> ]b4_user_actions[
<br> default: break;
<br> }
<br>+ /* User semantic actions sometimes alter yychar, and that requires
<br>+ that yytoken be updated with the new translation. We take the
<br>+ approach of translating immediately before every use of yytoken.
<br>+ One alternative is translating here after every semantic action,
<br>+ but that translation would be missed if the semantic action invokes
<br>+ YYABORT, YYACCEPT, or YYERROR immediately after altering yychar or
<br>+ if it invokes YYBACKUP. In the case of YYABORT or YYACCEPT, an
<br>+ incorrect destructor might then be invoked immediately. In the
<br>+ case of YYERROR or YYBACKUP, subsequent parser actions might lead
<br>+ to an incorrect destructor call or verbose syntax error message
<br>+ before the lookahead is translated. */
<br> YY_SYMBOL_PRINT ("-> $$ =", yyr1[yyn], &yyval, &yyloc);
<br>
<br> YYPOPSTACK (yylen);
<br>@@ -1448,6 +1458,10 @@ yyreduce:
<br> | yyerrlab -- here on detecting error |
<br> `------------------------------------*/
<br> yyerrlab:
<br>+ /* Make sure we have latest lookahead translation. See comments at
<br>+ user semantic actions for why this is necessary. */
<br>+ yytoken = YYTRANSLATE (yychar);
<br>+
<br> /* If not already recovering from an error, report this error. */
<br> if (!yyerrstatus)
<br> {
<br>@@ -1600,8 +1614,13 @@ yyexhaustedlab:
<br>
<br> yyreturn:
<br> if (yychar != YYEMPTY)
<br>- yydestruct ("Cleanup: discarding lookahead",
<br>- yytoken, &yylval]b4_locations_if([, &yylloc])[]b4_user_args[);
<br>+ {
<br>+ /* Make sure we have latest lookahead translation. See comments at
<br>+ user semantic actions for why this is necessary. */
<br>+ yytoken = YYTRANSLATE (yychar);
<br>+ yydestruct ("Cleanup: discarding lookahead",
<br>+ yytoken, &yylval]b4_locations_if([, &yylloc])[]b4_user_args[);
<br>+ }
<br> /* Do not reclaim the symbols of the rule which action triggered
<br> this YYABORT or YYACCEPT. */
<br> YYPOPSTACK (yylen);
<br>diff --git a/tests/actions.at b/tests/actions.at
<br>index d317920..144791c 100644
<br>--- a/tests/actions.at
<br>+++ b/tests/actions.at
<br>@@ -1408,3 +1408,74 @@ AT_MATCHES_CHECK([input.c], [[// TEST:Y:1 [;{}]*\n;\}$]], [[12]])
<br> AT_MATCHES_CHECK([input.c], [[#define TEST_MACRO_N \\\n\[\]"broken\\" \$ \@ \$\$ \@\$ \[\];\\\nstring;"\}]], [[2]])
<br>
<br> AT_CLEANUP
<br>+
<br>+
<br>+## -------------------------------------------------- ##
<br>+## Destroying lookahead assigned by semantic action. ##
<br>+## -------------------------------------------------- ##
<br>+
<br>+AT_SETUP([[Destroying lookahead assigned by semantic action]])
<br>+
<br>+AT_DATA_GRAMMAR([input.y],
<br>+[[
<br>+%code {
<br>+ #include <assert.h>
<br>+ #include <stdio.h>
<br>+ static void yyerror (char const *);
<br>+ static int yylex (void);
<br>+ #define USE(Var)
<br>+}
<br>+
<br>+%destructor { fprintf (stderr, "'a' destructor\n"); } 'a'
<br>+%destructor { fprintf (stderr, "'b' destructor\n"); } 'b'
<br>+
<br>+%%
<br>+
<br>+// In a previous version of Bison, yychar assigned by the semantic
<br>+// action below was not translated into yytoken before the lookahead was
<br>+// discarded and thus before its destructor (selected according to
<br>+// yytoken) was called in order to return from yyparse. This would
<br>+// happen even if YYACCEPT was performed in a later semantic action as
<br>+// long as only consistent states with default reductions were visited
<br>+// in between. However, we leave YYACCEPT in the same semantic action
<br>+// for this test in order to show that skeletons cannot simply translate
<br>+// immediately after every semantic action because a semantic action
<br>+// that has set yychar might not always return normally. Instead,
<br>+// skeletons must translate before every use of yytoken.
<br>+start: 'a' accept { USE($1); } ;
<br>+accept: /*empty*/ {
<br>+ assert (yychar == YYEMPTY);
<br>+ yychar = 'b';
<br>+ YYACCEPT;
<br>+} ;
<br>+
<br>+%%
<br>+
<br>+static void
<br>+yyerror (char const *msg)
<br>+{
<br>+ fprintf (stderr, "%s\n", msg);
<br>+}
<br>+
<br>+static int
<br>+yylex (void)
<br>+{
<br>+ static char const *input = "a";
<br>+ return *input++;
<br>+}
<br>+
<br>+int
<br>+main (void)
<br>+{
<br>+ return yyparse ();
<br>+}
<br>+]])
<br>+
<br>+AT_BISON_CHECK([[-o input.c input.y]])
<br>+AT_COMPILE([[input]])
<br>+AT_PARSER_CHECK([[./input]], [[0]], [],
<br>+[['b' destructor
<br>+'a' destructor
<br>+]])
<br>+
<br>+AT_CLEANUP
<br>diff --git a/tests/conflicts.at b/tests/conflicts.at
<br>index 26ec08d..11eacd4 100644
<br>--- a/tests/conflicts.at
<br>+++ b/tests/conflicts.at
<br>@@ -139,6 +139,121 @@ AT_CLEANUP
<br>
<br>
<br>
<br>+## ------------------------------------------- ##
<br>+## parse.error=verbose and consistent errors. ##
<br>+## ------------------------------------------- ##
<br>+
<br>+AT_SETUP([[parse.error=verbose and consistent errors]])
<br>+
<br>+m4_pushdef([AT_CONSISTENT_ERRORS_CHECK], [
<br>+
<br>+AT_BISON_CHECK([$1[ -o input.c input.y]])
<br>+AT_COMPILE([[input]])
<br>+
<br>+m4_pushdef([AT_EXPECTING], [m4_if($2, [ab], [[, expecting 'a' or 'b']],
<br>+ $2, [a], [[, expecting 'a']],
<br>+ $2, [b], [[, expecting 'b']])])
<br>+
<br>+AT_PARSER_CHECK([[./input]], [[1]], [],
<br>+[[syntax error, unexpected $end]AT_EXPECTING[
<br>+]])
<br>+
<br>+m4_popdef([AT_EXPECTING])
<br>+
<br>+])
<br>+
<br>+AT_DATA_GRAMMAR([input.y],
<br>+[[%code {
<br>+ #include <assert.h>
<br>+ #include <stdio.h>
<br>+ int yylex (void);
<br>+ void yyerror (char const *);
<br>+ #define USE(Var)
<br>+}
<br>+
<br>+%define parse.error verbose
<br>+
<br>+// The point isn't to test IELR here, but state merging happens to
<br>+// complicate the example.
<br>+%define lr.type ielr
<br>+
<br>+%nonassoc 'a'
<br>+
<br>+// If yylval=0 here, then we know that the 'a' destructor is being
<br>+// invoked incorrectly for the $end set in the semantic action below.
<br>+// All other tokens are returned by yylex, which sets yylval=1.
<br>+%destructor {
<br>+ if (!$$)
<br>+ fprintf (stderr, "Wrong destructor.\n");
<br>+} 'a'
<br>+
<br>+%%
<br>+
<br>+// The lookahead isn't needed before the error action is encountered.
<br>+// In a previous version of Bison, this was a problem as it meant yychar
<br>+// was not translated into yytoken before error recovery, which discards
<br>+// the lookahead and thus invokes a destructor that it selected
<br>+// according to the incorrect yytoken. That missing translation also
<br>+// would have caused the wrong unexpected token to be reported except
<br>+// that, due to another bug, the unexpected token is not reported at all
<br>+// because the error is detected in a consistent state with an error
<br>+// action. That bug still needs to be fixed.
<br>+start: consistent-error-on-a-a 'a' { USE ($2); } ;
<br>+
<br>+consistent-error-on-a-a:
<br>+ 'a' consistent-reduction
<br>+ | 'a' consistent-reduction 'a' { USE (($1, $3)); }
<br>+ ;
<br>+
<br>+consistent-reduction: /*empty*/ {
<br>+ assert (yychar == YYEMPTY);
<br>+ yylval = 0;
<br>+ yychar = 0;
<br>+} ;
<br>+
<br>+// Provide another context in which all rules are useful so that this
<br>+// test case looks a little more realistic.
<br>+start: 'b' consistent-error-on-a-a 'b' ;
<br>+
<br>+%%
<br>+
<br>+int
<br>+yylex (void)
<br>+{
<br>+ static char const *input = "a";
<br>+ yylval = 1;
<br>+ return *input++;
<br>+}
<br>+
<br>+void
<br>+yyerror (char const *msg)
<br>+{
<br>+ fprintf (stderr, "%s\n", msg);
<br>+}
<br>+
<br>+int
<br>+main (void)
<br>+{
<br>+ return yyparse ();
<br>+}
<br>+]])
<br>+
<br>+# See comments in grammar for why this test doesn't succeed.
<br>+AT_XFAIL_IF([[:]])
<br>+
<br>+AT_CONSISTENT_ERRORS_CHECK([], [[none]])
<br>+AT_CONSISTENT_ERRORS_CHECK([[-Dlr.default-reductions=consistent]], [[none]])
<br>+
<br>+# Canonical LR doesn't foresee the error for 'a'!
<br>+AT_CONSISTENT_ERRORS_CHECK([[-Dlr.default-reductions=accepting]], [[a]])
<br>+AT_CONSISTENT_ERRORS_CHECK([[-Flr.type=canonical-lr]], [[a]])
<br>+
<br>+m4_popdef([AT_CONSISTENT_ERRORS_CHECK])
<br>+
<br>+AT_CLEANUP
<br>+
<br>+
<br>+
<br> ## ------------------------- ##
<br> ## Unresolved SR Conflicts. ##
<br> ## ------------------------- ##
<br>--
<br>1.5.4.3
<br><br><br><br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Patches-f1505.html" embed="fixTarget[1505]" target="_top" >Gnu - Bison - Patches</a></p>tag:old.nabble.com,2006:post-26816878Re: several messages2009-12-16T11:21:45Z2009-12-16T11:21:45ZJoel E. Denny-4On Wed, 16 Dec 2009, Joel E. Denny wrote:
<br><br>> > Avoid warnings from gcc -Wunused y.tab.c.
<br>> > Simplify y.tab.c when location tracking is disabled.
<br>>
<br>> Including the -Wundef correction, I pushed those patches and the following
<br>> patches to master, branch-2.5, and branch-2.4.2 except that I omitted the
<br>> b4_locations_if changes from branch-2.4.2.
<br><br>I also pushed this patch to all three branches to help us avoid this
<br>problem in the future.
<br><br>>From a603c6e0bb8b9e79d4136ee0fd0d1bf6a7d010ef Mon Sep 17 00:00:00 2001
<br>From: Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26816878&i=0" target="_top" rel="nofollow">jdenny@...</a>>
<br>Date: Wed, 16 Dec 2009 13:19:19 -0500
<br>Subject: [PATCH] Add gcc's -Wundef to test suite and fix another warning from it.
<br><br>* NEWS (2.4.2): Update description of -Wundef fix.
<br>* configure.ac (WARN_CXXFLAGS_TEST): New substitution.
<br>(WARN_CFLAGS_TEST): New substitution.
<br>* data/glr.c: Avoid warning about __STRICT_ANSI__.
<br>* tests/atlocal.in (CFLAGS): Use WARN_CFLAGS_TEST instead of
<br>WARN_CFLAGS.
<br>(NO_WERROR_CFLAGS): Likewise.
<br>(CXXFLAGS): Use WARN_CXXFLAGS_TEST instead of WARN_CXXFLAGS.
<br>---
<br> ChangeLog | 12 ++++++++++++
<br> NEWS | 5 +++--
<br> configure.ac | 6 ++++++
<br> data/glr.c | 3 ++-
<br> tests/atlocal.in | 6 +++---
<br> 5 files changed, 26 insertions(+), 6 deletions(-)
<br><br>diff --git a/ChangeLog b/ChangeLog
<br>index 97c8f6e..2817f11 100644
<br>--- a/ChangeLog
<br>+++ b/ChangeLog
<br>@@ -1,5 +1,17 @@
<br> 2009-12-16 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26816878&i=1" target="_top" rel="nofollow">jdenny@...</a>>
<br>
<br>+ Add gcc's -Wundef to test suite and fix another warning from it.
<br>+ * NEWS (2.4.2): Update description of -Wundef fix.
<br>+ * configure.ac (WARN_CXXFLAGS_TEST): New substitution.
<br>+ (WARN_CFLAGS_TEST): New substitution.
<br>+ * data/glr.c: Avoid warning about __STRICT_ANSI__.
<br>+ * tests/atlocal.in (CFLAGS): Use WARN_CFLAGS_TEST instead of
<br>+ WARN_CFLAGS.
<br>+ (NO_WERROR_CFLAGS): Likewise.
<br>+ (CXXFLAGS): Use WARN_CXXFLAGS_TEST instead of WARN_CXXFLAGS.
<br>+
<br>+2009-12-16 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26816878&i=2" target="_top" rel="nofollow">jdenny@...</a>>
<br>+
<br> * data/yacc.c: Reformat m4 a little.
<br>
<br> 2009-12-16 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26816878&i=3" target="_top" rel="nofollow">jdenny@...</a>>
<br>diff --git a/NEWS b/NEWS
<br>index 186b8e0..92567dd 100644
<br>--- a/NEWS
<br>+++ b/NEWS
<br>@@ -216,8 +216,9 @@ Bison News
<br>
<br> ** Detection of GNU M4 1.4.6 or newer during configure is improved.
<br>
<br>-** Warnings from gcc's -Wundef option about undefined YYENABLE_NLS and
<br>- YYLTYPE_IS_TRIVIAL in C/C++ parsers are now avoided.
<br>+** Warnings from gcc's -Wundef option about undefined YYENABLE_NLS,
<br>+ YYLTYPE_IS_TRIVIAL, and __STRICT_ANSI__ in C/C++ parsers are now
<br>+ avoided.
<br>
<br> ** %code is now a permanent feature.
<br>
<br>diff --git a/configure.ac b/configure.ac
<br>index 4f1b20f..8e9fc29 100644
<br>--- a/configure.ac
<br>+++ b/configure.ac
<br>@@ -68,6 +68,12 @@ if test "${enableval}" = yes; then
<br> gl_WARN_ADD([-Wshadow])
<br> gl_WARN_ADD([-Wstrict-prototypes])
<br> AC_SUBST([WARN_CFLAGS])
<br>+ # Warnings for the test suite only.
<br>+ gl_WARN_ADD([-Wundef], [WARN_CFLAGS_TEST])
<br>+ WARN_CXXFLAGS_TEST="$WARN_CXXFLAGS $WARN_CFLAGS_TEST"
<br>+ WARN_CFLAGS_TEST="$WARN_CFLAGS $WARN_CFLAGS_TEST"
<br>+ AC_SUBST([WARN_CXXFLAGS_TEST])
<br>+ AC_SUBST([WARN_CFLAGS_TEST])
<br> AC_DEFINE([lint], 1, [Define to 1 if the compiler is checking for lint.])
<br> fi
<br>
<br>diff --git a/data/glr.c b/data/glr.c
<br>index e73b251..8c964e5 100644
<br>--- a/data/glr.c
<br>+++ b/data/glr.c
<br>@@ -318,7 +318,8 @@ b4_percent_code_get[]dnl
<br> #ifndef __attribute__
<br> /* This feature is available in gcc versions 2.5 and later. */
<br> # if (! defined __GNUC__ || __GNUC__ < 2 \
<br>- || (__GNUC__ == 2 && __GNUC_MINOR__ < 5) || __STRICT_ANSI__)
<br>+ || (__GNUC__ == 2 && __GNUC_MINOR__ < 5) \
<br>+ || (defined __STRICT_ANSI__ && __STRICT_ANSI__))
<br> # define __attribute__(Spec) /* empty */
<br> # endif
<br> #endif
<br>diff --git a/tests/atlocal.in b/tests/atlocal.in
<br>index fb1cef6..7277a59 100644
<br>--- a/tests/atlocal.in
<br>+++ b/tests/atlocal.in
<br>@@ -8,11 +8,11 @@
<br> : ${CC='@CC@'}
<br>
<br> # We want no optimization.
<br>-CFLAGS='@O0CFLAGS@ @WARN_CFLAGS@ @WERROR_CFLAGS@'
<br>+CFLAGS='@O0CFLAGS@ @WARN_CFLAGS_TEST@ @WERROR_CFLAGS@'
<br>
<br> # Sometimes a test group needs to ignore gcc warnings, so it locally
<br> # sets CFLAGS to this.
<br>-NO_WERROR_CFLAGS='@O0CFLAGS@ @WARN_CFLAGS@'
<br>+NO_WERROR_CFLAGS='@O0CFLAGS@ @WARN_CFLAGS_TEST@'
<br>
<br> # We need `config.h'.
<br> CPPFLAGS="-I$abs_top_builddir/lib @CPPFLAGS@"
<br>@@ -27,7 +27,7 @@ GCC='@GCC@'
<br> BISON_CXX_WORKS='@BISON_CXX_WORKS@'
<br>
<br> # We want no optimization with C++, too.
<br>-CXXFLAGS='@O0CXXFLAGS@ @WARN_CXXFLAGS@ @WERROR_CFLAGS@'
<br>+CXXFLAGS='@O0CXXFLAGS@ @WARN_CXXFLAGS_TEST@ @WERROR_CFLAGS@'
<br>
<br> # Are special link options needed?
<br> LDFLAGS='@LDFLAGS@'
<br>--
<br>1.5.4.3
<br><br><br><br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Patches-f1505.html" embed="fixTarget[1505]" target="_top" >Gnu - Bison - Patches</a></p>tag:old.nabble.com,2006:post-26807059Re: several messages2009-12-15T23:05:21Z2009-12-15T23:05:21ZJoel E. Denny-4On Tue, 15 Dec 2009, Joel E. Denny wrote:
<br><br>> > Thanks for the catch. I should have been saying -Wundef all this
<br>> > time.
<br><br>> Two following two patches follow.
<br>>
<br>> Avoid warnings from gcc -Wunused y.tab.c.
<br>> Simplify y.tab.c when location tracking is disabled.
<br><br>Including the -Wundef correction, I pushed those patches and the following
<br>patches to master, branch-2.5, and branch-2.4.2 except that I omitted the
<br>b4_locations_if changes from branch-2.4.2.
<br><br>Thanks again for your contribution.
<br><br>From c938d650dbddd3d94b3543bab76e1afce99c409d Mon Sep 17 00:00:00 2001
<br>From: Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26807059&i=0" target="_top" rel="nofollow">jdenny@...</a>>
<br>Date: Wed, 16 Dec 2009 01:14:44 -0500
<br>Subject: [PATCH] Document gcc -Wundef fix.
<br><br>* NEWS (2.4.2): Here.
<br>* THANKS (Jonathan Nieder): Add.
<br>---
<br> ChangeLog | 6 ++++++
<br> NEWS | 3 +++
<br> THANKS | 1 +
<br> 3 files changed, 10 insertions(+), 0 deletions(-)
<br><br>diff --git a/ChangeLog b/ChangeLog
<br>index 029826f..3fedc38 100644
<br>--- a/ChangeLog
<br>+++ b/ChangeLog
<br>@@ -1,3 +1,9 @@
<br>+2009-12-16 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26807059&i=1" target="_top" rel="nofollow">jdenny@...</a>>
<br>+
<br>+ Document gcc -Wundef fix.
<br>+ * NEWS (2.4.2): Here.
<br>+ * THANKS (Jonathan Nieder): Add.
<br>+
<br> 2009-12-15 Jonathan Nieder <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26807059&i=2" target="_top" rel="nofollow">jrnieder@...</a>> (tiny change)
<br>
<br> Simplify y.tab.c when location tracking is disabled.
<br>diff --git a/NEWS b/NEWS
<br>index d585139..186b8e0 100644
<br>--- a/NEWS
<br>+++ b/NEWS
<br>@@ -216,6 +216,9 @@ Bison News
<br>
<br> ** Detection of GNU M4 1.4.6 or newer during configure is improved.
<br>
<br>+** Warnings from gcc's -Wundef option about undefined YYENABLE_NLS and
<br>+ YYLTYPE_IS_TRIVIAL in C/C++ parsers are now avoided.
<br>+
<br> ** %code is now a permanent feature.
<br>
<br> A traditional Yacc prologue directive is written in the form:
<br>diff --git a/THANKS b/THANKS
<br>index 67d6663..0a203ed 100644
<br>--- a/THANKS
<br>+++ b/THANKS
<br>@@ -46,6 +46,7 @@ Jesse Thilo <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26807059&i=3" target="_top" rel="nofollow">jthilo@...</a>
<br> Jim Kent <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26807059&i=4" target="_top" rel="nofollow">jkent@...</a>
<br> Jim Meyering <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26807059&i=5" target="_top" rel="nofollow">jim@...</a>
<br> Joel E. Denny <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26807059&i=6" target="_top" rel="nofollow">jdenny@...</a>
<br>+Jonathan Nieder <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26807059&i=7" target="_top" rel="nofollow">jrnieder@...</a>
<br> Juan Manuel Guerrero <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26807059&i=8" target="_top" rel="nofollow">juan.guerrero@...</a>
<br> Kees Zeelenberg <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26807059&i=9" target="_top" rel="nofollow">kzlg@...</a>
<br> Keith Browne <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26807059&i=10" target="_top" rel="nofollow">kbrowne@...</a>
<br>--
<br>1.5.4.3
<br><br><br>From 19750d310c1f21b72fc26d0b8905fac08d175130 Mon Sep 17 00:00:00 2001
<br>From: Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26807059&i=11" target="_top" rel="nofollow">jdenny@...</a>>
<br>Date: Wed, 16 Dec 2009 01:41:17 -0500
<br>Subject: [PATCH] * data/yacc.c: Reformat m4 a little.
<br><br>---
<br> ChangeLog | 4 ++++
<br> data/yacc.c | 13 +++++++------
<br> 2 files changed, 11 insertions(+), 6 deletions(-)
<br><br>diff --git a/ChangeLog b/ChangeLog
<br>index 3fedc38..97c8f6e 100644
<br>--- a/ChangeLog
<br>+++ b/ChangeLog
<br>@@ -1,5 +1,9 @@
<br> 2009-12-16 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26807059&i=12" target="_top" rel="nofollow">jdenny@...</a>>
<br>
<br>+ * data/yacc.c: Reformat m4 a little.
<br>+
<br>+2009-12-16 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26807059&i=13" target="_top" rel="nofollow">jdenny@...</a>>
<br>+
<br> Document gcc -Wundef fix.
<br> * NEWS (2.4.2): Here.
<br> * THANKS (Jonathan Nieder): Add.
<br>diff --git a/data/yacc.c b/data/yacc.c
<br>index 711d41d..11ebbf1 100644
<br>--- a/data/yacc.c
<br>+++ b/data/yacc.c
<br>@@ -660,9 +660,9 @@ while (YYID (0))
<br> YYRHSLOC (Rhs, 0).last_column; \
<br> } \
<br> while (YYID (0))
<br>-#endif
<br>+#endif]b4_locations_if([[
<br>+
<br>
<br>-]b4_locations_if([[
<br> /* YY_LOCATION_PRINT -- Print the location on the stream.
<br> This macro was not mandated originally: define only if we know
<br> we won't break user code: when these are the locations we know. */
<br>@@ -676,14 +676,15 @@ while (YYID (0))
<br> # else
<br> # define YY_LOCATION_PRINT(File, Loc) ((void) 0)
<br> # endif
<br>-#endif
<br>-]], [[
<br>+#endif]], [[
<br>+
<br>+
<br> /* This macro is provided for backward compatibility. */
<br>
<br> #ifndef YY_LOCATION_PRINT
<br> # define YY_LOCATION_PRINT(File, Loc) ((void) 0)
<br>-#endif
<br>-]])[
<br>+#endif]])[
<br>+
<br>
<br> /* YYLEX -- calling `yylex' with the right arguments. */
<br>
<br>--
<br>1.5.4.3
<br><br><br><br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Patches-f1505.html" embed="fixTarget[1505]" target="_top" >Gnu - Bison - Patches</a></p>tag:old.nabble.com,2006:post-26806297Re: several messages2009-12-15T20:56:16Z2009-12-15T20:56:16ZJoel E. Denny-4On Tue, 15 Dec 2009, Jonathan Nieder wrote:
<br><br>> > By the way, which version of gcc are you using? The versions I've tried
<br>> > don't produce this warning for -Wunused.
<br>>
<br>> Thanks for the catch. I should have been saying -Wundef all this
<br>> time.
<br><br>Ah, thanks.
<br><br>> Could you correct this or should I resend?
<br><br>I'll get it.
<br><br>> Sorry for the trouble,
<br><br>Not a problem.
<br><br><br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Patches-f1505.html" embed="fixTarget[1505]" target="_top" >Gnu - Bison - Patches</a></p>tag:old.nabble.com,2006:post-26806263Re: several messages2009-12-15T20:50:33Z2009-12-15T20:50:33ZJonathan NiederJoel E. Denny wrote:
<br>> On Tue, 15 Dec 2009, Jonathan Nieder wrote:
<br><br>>> Avoid warnings from gcc -Wunused y.tab.c.
<br>>
<br>> By the way, which version of gcc are you using? The versions I've tried
<br>> don't produce this warning for -Wunused.
<br><br>Thanks for the catch. I should have been saying -Wundef all this
<br>time.
<br><br>Could you correct this or should I resend?
<br><br>Sorry for the trouble,
<br>Jonathan
<br><br><br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Patches-f1505.html" embed="fixTarget[1505]" target="_top" >Gnu - Bison - Patches</a></p>tag:old.nabble.com,2006:post-26806216Re: several messages2009-12-15T20:40:49Z2009-12-15T20:40:49ZJoel E. Denny-4On Tue, 15 Dec 2009, Jonathan Nieder wrote:
<br><br>> > I want to include the rest of your changes in release 2.4.2, which is
<br>> > purely a bug fix release, so please submit the b4_locations_if change in a
<br>> > separate patch.
<br>>
<br>> Oh! Sorry about that.
<br>>
<br>> Two following two patches follow.
<br><br>Thanks! I'll push them soon.
<br><br>> Avoid warnings from gcc -Wunused y.tab.c.
<br><br>By the way, which version of gcc are you using? The versions I've tried
<br>don't produce this warning for -Wunused.
<br><br><br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Patches-f1505.html" embed="fixTarget[1505]" target="_top" >Gnu - Bison - Patches</a></p>tag:old.nabble.com,2006:post-26806119Re: several messages2009-12-15T20:20:38Z2009-12-15T20:20:38ZJonathan NiederHi again,
<br><br>Joel E. Denny wrote:
<br><br>> I guess I'm just not used to seeing so many emails for one small issue.
<br>> I'm used to seeing patches appended to the main email that introduces
<br>> them.
<br><br>Okay, I’ll try this.
<br><br>> I want to include the rest of your changes in release 2.4.2, which is
<br>> purely a bug fix release, so please submit the b4_locations_if change in a
<br>> separate patch.
<br><br>Oh! Sorry about that.
<br><br>Two following two patches follow.
<br><br> Avoid warnings from gcc -Wunused y.tab.c.
<br> Simplify y.tab.c when location tracking is disabled.
<br><br> ChangeLog | 15 +++++++++++++++
<br> data/glr.c | 4 ++--
<br> data/lalr1.cc | 2 +-
<br> data/yacc.c | 14 ++++++++++----
<br> 4 files changed, 28 insertions(+), 7 deletions(-)
<br><br>-- %< ------- patch 1 -------
<br>From 23d4e466fcf6428cad857149d51c25475a870d4d Mon Sep 17 00:00:00 2001
<br>From: Jonathan Nieder <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26806119&i=0" target="_top" rel="nofollow">jrnieder@...</a>>
<br>Date: Tue, 15 Dec 2009 22:03:18 -0600
<br>Subject: Avoid warnings from gcc -Wunused y.tab.c.
<br><br>* data/glr.c: Check if YYENABLE_NLS and YYLTYPE_IS_TRIVIAL are
<br>defined before using them.
<br>* data/lalr1.cc: Likewise.
<br>* data/yacc.c: Likewise.
<br>---
<br> ChangeLog | 8 ++++++++
<br> data/glr.c | 4 ++--
<br> data/lalr1.cc | 2 +-
<br> data/yacc.c | 6 +++---
<br> 4 files changed, 14 insertions(+), 6 deletions(-)
<br><br>diff --git a/ChangeLog b/ChangeLog
<br>index 2c5d33e..e2a323e 100644
<br>--- a/ChangeLog
<br>+++ b/ChangeLog
<br>@@ -1,3 +1,11 @@
<br>+2009-12-15 Jonathan Nieder <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26806119&i=1" target="_top" rel="nofollow">jrnieder@...</a>>
<br>+
<br>+ Avoid warnings from gcc -Wunused y.tab.c.
<br>+ * data/glr.c: Check if YYENABLE_NLS and YYLTYPE_IS_TRIVIAL are
<br>+ defined before using them.
<br>+ * data/lalr1.cc: Likewise.
<br>+ * data/yacc.c: Likewise.
<br>+
<br> 2009-12-15 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26806119&i=2" target="_top" rel="nofollow">jdenny@...</a>>
<br>
<br> autoconf: update to latest for fix of M4 detection.
<br>diff --git a/data/glr.c b/data/glr.c
<br>index 6f1f7ee..e73b251 100644
<br>--- a/data/glr.c
<br>+++ b/data/glr.c
<br>@@ -256,7 +256,7 @@ b4_percent_code_get[]dnl
<br> #include <string.h>
<br>
<br> #ifndef YY_
<br>-# if YYENABLE_NLS
<br>+# if defined YYENABLE_NLS && YYENABLE_NLS
<br> # if ENABLE_NLS
<br> # include <libintl.h> /* INFRINGES ON USER NAME SPACE */
<br> # define YY_(msgid) dgettext ("bison-runtime", msgid)
<br>@@ -2264,7 +2264,7 @@ yyrecoverSyntaxError (yyGLRStack* yystackp]b4_user_formals[)
<br> yychar = YYEMPTY;
<br> yylval = yyval_default;
<br> ]b4_locations_if([
<br>-#if YYLTYPE_IS_TRIVIAL
<br>+#if defined YYLTYPE_IS_TRIVIAL && YYLTYPE_IS_TRIVIAL
<br> yylloc.first_line = yylloc.last_line = ]b4_location_initial_line[;
<br> yylloc.first_column = yylloc.last_column = ]b4_location_initial_column[;
<br> #endif
<br>diff --git a/data/lalr1.cc b/data/lalr1.cc
<br>index 1a2777e..9e0d522 100644
<br>--- a/data/lalr1.cc
<br>+++ b/data/lalr1.cc
<br>@@ -398,7 +398,7 @@ m4_if(b4_prefix, [yy], [],
<br> b4_percent_code_get[]dnl
<br>
<br> [#ifndef YY_
<br>-# if YYENABLE_NLS
<br>+# if defined YYENABLE_NLS && YYENABLE_NLS
<br> # if ENABLE_NLS
<br> # include <libintl.h> /* FIXME: INFRINGES ON USER NAME SPACE */
<br> # define YY_(msgid) dgettext ("bison-runtime", msgid)
<br>diff --git a/data/yacc.c b/data/yacc.c
<br>index ea2ae14..a7d7948 100644
<br>--- a/data/yacc.c
<br>+++ b/data/yacc.c
<br>@@ -374,7 +374,7 @@ typedef short int yytype_int16;
<br> #define YYSIZE_MAXIMUM ((YYSIZE_T) -1)
<br>
<br> #ifndef YY_
<br>-# if YYENABLE_NLS
<br>+# if defined YYENABLE_NLS && YYENABLE_NLS
<br> # if ENABLE_NLS
<br> # include <libintl.h> /* INFRINGES ON USER NAME SPACE */
<br> # define YY_(msgid) dgettext ("bison-runtime", msgid)
<br>@@ -668,7 +668,7 @@ while (YYID (0))
<br> we won't break user code: when these are the locations we know. */
<br>
<br> #ifndef YY_LOCATION_PRINT
<br>-# if YYLTYPE_IS_TRIVIAL
<br>+# if defined YYLTYPE_IS_TRIVIAL && YYLTYPE_IS_TRIVIAL
<br> # define YY_LOCATION_PRINT(File, Loc) \
<br> fprintf (File, "%d.%d-%d.%d", \
<br> (Loc).first_line, (Loc).first_column, \
<br>@@ -1192,7 +1192,7 @@ b4_c_function_def([[yyparse]], [[int]], b4_parse_param)[
<br> yyvsp = yyvs;]b4_locations_if([[
<br> yylsp = yyls;
<br>
<br>-#if YYLTYPE_IS_TRIVIAL
<br>+#if defined YYLTYPE_IS_TRIVIAL && YYLTYPE_IS_TRIVIAL
<br> /* Initialize the default location before parsing starts. */
<br> yylloc.first_line = yylloc.last_line = ]b4_location_initial_line[;
<br> yylloc.first_column = yylloc.last_column = ]b4_location_initial_column[;
<br>--
<br>1.6.5.4
<br><br>-- >% ----- end of patch 1 -----
<br>-- %< ----- patch 2 -----
<br>From 83e581279b9485ddd26a25c84cbd9c9633e4f5da Mon Sep 17 00:00:00 2001
<br>From: Jonathan Nieder <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26806119&i=3" target="_top" rel="nofollow">jrnieder@...</a>>
<br>Date: Tue, 15 Dec 2009 22:07:46 -0600
<br>Subject: Simplify y.tab.c when location tracking is disabled.
<br><br>* data/yacc.c: Do not check YYLTYPE_IS_TRIVIAL if location
<br>tracking is not enabled. Instead, unconditionally define
<br>YY_LOCATION_PRINT as a no-op for backward compatibility.
<br>---
<br> ChangeLog | 7 +++++++
<br> data/yacc.c | 8 +++++++-
<br> 2 files changed, 14 insertions(+), 1 deletions(-)
<br><br>diff --git a/ChangeLog b/ChangeLog
<br>index e2a323e..30aa283 100644
<br>--- a/ChangeLog
<br>+++ b/ChangeLog
<br>@@ -1,5 +1,12 @@
<br> 2009-12-15 Jonathan Nieder <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26806119&i=4" target="_top" rel="nofollow">jrnieder@...</a>>
<br>
<br>+ Simplify y.tab.c when location tracking is disabled.
<br>+ * data/yacc.c: Do not check YYLTYPE_IS_TRIVIAL if location
<br>+ tracking is not enabled. Instead, unconditionally define
<br>+ YY_LOCATION_PRINT as a no-op for backward compatibility.
<br>+
<br>+2009-12-15 Jonathan Nieder <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26806119&i=5" target="_top" rel="nofollow">jrnieder@...</a>>
<br>+
<br> Avoid warnings from gcc -Wunused y.tab.c.
<br> * data/glr.c: Check if YYENABLE_NLS and YYLTYPE_IS_TRIVIAL are
<br> defined before using them.
<br>diff --git a/data/yacc.c b/data/yacc.c
<br>index a7d7948..711d41d 100644
<br>--- a/data/yacc.c
<br>+++ b/data/yacc.c
<br>@@ -662,7 +662,7 @@ while (YYID (0))
<br> while (YYID (0))
<br> #endif
<br>
<br>-
<br>+]b4_locations_if([[
<br> /* YY_LOCATION_PRINT -- Print the location on the stream.
<br> This macro was not mandated originally: define only if we know
<br> we won't break user code: when these are the locations we know. */
<br>@@ -677,7 +677,13 @@ while (YYID (0))
<br> # define YY_LOCATION_PRINT(File, Loc) ((void) 0)
<br> # endif
<br> #endif
<br>+]], [[
<br>+/* This macro is provided for backward compatibility. */
<br>
<br>+#ifndef YY_LOCATION_PRINT
<br>+# define YY_LOCATION_PRINT(File, Loc) ((void) 0)
<br>+#endif
<br>+]])[
<br>
<br> /* YYLEX -- calling `yylex' with the right arguments. */
<br>
<br>--
<br>1.6.5.4
<br><br>-- >% ------ end of patch 2 -----
<br><br><br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Patches-f1505.html" embed="fixTarget[1505]" target="_top" >Gnu - Bison - Patches</a></p>tag:old.nabble.com,2006:post-26805984Re: [PATCH v2] Avoid warnings from gcc -Wunused y.tab.c2009-12-15T20:00:12Z2009-12-15T20:00:12ZJoel E. Denny-4Hi Jonathan,
<br><br>On Tue, 15 Dec 2009, Jonathan Nieder wrote:
<br><br>> Here is an updated patch incorporating all your suggestions except for
<br>> the last one.
<br><br>I didn't see this before responding to your previous email.
<br><br>> Joel E. Denny wrote:
<br>>
<br>> > How does the b4_locations_if part of this change help? If it isn't really
<br>> > necessary, then let's drop it. As far as I can tell, the addition of
<br>> > `defined YYLTYPE_IS_TRIVIAL &&' is sufficient.
<br>>
<br>> I tried to make it a little clearer what the conditional is for in
<br>> this version. If you still think the conditional should be dropped,
<br>> let me know and I’ll send a version without it.
<br><br>Ok, thanks, but I'd still like it to be a separate patch.
<br><div class='shrinkable-quote'><div class='shrinkable-quote'><br>> >From 81895390b3b97540e4ade42031449700d10d62bd Mon Sep 17 00:00:00 2001
<br>> From: Jonathan Nieder <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26805984&i=0" target="_top" rel="nofollow">jrnieder@...</a>>
<br>> Date: Sat, 12 Dec 2009 15:43:18 -0600
<br>> Subject: [PATCH] Avoid warnings from gcc -Wunused y.tab.c
<br>>
<br>> Building bison-produced parsers can result in a few warnings:
<br>>
<br>> y.tab.c:383:6: warning: "YYENABLE_NLS" is not defined
<br>> y.tab.c:1271:6: warning: "YYLTYPE_IS_TRIVIAL" is not defined
<br>>
<br>> These warnings are not revealing any actual problem (it is
<br>> perfectly reasonable to use #if FOO even if FOO is not defined).
<br>> Suppress them by checking if preprocessor symbols are defined
<br>> before using their values.
<br>>
<br>> While at it, do not use the YYLTYPE_IS_TRIVIAL symbol in the
<br>> generated parser source if location tracking is not enabled.
<br>> (Most uses were protected by b4_locations_if already.)
</div></div>I'm sorry I didn't make this clear before, but ChangeLog entries and git
<br>log entries should be the same (except that the blank line following the
<br>first line of a git log entry is usually removed in the ChangeLog entry).
<br><br>If you'd like, I can go ahead and revise your patch. If I don't hear from
<br>you again by tomorrow, I'll probably just go ahead and do that.
<br><br>Thanks.<p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Patches-f1505.html" embed="fixTarget[1505]" target="_top" >Gnu - Bison - Patches</a></p>tag:old.nabble.com,2006:post-26805819Re: several messages2009-12-15T19:37:54Z2009-12-15T19:37:54ZJoel E. Denny-4On Tue, 15 Dec 2009, Jonathan Nieder wrote:
<br><br>> > 3. After committing in your local repo, email us the output of:
<br>> >
<br>> > git format-patch --stdout -1
<br>> >
<br>> > That will make your patch easier to apply.
<br>>
<br>> I am not sure I understand this one.
<br><br>My mistake. By the time I wrote that, I forgot you were actually using
<br>format-patch, and so I just wrote the usual recipe.
<br><br>> Do you mean that I should keep
<br>> the "From " mbox delimiter and From:, Date:, etc fields inline in my
<br>> message?
<br><br>I guess I'm just not used to seeing so many emails for one small issue.
<br>I'm used to seeing patches appended to the main email that introduces
<br>them. However, feel free to send patches separately if that's somehow
<br>easier for you.
<br><div class='shrinkable-quote'><br>> >> --- a/data/yacc.c
<br>> >> +++ b/data/yacc.c
<br>> >> @@ -668,7 +668,7 @@ while (YYID (0))
<br>> >> we won't break user code: when these are the locations we know. */
<br>> >>
<br>> >> #ifndef YY_LOCATION_PRINT
<br>> >> -# if YYLTYPE_IS_TRIVIAL
<br>> >> +]b4_locations_if([[# if defined YYLTYPE_IS_TRIVIAL && YYLTYPE_IS_TRIVIAL
<br>> >> # define YY_LOCATION_PRINT(File, Loc) \
<br>> >> fprintf (File, "%d.%d-%d.%d", \
<br>> >> (Loc).first_line, (Loc).first_column, \
<br>> >> @@ -676,7 +676,9 @@ while (YYID (0))
<br>> >> # else
<br>> >> # define YY_LOCATION_PRINT(File, Loc) ((void) 0)
<br>> >> # endif
<br>> >> -#endif
<br>> >> +]], [[ /* This macro is maintained here in case user code is relying on it. */
<br>> >> +# define YY_LOCATION_PRINT(File, Loc) ((void) 0)
<br>> >> +]])[#endif
<br>> >
<br>> > How does the b4_locations_if part of this change help?
<br>>
<br>> Not much, admittedly. Here is why I did it anyway:
</div><br>I understand your reasoning, but it sounds like this change is distinct
<br>from the rest of your changes, it does not fix any bugs, and it might
<br>somehow create a backward compatibility issue for someone somewhere.
<br>That is, even though neither YY_LOCATION_PRINT nor YYLTYPE_IS_TRIVIAL is
<br>actually documented for users, YY_LOCATION_PRINT was put there for some
<br>purpose, and the discrepancy you're eliminating between yacc.c and glr.c
<br>might somehow be intentional.
<br><br>I want to include the rest of your changes in release 2.4.2, which is
<br>purely a bug fix release, so please submit the b4_locations_if change in a
<br>separate patch.
<br><br><br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Patches-f1505.html" embed="fixTarget[1505]" target="_top" >Gnu - Bison - Patches</a></p>tag:old.nabble.com,2006:post-26805780[PATCH v2] Avoid warnings from gcc -Wunused y.tab.c2009-12-15T19:33:02Z2009-12-15T19:33:02ZJonathan NiederHi Joel,
<br><br>Here is an updated patch incorporating all your suggestions except for
<br>the last one.
<br><br>Joel E. Denny wrote:
<br><br>> How does the b4_locations_if part of this change help? If it isn't really
<br>> necessary, then let's drop it. As far as I can tell, the addition of
<br>> `defined YYLTYPE_IS_TRIVIAL &&' is sufficient.
<br><br>I tried to make it a little clearer what the conditional is for in
<br>this version. If you still think the conditional should be dropped,
<br>let me know and I’ll send a version without it.
<br><br>Ciao,
<br>-- %< --
<br>From 81895390b3b97540e4ade42031449700d10d62bd Mon Sep 17 00:00:00 2001
<br>From: Jonathan Nieder <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26805780&i=0" target="_top" rel="nofollow">jrnieder@...</a>>
<br>Date: Sat, 12 Dec 2009 15:43:18 -0600
<br>Subject: [PATCH] Avoid warnings from gcc -Wunused y.tab.c
<br><br>Building bison-produced parsers can result in a few warnings:
<br><br> y.tab.c:383:6: warning: "YYENABLE_NLS" is not defined
<br> y.tab.c:1271:6: warning: "YYLTYPE_IS_TRIVIAL" is not defined
<br><br>These warnings are not revealing any actual problem (it is
<br>perfectly reasonable to use #if FOO even if FOO is not defined).
<br>Suppress them by checking if preprocessor symbols are defined
<br>before using their values.
<br><br>While at it, do not use the YYLTYPE_IS_TRIVIAL symbol in the
<br>generated parser source if location tracking is not enabled.
<br>(Most uses were protected by b4_locations_if already.)
<br>---
<br> ChangeLog | 15 +++++++++++++++
<br> data/glr.c | 4 ++--
<br> data/lalr1.cc | 2 +-
<br> data/yacc.c | 14 ++++++++++----
<br> 4 files changed, 28 insertions(+), 7 deletions(-)
<br><br>diff --git a/ChangeLog b/ChangeLog
<br>index 2c5d33e..1f7fd6b 100644
<br>--- a/ChangeLog
<br>+++ b/ChangeLog
<br>@@ -1,3 +1,18 @@
<br>+2009-12-15 Jonathan Nieder <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26805780&i=1" target="_top" rel="nofollow">jrnieder@...</a>>
<br>+
<br>+ Check defined FOO && FOO instead of FOO when testing
<br>+ user-defined preprocessor symbols.
<br>+ This avoids warnings from gcc -Wunused y.tab.c.
<br>+ * data/glr.c: Test if YYENABLE_NLS is defined before using it.
<br>+ * data/lalr1.cc: Likewise.
<br>+ * data/yacc.c: Likewise.
<br>+ * data/glr.c: Test if YYLTYPE_IS_TRIVIAL is defined before
<br>+ using it.
<br>+ * data/yacc.c: If location tracking is enabled, test if
<br>+ YYLTYPE_IS_TRIVIAL is defined before using it. Otherwise, do
<br>+ not use it: define YY_LOCATION_PRINT as a no-op unconditionally
<br>+ and add a comment explaining why.
<br>+
<br> 2009-12-15 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26805780&i=2" target="_top" rel="nofollow">jdenny@...</a>>
<br>
<br> autoconf: update to latest for fix of M4 detection.
<br>diff --git a/data/glr.c b/data/glr.c
<br>index 6f1f7ee..e73b251 100644
<br>--- a/data/glr.c
<br>+++ b/data/glr.c
<br>@@ -256,7 +256,7 @@ b4_percent_code_get[]dnl
<br> #include <string.h>
<br>
<br> #ifndef YY_
<br>-# if YYENABLE_NLS
<br>+# if defined YYENABLE_NLS && YYENABLE_NLS
<br> # if ENABLE_NLS
<br> # include <libintl.h> /* INFRINGES ON USER NAME SPACE */
<br> # define YY_(msgid) dgettext ("bison-runtime", msgid)
<br>@@ -2264,7 +2264,7 @@ yyrecoverSyntaxError (yyGLRStack* yystackp]b4_user_formals[)
<br> yychar = YYEMPTY;
<br> yylval = yyval_default;
<br> ]b4_locations_if([
<br>-#if YYLTYPE_IS_TRIVIAL
<br>+#if defined YYLTYPE_IS_TRIVIAL && YYLTYPE_IS_TRIVIAL
<br> yylloc.first_line = yylloc.last_line = ]b4_location_initial_line[;
<br> yylloc.first_column = yylloc.last_column = ]b4_location_initial_column[;
<br> #endif
<br>diff --git a/data/lalr1.cc b/data/lalr1.cc
<br>index 1a2777e..9e0d522 100644
<br>--- a/data/lalr1.cc
<br>+++ b/data/lalr1.cc
<br>@@ -398,7 +398,7 @@ m4_if(b4_prefix, [yy], [],
<br> b4_percent_code_get[]dnl
<br>
<br> [#ifndef YY_
<br>-# if YYENABLE_NLS
<br>+# if defined YYENABLE_NLS && YYENABLE_NLS
<br> # if ENABLE_NLS
<br> # include <libintl.h> /* FIXME: INFRINGES ON USER NAME SPACE */
<br> # define YY_(msgid) dgettext ("bison-runtime", msgid)
<br>diff --git a/data/yacc.c b/data/yacc.c
<br>index ea2ae14..04c6a90 100644
<br>--- a/data/yacc.c
<br>+++ b/data/yacc.c
<br>@@ -374,7 +374,7 @@ typedef short int yytype_int16;
<br> #define YYSIZE_MAXIMUM ((YYSIZE_T) -1)
<br>
<br> #ifndef YY_
<br>-# if YYENABLE_NLS
<br>+# if defined YYENABLE_NLS && YYENABLE_NLS
<br> # if ENABLE_NLS
<br> # include <libintl.h> /* INFRINGES ON USER NAME SPACE */
<br> # define YY_(msgid) dgettext ("bison-runtime", msgid)
<br>@@ -662,13 +662,13 @@ while (YYID (0))
<br> while (YYID (0))
<br> #endif
<br>
<br>-
<br>+]b4_locations_if([[
<br> /* YY_LOCATION_PRINT -- Print the location on the stream.
<br> This macro was not mandated originally: define only if we know
<br> we won't break user code: when these are the locations we know. */
<br>
<br> #ifndef YY_LOCATION_PRINT
<br>-# if YYLTYPE_IS_TRIVIAL
<br>+# if defined YYLTYPE_IS_TRIVIAL && YYLTYPE_IS_TRIVIAL
<br> # define YY_LOCATION_PRINT(File, Loc) \
<br> fprintf (File, "%d.%d-%d.%d", \
<br> (Loc).first_line, (Loc).first_column, \
<br>@@ -677,7 +677,13 @@ while (YYID (0))
<br> # define YY_LOCATION_PRINT(File, Loc) ((void) 0)
<br> # endif
<br> #endif
<br>+]], [[
<br>+/* This macro is provided for backwards compatibility. */
<br>
<br>+#ifndef YY_LOCATION_PRINT
<br>+# define YY_LOCATION_PRINT(File, Loc) ((void) 0)
<br>+#endif
<br>+]])[
<br>
<br> /* YYLEX -- calling `yylex' with the right arguments. */
<br>
<br>@@ -1192,7 +1198,7 @@ b4_c_function_def([[yyparse]], [[int]], b4_parse_param)[
<br> yyvsp = yyvs;]b4_locations_if([[
<br> yylsp = yyls;
<br>
<br>-#if YYLTYPE_IS_TRIVIAL
<br>+#if defined YYLTYPE_IS_TRIVIAL && YYLTYPE_IS_TRIVIAL
<br> /* Initialize the default location before parsing starts. */
<br> yylloc.first_line = yylloc.last_line = ]b4_location_initial_line[;
<br> yylloc.first_column = yylloc.last_column = ]b4_location_initial_column[;
<br>--
<br>1.6.5.4
<br><br><br><br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Patches-f1505.html" embed="fixTarget[1505]" target="_top" >Gnu - Bison - Patches</a></p>tag:old.nabble.com,2006:post-26805166Re: several messages2009-12-15T18:10:14Z2009-12-15T18:10:14ZJonathan NiederJoel E. Denny wrote:
<br><br>> If you have time, please do the following:
<br>>
<br>> 1. Address the comments I've inlined in your patches below.
<br>>
<br>> 2. Write a ChangeLog entry. One for all of the changes you posted seems
<br>> fine. Please try to follow the format you see in existing ChangeLog
<br>> entries.
<br><br>Sure thing. Thanks for the tips.
<br><br>> 3. After committing in your local repo, email us the output of:
<br>>
<br>> git format-patch --stdout -1
<br>>
<br>> That will make your patch easier to apply.
<br><br>I am not sure I understand this one. Do you mean that I should keep
<br>the "From " mbox delimiter and From:, Date:, etc fields inline in my
<br>message?
<br><br>> On Sat, 12 Dec 2009, Jonathan Nieder wrote:
<br><div class='shrinkable-quote'><br>>> diff --git a/src/parse-gram.c b/src/parse-gram.c
<br>>> index 4b6d559..0f509e6 100644
<br>>> --- a/src/parse-gram.c
<br>>> +++ b/src/parse-gram.c
<br>>
<br>>> diff --git a/src/parse-gram.h b/src/parse-gram.h
<br>>> index 894b5dd..fcf3255 100644
<br>>> --- a/src/parse-gram.h
<br>>> +++ b/src/parse-gram.h
<br>>
<br>> When posting patches, it's best if you remove all changes that were
<br>> automatically generated during the build. We can regenerate those
<br>> ourselves, and we don't want to read them while reviewing your patches.
</div><br>That patch consisted only of automatically generated changes. I’ll
<br>omit it and combine the other two when I reroll.
<br><div class='shrinkable-quote'><br>>> --- a/data/yacc.c
<br>>> +++ b/data/yacc.c
<br>>> @@ -668,7 +668,7 @@ while (YYID (0))
<br>>> we won't break user code: when these are the locations we know. */
<br>>>
<br>>> #ifndef YY_LOCATION_PRINT
<br>>> -# if YYLTYPE_IS_TRIVIAL
<br>>> +]b4_locations_if([[# if defined YYLTYPE_IS_TRIVIAL && YYLTYPE_IS_TRIVIAL
<br>>> # define YY_LOCATION_PRINT(File, Loc) \
<br>>> fprintf (File, "%d.%d-%d.%d", \
<br>>> (Loc).first_line, (Loc).first_column, \
<br>>> @@ -676,7 +676,9 @@ while (YYID (0))
<br>>> # else
<br>>> # define YY_LOCATION_PRINT(File, Loc) ((void) 0)
<br>>> # endif
<br>>> -#endif
<br>>> +]], [[ /* This macro is maintained here in case user code is relying on it. */
<br>>> +# define YY_LOCATION_PRINT(File, Loc) ((void) 0)
<br>>> +]])[#endif
<br>>
<br>> How does the b4_locations_if part of this change help?
</div><br>Not much, admittedly. Here is why I did it anyway:
<br><br> 1. It makes sure that YYLTYPE_IS_TRIVIAL does not get used if
<br> locations are turned off. Since that symbol is in yacc’s
<br> namespace anyway, that is not a big deal. But every other
<br> use of YYLTYPE_IS_TRIVIAL is already protected, and for peace
<br> of mind I think that is worth maintaining.
<br><br> In particular, it is a little clearer that this preprocessor
<br> symbol is not the cause of any problem that appears with
<br> locations turned off when grepping for YYLTPYE_IS_TRIVIAL in
<br> the generated parser yields no matches.
<br><br> 2. It makes the generated y.tab.c simpler if locations are turned
<br> off. I do think that making the y.tab.c less mysterious when
<br> possible is worth while.
<br><br> 3. It makes what happens in the locations-turned-off case a little
<br> clearer. In my opinion, except for compatibility reasons, it
<br> doesn’t make sense to define the YY_LOCATION_PRINT macro when
<br> locations are turned off. Writing the code this way clarifies
<br> that we are aware we are defining it and why.
<br><br>On the other hand, it makes the source uglier. So I’m not sure what
<br>the right thing to do is.
<br><br>Jonathan
<br><br><br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Patches-f1505.html" embed="fixTarget[1505]" target="_top" >Gnu - Bison - Patches</a></p>tag:old.nabble.com,2006:post-26804579Re: several messages2009-12-15T16:58:31Z2009-12-15T16:58:31ZJoel E. Denny-4On Sat, 12 Dec 2009, Jonathan Nieder wrote:
<br><div class='shrinkable-quote'><br>> I noticed a few warnings when building mawk:
<br>>
<br>> compiling parse.c
<br>> y.tab.c:383:6: warning: "YYENABLE_NLS" is not defined
<br>> y.tab.c:1271:6: warning: "YYLTYPE_IS_TRIVIAL" is not defined
<br>>
<br>> These warnings are not revealing any actual problem (it is perfectly
<br>> reasonable to use #if FOO even if FOO is not defined). They look
<br>> simple to fix, and a quick Google search shows all kinds of projects
<br>> unnecessarily defining these symbols to work around the warnings, so I
<br>> figure, why not suppress them here at the source?
</div><br>Sounds good to me.
<br><br>> These two patches do that. Perhaps they could be useful. The result
<br>> still passes the test suite here.
<br><br>Thanks.
<br><br>> Jonathan Nieder (2):
<br>> Check if YYENABLE_NLS is defined before checking its value
<br>> Do not depend on YYLTYPE_IS_TRIVIAL unless locations are requested
<br>>
<br>> data/glr.c | 4 ++--
<br>> data/lalr1.cc | 2 +-
<br>> data/yacc.c | 10 ++++++----
<br>> 3 files changed, 9 insertions(+), 7 deletions(-)
<br><br>Good, you're using git. If you have time, please do the following:
<br><br>1. Address the comments I've inlined in your patches below.
<br><br>2. Write a ChangeLog entry. One for all of the changes you posted seems
<br>fine. Please try to follow the format you see in existing ChangeLog
<br>entries.
<br><br>3. After committing in your local repo, email us the output of:
<br><br> git format-patch --stdout -1
<br><br>That will make your patch easier to apply.
<br><br>> diff --git a/src/parse-gram.c b/src/parse-gram.c
<br>> index 4b6d559..0f509e6 100644
<br>> --- a/src/parse-gram.c
<br>> +++ b/src/parse-gram.c
<br><br>> diff --git a/src/parse-gram.h b/src/parse-gram.h
<br>> index 894b5dd..fcf3255 100644
<br>> --- a/src/parse-gram.h
<br>> +++ b/src/parse-gram.h
<br><br>When posting patches, it's best if you remove all changes that were
<br>automatically generated during the build. We can regenerate those
<br>ourselves, and we don't want to read them while reviewing your patches.
<br><br>> Also check if YYLTYPE_IS_TRIVIAL is defined when using its value,
<br>> to avoid warnings if it is not defined.
<br>> ---
<br>> I could not tell from the documentation whether YY_LOCATION_PRINT
<br>> needs to be defined when locations are disabled. This patches
<br>> leaves it in to be safe.
<br><div class='shrinkable-quote'><br>> diff --git a/data/yacc.c b/data/yacc.c
<br>> index af38314..0d171a7 100644
<br>> --- a/data/yacc.c
<br>> +++ b/data/yacc.c
<br>> @@ -668,7 +668,7 @@ while (YYID (0))
<br>> we won't break user code: when these are the locations we know. */
<br>>
<br>> #ifndef YY_LOCATION_PRINT
<br>> -# if YYLTYPE_IS_TRIVIAL
<br>> +]b4_locations_if([[# if defined YYLTYPE_IS_TRIVIAL && YYLTYPE_IS_TRIVIAL
<br>> # define YY_LOCATION_PRINT(File, Loc) \
<br>> fprintf (File, "%d.%d-%d.%d", \
<br>> (Loc).first_line, (Loc).first_column, \
<br>> @@ -676,7 +676,9 @@ while (YYID (0))
<br>> # else
<br>> # define YY_LOCATION_PRINT(File, Loc) ((void) 0)
<br>> # endif
<br>> -#endif
<br>> +]], [[ /* This macro is maintained here in case user code is relying on it. */
<br>> +# define YY_LOCATION_PRINT(File, Loc) ((void) 0)
<br>> +]])[#endif
</div><br>How does the b4_locations_if part of this change help? If it isn't really
<br>necessary, then let's drop it. As far as I can tell, the addition of
<br>`defined YYLTYPE_IS_TRIVIAL &&' is sufficient.
<br><br>Thanks again.
<br><br><br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Patches-f1505.html" embed="fixTarget[1505]" target="_top" >Gnu - Bison - Patches</a></p>tag:old.nabble.com,2006:post-26804260Re: Non-portable m4 check in autoconf 2.64, 2.65, HEAD2009-12-15T16:18:01Z2009-12-15T16:18:01ZJoel E. Denny-4Hi Eric,
<br><br>On Fri, 27 Nov 2009, Eric Blake wrote:
<br><br>> According to Harald van Dijk on 11/26/2009 11:32 AM:
<br>> > The configure script of autoconf 2.64/2.65/HEAD fails to detect a
<br>> > suitable m4 on my system.
<br>> > I have GNU M4 1.4.13. The problem is that I'm using dash for /bin/sh
<br>> > (and my dash supports LINENO so configure doesn't silently use bash
<br>> > instead), and the m4 detection doesn't work when echo \1 prints ^A.
<br>>
<br>> That means bison needs to update its autoconf submodule to today, or it
<br>> will suffer from the same problem in its reuse of the broken m4.m4 test.
<br><br>Thanks. I pushed the first patch below to master and branch-2.5. I
<br>pushed the second to branch-2.4.2.
<br><br>From c843aaab352ab1f40c7d218fb5bb2c6fd6fd3a88 Mon Sep 17 00:00:00 2001
<br>From: Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26804260&i=0" target="_top" rel="nofollow">jdenny@...</a>>
<br>Date: Tue, 15 Dec 2009 18:56:52 -0500
<br>Subject: [PATCH] autoconf: update to latest for fix of M4 detection.
<br><br>Reported by Eric Blake.
<br>* submodules/autoconf: Update.
<br>---
<br> ChangeLog | 6 ++++++
<br> submodules/autoconf | 2 +-
<br> 2 files changed, 7 insertions(+), 1 deletions(-)
<br><br>diff --git a/ChangeLog b/ChangeLog
<br>index 897b54a..2c5d33e 100644
<br>--- a/ChangeLog
<br>+++ b/ChangeLog
<br>@@ -1,5 +1,11 @@
<br> 2009-12-15 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26804260&i=1" target="_top" rel="nofollow">jdenny@...</a>>
<br>
<br>+ autoconf: update to latest for fix of M4 detection.
<br>+ Reported by Eric Blake.
<br>+ * submodules/autoconf: Update.
<br>+
<br>+2009-12-15 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26804260&i=2" target="_top" rel="nofollow">jdenny@...</a>>
<br>+
<br> portability: use -DGNULIB_POSIXCHECK.
<br> Reported by Eric Blake. See discussions at
<br> <<a href="http://lists.gnu.org/archive/html/bug-bison/2009-09/msg00008.html" target="_top" rel="nofollow">http://lists.gnu.org/archive/html/bug-bison/2009-09/msg00008.html</a>>
<br>diff --git a/submodules/autoconf b/submodules/autoconf
<br>index 936160f..81074ea 160000
<br>--- a/submodules/autoconf
<br>+++ b/submodules/autoconf
<br>@@ -1 +1 @@
<br>-Subproject commit 936160f7619cb7fddbc424bb5cb9a2f997d8e166
<br>+Subproject commit 81074eaa201a289d11b6be05e0c78afbe5ff79f5
<br>--
<br>1.5.4.3
<br><br>From dfff6628178e75cc2d23ad574f14bc540249258c Mon Sep 17 00:00:00 2001
<br>From: Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26804260&i=3" target="_top" rel="nofollow">jdenny@...</a>>
<br>Date: Tue, 15 Dec 2009 19:11:42 -0500
<br>Subject: [PATCH] Import M4 detection fix from Autoconf.
<br><br>Reported by Eric Blake.
<br>* m4/m4.m4: Copy serial 10.
<br>---
<br> ChangeLog | 6 ++++++
<br> m4/m4.m4 | 4 ++--
<br> 2 files changed, 8 insertions(+), 2 deletions(-)
<br><br>diff --git a/ChangeLog b/ChangeLog
<br>index c6a2750..d419d50 100644
<br>--- a/ChangeLog
<br>+++ b/ChangeLog
<br>@@ -1,3 +1,9 @@
<br>+2009-12-15 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26804260&i=4" target="_top" rel="nofollow">jdenny@...</a>>
<br>+
<br>+ Import M4 detection fix from Autoconf.
<br>+ Reported by Eric Blake.
<br>+ * m4/m4.m4: Copy serial 10.
<br>+
<br> 2009-12-14 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26804260&i=5" target="_top" rel="nofollow">jdenny@...</a>>
<br>
<br> Enable assertion output and --disable-assert for configure.
<br>diff --git a/m4/m4.m4 b/m4/m4.m4
<br>index 096fb28..c6b64b0 100644
<br>--- a/m4/m4.m4
<br>+++ b/m4/m4.m4
<br>@@ -1,4 +1,4 @@
<br>-# m4.m4 serial 9
<br>+# m4.m4 serial 10
<br>
<br> # Copyright (C) 2000, 2006, 2007, 2008, 2009 Free Software Foundation,
<br> # Inc.
<br>@@ -29,7 +29,7 @@ AC_PATH_PROGS_FEATURE_CHECK([M4], [m4 gm4 gnum4],
<br> ac_snippet=change'quote(<,>)in''dir(<if''def>,mac,bug)'
<br> ac_snippet=${ac_snippet}pat'subst(a,\(b\)\|\(a\),\1)d'nl
<br> test -z "`$ac_path_M4 -F conftest.m4f </dev/null 2>&1`" \
<br>- && test -z "`echo $ac_snippet | $ac_path_M4 --trace=mac 2>&1`" \
<br>+ && test -z "`AS_ECHO([$ac_snippet]) | $ac_path_M4 --trace=mac 2>&1`" \
<br> && test -f conftest.m4f \
<br> && ac_cv_path_M4=$ac_path_M4 ac_path_M4_found=:
<br> rm -f conftest.m4f],
<br>--
<br>1.5.4.3
<br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Bugs-f1503.html" embed="fixTarget[1503]" target="_top" >Gnu - Bison - Bugs</a></p>tag:old.nabble.com,2006:post-26803846Re: [PATCH] build: avoid printf format mismatch warnings2009-12-15T15:33:43Z2009-12-15T15:33:43ZJoel E. Denny-4On Sun, 11 Oct 2009, Joel E. Denny wrote:
<br><br>> <a href="http://lists.gnu.org/archive/html/bug-gnulib/2009-10/msg00108.html" target="_top" rel="nofollow">http://lists.gnu.org/archive/html/bug-gnulib/2009-10/msg00108.html</a><br>>
<br>> Based on that discussion, I'm thinking of pushing the following. I'm
<br>> hesitating for a few reasons:
<br>>
<br>> 1. I'm not sure why bootstrap.conf was careful to exclude
<br>> m4/printf-posix.m4. Maybe to avoid bloat?
<br><br>Given that the comments say "we don't need them", I'm assuming the point
<br>was to avoid bloat. However, -DGNULIB_POSIXCHECK recommends printf-posix,
<br>so I'm removing the exclusion of m4/printf-posix.m4.
<br><br>> 2. -DGNULIB_POSIXCHECK suggests realloc-posix for the sake of
<br>> vasnprintf.c, so it seems realloc-posix ought to be a dependency of
<br>> vasnprintf-posix. I'll explore further later and maybe take it to the
<br>> gnulib maintainers.
<br><br>On second thought, maybe realloc-posix doesn't fix any feature of realloc
<br>actually needed by vasnprintf-posix, but -DGNULIB_POSIXCHECK of course
<br>isn't precise enough to realize that. I'm not going to pursue this with
<br>the gnulib maintainers.
<br><br>> 3. -DGNULIB_POSIXCHECK is suggesting a lot of other modules. We need to
<br>> explore those at some point.
<br><br>Later.
<br><br>> 4. I'm trying to decide whether to apply the patch to branch-2.4.2.
<br><br>Because I really don't know what I'm doing, I'm not going to risk that.
<br><br>I pushed the following to master and branch-2.5. It extends the patch I
<br>proposed in the previous email by linking gnulib libraries in the test
<br>suite parsers. Some tests were failing without that.
<br><br>>From 5c99151aeb4396bf35b482e27075948c33088afa Mon Sep 17 00:00:00 2001
<br>From: Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26803846&i=0" target="_top" rel="nofollow">jdenny@...</a>>
<br>Date: Tue, 15 Dec 2009 00:15:41 -0500
<br>Subject: [PATCH] portability: use -DGNULIB_POSIXCHECK.
<br><br>Reported by Eric Blake. See discussions at
<br><<a href="http://lists.gnu.org/archive/html/bug-bison/2009-09/msg00008.html" target="_top" rel="nofollow">http://lists.gnu.org/archive/html/bug-bison/2009-09/msg00008.html</a>>
<br>and
<br><<a href="http://lists.gnu.org/archive/html/bug-gnulib/2009-10/msg00108.html" target="_top" rel="nofollow">http://lists.gnu.org/archive/html/bug-gnulib/2009-10/msg00108.html</a>>.
<br>* HACKING (Release checks): Suggest -DGNULIB_POSIXCHECK.
<br>* bootstrap.conf (gnulib_modules): Add all the printf modules
<br>suggested by -DGNULIB_POSIXCHECK. Add realloc-posix as
<br>suggested by -DGNULIB_POSIXCHECK for gnulib's own vasnprintf.c.
<br>(excluded_files): Remove m4/printf-posix.m4.
<br>* tests/atlocal.in (LIBS): As for LDADD in src/local.mk, add
<br>lib/libbison.a so gnulib libraries can be linked.
<br>---
<br> ChangeLog | 15 +++++++++++++++
<br> HACKING | 3 +++
<br> bootstrap.conf | 11 +++++++----
<br> lib/.cvsignore | 42 ++++++++++++++++++++++++++++++++++++++++++
<br> lib/.gitignore | 42 ++++++++++++++++++++++++++++++++++++++++++
<br> m4/.cvsignore | 33 +++++++++++++++++++++++++++++++++
<br> m4/.gitignore | 33 +++++++++++++++++++++++++++++++++
<br> tests/atlocal.in | 2 +-
<br> 8 files changed, 176 insertions(+), 5 deletions(-)
<br><br>diff --git a/ChangeLog b/ChangeLog
<br>index 2c4baf4..897b54a 100644
<br>--- a/ChangeLog
<br>+++ b/ChangeLog
<br>@@ -1,5 +1,20 @@
<br> 2009-12-15 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26803846&i=1" target="_top" rel="nofollow">jdenny@...</a>>
<br>
<br>+ portability: use -DGNULIB_POSIXCHECK.
<br>+ Reported by Eric Blake. See discussions at
<br>+ <<a href="http://lists.gnu.org/archive/html/bug-bison/2009-09/msg00008.html" target="_top" rel="nofollow">http://lists.gnu.org/archive/html/bug-bison/2009-09/msg00008.html</a>>
<br>+ and
<br>+ <<a href="http://lists.gnu.org/archive/html/bug-gnulib/2009-10/msg00108.html" target="_top" rel="nofollow">http://lists.gnu.org/archive/html/bug-gnulib/2009-10/msg00108.html</a>>.
<br>+ * HACKING (Release checks): Suggest -DGNULIB_POSIXCHECK.
<br>+ * bootstrap.conf (gnulib_modules): Add all the printf modules
<br>+ suggested by -DGNULIB_POSIXCHECK. Add realloc-posix as
<br>+ suggested by -DGNULIB_POSIXCHECK for gnulib's own vasnprintf.c.
<br>+ (excluded_files): Remove m4/printf-posix.m4.
<br>+ * tests/atlocal.in (LIBS): As for LDADD in src/local.mk, add
<br>+ lib/libbison.a so gnulib libraries can be linked.
<br>+
<br>+2009-12-15 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26803846&i=2" target="_top" rel="nofollow">jdenny@...</a>>
<br>+
<br> gnulib: update for fix of fprintf-posix, which we'll use soon.
<br> * etc/prefix-gnulib-mk (prefix): Adjust regex for make file
<br> targets so that gnulib's new arg-nonnull.h and link-warning.h
<br>diff --git a/HACKING b/HACKING
<br>index 82f07bb..4edb98b 100644
<br>--- a/HACKING
<br>+++ b/HACKING
<br>@@ -172,6 +172,9 @@ release:
<br> that 1. Bison compiles cleanly, 2. the parsers it produces compile
<br> cleanly too.
<br>
<br>+- Build with -DGNULIB_POSIXCHECK. It suggests gnulib modules that can
<br>+ fix portability issues.
<br>+
<br> - run `make maintainer-check' which:
<br> - runs `valgrind -q bison' to run Bison under Valgrind.
<br> - runs the parsers under Valgrind.
<br>diff --git a/bootstrap.conf b/bootstrap.conf
<br>index dff8d80..4522d8c 100644
<br>--- a/bootstrap.conf
<br>+++ b/bootstrap.conf
<br>@@ -20,9 +20,13 @@ gnulib_modules='
<br> announce-gen argmatch assert config-h c-strcase configmake dirname
<br> error extensions fopen-safer gendocs getopt-gnu gettext
<br> git-version-gen hash inttypes javacomp-script javaexec-script
<br>- maintainer-makefile malloc mbswidth obstack quote quotearg stdbool
<br>- stpcpy strerror strtoul strverscmp unistd unistd-safer unlocked-io
<br>- update-copyright unsetenv verify warnings xalloc xalloc-die xstrndup
<br>+ maintainer-makefile malloc mbswidth obstack quote quotearg
<br>+ realloc-posix stdbool stpcpy strerror strtoul strverscmp unistd
<br>+ unistd-safer unlocked-io update-copyright unsetenv verify warnings
<br>+ xalloc xalloc-die xstrndup
<br>+
<br>+ fprintf-posix printf-posix snprintf-posix sprintf-posix
<br>+ vsnprintf-posix vsprintf-posix
<br> '
<br>
<br> # Additional xgettext options to use. Use "\\\newline" to break lines.
<br>@@ -52,7 +56,6 @@ excluded_files='
<br> m4/lcmessage.m4
<br> m4/lock.m4
<br> m4/longdouble.m4
<br>- m4/printf-posix.m4
<br> m4/signed.m4
<br> m4/size_max.m4
<br> m4/uintmax_t.m4
<br>diff --git a/tests/atlocal.in b/tests/atlocal.in
<br>index 9264a40..fb1cef6 100644
<br>--- a/tests/atlocal.in
<br>+++ b/tests/atlocal.in
<br>@@ -33,7 +33,7 @@ CXXFLAGS='@O0CXXFLAGS@ @WARN_CXXFLAGS@ @WERROR_CFLAGS@'
<br> LDFLAGS='@LDFLAGS@'
<br>
<br> # Are special libraries needed?
<br>-LIBS='@LIBS@ @INTLLIBS@'
<br>+LIBS="$abs_top_builddir/lib/libbison.a @LIBS@ @INTLLIBS@"
<br>
<br> # Empty if no javac was found
<br> CONF_JAVAC='@CONF_JAVAC@'
<br>--
<br>1.5.4.3
<br><br><br><br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Patches-f1505.html" embed="fixTarget[1505]" target="_top" >Gnu - Bison - Patches</a></p>tag:old.nabble.com,2006:post-26803843Re: [PATCH] build: avoid printf format mismatch warnings2009-12-15T15:33:43Z2009-12-15T15:33:43ZJoel E. Denny-4On Sun, 11 Oct 2009, Joel E. Denny wrote:
<br><br>> <a href="http://lists.gnu.org/archive/html/bug-gnulib/2009-10/msg00108.html" target="_top" rel="nofollow">http://lists.gnu.org/archive/html/bug-gnulib/2009-10/msg00108.html</a><br>>
<br>> Based on that discussion, I'm thinking of pushing the following. I'm
<br>> hesitating for a few reasons:
<br>>
<br>> 1. I'm not sure why bootstrap.conf was careful to exclude
<br>> m4/printf-posix.m4. Maybe to avoid bloat?
<br><br>Given that the comments say "we don't need them", I'm assuming the point
<br>was to avoid bloat. However, -DGNULIB_POSIXCHECK recommends printf-posix,
<br>so I'm removing the exclusion of m4/printf-posix.m4.
<br><br>> 2. -DGNULIB_POSIXCHECK suggests realloc-posix for the sake of
<br>> vasnprintf.c, so it seems realloc-posix ought to be a dependency of
<br>> vasnprintf-posix. I'll explore further later and maybe take it to the
<br>> gnulib maintainers.
<br><br>On second thought, maybe realloc-posix doesn't fix any feature of realloc
<br>actually needed by vasnprintf-posix, but -DGNULIB_POSIXCHECK of course
<br>isn't precise enough to realize that. I'm not going to pursue this with
<br>the gnulib maintainers.
<br><br>> 3. -DGNULIB_POSIXCHECK is suggesting a lot of other modules. We need to
<br>> explore those at some point.
<br><br>Later.
<br><br>> 4. I'm trying to decide whether to apply the patch to branch-2.4.2.
<br><br>Because I really don't know what I'm doing, I'm not going to risk that.
<br><br>I pushed the following to master and branch-2.5. It extends the patch I
<br>proposed in the previous email by linking gnulib libraries in the test
<br>suite parsers. Some tests were failing without that.
<br><br>>From 5c99151aeb4396bf35b482e27075948c33088afa Mon Sep 17 00:00:00 2001
<br>From: Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26803843&i=0" target="_top" rel="nofollow">jdenny@...</a>>
<br>Date: Tue, 15 Dec 2009 00:15:41 -0500
<br>Subject: [PATCH] portability: use -DGNULIB_POSIXCHECK.
<br><br>Reported by Eric Blake. See discussions at
<br><<a href="http://lists.gnu.org/archive/html/bug-bison/2009-09/msg00008.html" target="_top" rel="nofollow">http://lists.gnu.org/archive/html/bug-bison/2009-09/msg00008.html</a>>
<br>and
<br><<a href="http://lists.gnu.org/archive/html/bug-gnulib/2009-10/msg00108.html" target="_top" rel="nofollow">http://lists.gnu.org/archive/html/bug-gnulib/2009-10/msg00108.html</a>>.
<br>* HACKING (Release checks): Suggest -DGNULIB_POSIXCHECK.
<br>* bootstrap.conf (gnulib_modules): Add all the printf modules
<br>suggested by -DGNULIB_POSIXCHECK. Add realloc-posix as
<br>suggested by -DGNULIB_POSIXCHECK for gnulib's own vasnprintf.c.
<br>(excluded_files): Remove m4/printf-posix.m4.
<br>* tests/atlocal.in (LIBS): As for LDADD in src/local.mk, add
<br>lib/libbison.a so gnulib libraries can be linked.
<br>---
<br> ChangeLog | 15 +++++++++++++++
<br> HACKING | 3 +++
<br> bootstrap.conf | 11 +++++++----
<br> lib/.cvsignore | 42 ++++++++++++++++++++++++++++++++++++++++++
<br> lib/.gitignore | 42 ++++++++++++++++++++++++++++++++++++++++++
<br> m4/.cvsignore | 33 +++++++++++++++++++++++++++++++++
<br> m4/.gitignore | 33 +++++++++++++++++++++++++++++++++
<br> tests/atlocal.in | 2 +-
<br> 8 files changed, 176 insertions(+), 5 deletions(-)
<br><br>diff --git a/ChangeLog b/ChangeLog
<br>index 2c4baf4..897b54a 100644
<br>--- a/ChangeLog
<br>+++ b/ChangeLog
<br>@@ -1,5 +1,20 @@
<br> 2009-12-15 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26803843&i=1" target="_top" rel="nofollow">jdenny@...</a>>
<br>
<br>+ portability: use -DGNULIB_POSIXCHECK.
<br>+ Reported by Eric Blake. See discussions at
<br>+ <<a href="http://lists.gnu.org/archive/html/bug-bison/2009-09/msg00008.html" target="_top" rel="nofollow">http://lists.gnu.org/archive/html/bug-bison/2009-09/msg00008.html</a>>
<br>+ and
<br>+ <<a href="http://lists.gnu.org/archive/html/bug-gnulib/2009-10/msg00108.html" target="_top" rel="nofollow">http://lists.gnu.org/archive/html/bug-gnulib/2009-10/msg00108.html</a>>.
<br>+ * HACKING (Release checks): Suggest -DGNULIB_POSIXCHECK.
<br>+ * bootstrap.conf (gnulib_modules): Add all the printf modules
<br>+ suggested by -DGNULIB_POSIXCHECK. Add realloc-posix as
<br>+ suggested by -DGNULIB_POSIXCHECK for gnulib's own vasnprintf.c.
<br>+ (excluded_files): Remove m4/printf-posix.m4.
<br>+ * tests/atlocal.in (LIBS): As for LDADD in src/local.mk, add
<br>+ lib/libbison.a so gnulib libraries can be linked.
<br>+
<br>+2009-12-15 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26803843&i=2" target="_top" rel="nofollow">jdenny@...</a>>
<br>+
<br> gnulib: update for fix of fprintf-posix, which we'll use soon.
<br> * etc/prefix-gnulib-mk (prefix): Adjust regex for make file
<br> targets so that gnulib's new arg-nonnull.h and link-warning.h
<br>diff --git a/HACKING b/HACKING
<br>index 82f07bb..4edb98b 100644
<br>--- a/HACKING
<br>+++ b/HACKING
<br>@@ -172,6 +172,9 @@ release:
<br> that 1. Bison compiles cleanly, 2. the parsers it produces compile
<br> cleanly too.
<br>
<br>+- Build with -DGNULIB_POSIXCHECK. It suggests gnulib modules that can
<br>+ fix portability issues.
<br>+
<br> - run `make maintainer-check' which:
<br> - runs `valgrind -q bison' to run Bison under Valgrind.
<br> - runs the parsers under Valgrind.
<br>diff --git a/bootstrap.conf b/bootstrap.conf
<br>index dff8d80..4522d8c 100644
<br>--- a/bootstrap.conf
<br>+++ b/bootstrap.conf
<br>@@ -20,9 +20,13 @@ gnulib_modules='
<br> announce-gen argmatch assert config-h c-strcase configmake dirname
<br> error extensions fopen-safer gendocs getopt-gnu gettext
<br> git-version-gen hash inttypes javacomp-script javaexec-script
<br>- maintainer-makefile malloc mbswidth obstack quote quotearg stdbool
<br>- stpcpy strerror strtoul strverscmp unistd unistd-safer unlocked-io
<br>- update-copyright unsetenv verify warnings xalloc xalloc-die xstrndup
<br>+ maintainer-makefile malloc mbswidth obstack quote quotearg
<br>+ realloc-posix stdbool stpcpy strerror strtoul strverscmp unistd
<br>+ unistd-safer unlocked-io update-copyright unsetenv verify warnings
<br>+ xalloc xalloc-die xstrndup
<br>+
<br>+ fprintf-posix printf-posix snprintf-posix sprintf-posix
<br>+ vsnprintf-posix vsprintf-posix
<br> '
<br>
<br> # Additional xgettext options to use. Use "\\\newline" to break lines.
<br>@@ -52,7 +56,6 @@ excluded_files='
<br> m4/lcmessage.m4
<br> m4/lock.m4
<br> m4/longdouble.m4
<br>- m4/printf-posix.m4
<br> m4/signed.m4
<br> m4/size_max.m4
<br> m4/uintmax_t.m4
<br>diff --git a/tests/atlocal.in b/tests/atlocal.in
<br>index 9264a40..fb1cef6 100644
<br>--- a/tests/atlocal.in
<br>+++ b/tests/atlocal.in
<br>@@ -33,7 +33,7 @@ CXXFLAGS='@O0CXXFLAGS@ @WARN_CXXFLAGS@ @WERROR_CFLAGS@'
<br> LDFLAGS='@LDFLAGS@'
<br>
<br> # Are special libraries needed?
<br>-LIBS='@LIBS@ @INTLLIBS@'
<br>+LIBS="$abs_top_builddir/lib/libbison.a @LIBS@ @INTLLIBS@"
<br>
<br> # Empty if no javac was found
<br> CONF_JAVAC='@CONF_JAVAC@'
<br>--
<br>1.5.4.3
<br><br><br><br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Bugs-f1503.html" embed="fixTarget[1503]" target="_top" >Gnu - Bison - Bugs</a></p>tag:old.nabble.com,2006:post-26802922[PATCH] gnulib: update for fix of fprintf-posix, which we'll use soon.2009-12-15T14:26:23Z2009-12-15T14:26:23ZJoel E. Denny-4I pushed this to master.
<br><br>From 75f94c6ddfe609b1b4dca76ea447b6271593265e Mon Sep 17 00:00:00 2001
<br>From: Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26802922&i=0" target="_top" rel="nofollow">jdenny@...</a>>
<br>Date: Tue, 15 Dec 2009 17:20:31 -0500
<br>Subject: [PATCH] gnulib: update for fix of fprintf-posix, which we'll use soon.
<br><br>* etc/prefix-gnulib-mk (prefix): Adjust regex for make file
<br>targets so that gnulib's new arg-nonnull.h and link-warning.h
<br>are matched.
<br>* gnulib: Update.
<br>---
<br> ChangeLog | 8 ++++++++
<br> build-aux/.cvsignore | 1 +
<br> build-aux/.gitignore | 1 +
<br> etc/prefix-gnulib-mk | 2 +-
<br> gnulib | 2 +-
<br> lib/.cvsignore | 5 +++++
<br> lib/.gitignore | 5 +++++
<br> m4/.cvsignore | 3 +++
<br> m4/.gitignore | 3 +++
<br> 9 files changed, 28 insertions(+), 2 deletions(-)
<br><br>diff --git a/ChangeLog b/ChangeLog
<br>index e465744..2c4baf4 100644
<br>--- a/ChangeLog
<br>+++ b/ChangeLog
<br>@@ -1,3 +1,11 @@
<br>+2009-12-15 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26802922&i=1" target="_top" rel="nofollow">jdenny@...</a>>
<br>+
<br>+ gnulib: update for fix of fprintf-posix, which we'll use soon.
<br>+ * etc/prefix-gnulib-mk (prefix): Adjust regex for make file
<br>+ targets so that gnulib's new arg-nonnull.h and link-warning.h
<br>+ are matched.
<br>+ * gnulib: Update.
<br>+
<br> 2009-12-14 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26802922&i=2" target="_top" rel="nofollow">jdenny@...</a>>
<br>
<br> Enable assertion output and --disable-assert for configure.
<br>diff --git a/etc/prefix-gnulib-mk b/etc/prefix-gnulib-mk
<br>index e969663..1bff69b 100755
<br>--- a/etc/prefix-gnulib-mk
<br>+++ b/etc/prefix-gnulib-mk
<br>@@ -90,7 +90,7 @@ sub prefix ($)
<br> # Prefix all the occurrence of files in rules. If there is nothing
<br> # after in the :, it's probably a phony target, or a suffix rule.
<br> # Don't touch it.
<br>- s{^([\w./]+ *: *\w.*)$}
<br>+ s{^([-\w/]+\.[-\w.]+ *: *\S.*)$}
<br> {prefix_words($1)}gem;
<br>
<br> # Prefix files in variables.
<br>diff --git a/gnulib b/gnulib
<br>index 3ef6401..3fd9a2d 160000
<br>--- a/gnulib
<br>+++ b/gnulib
<br>@@ -1 +1 @@
<br>-Subproject commit 3ef64012bf653379fda629f5a6b619be4c1f2e69
<br>+Subproject commit 3fd9a2df88c558b27be3a585ca2978dbf9b5da53
<br><br><br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Patches-f1505.html" embed="fixTarget[1505]" target="_top" >Gnu - Bison - Patches</a></p>tag:old.nabble.com,2006:post-26789366Re: AT_DATA and AS_ECHO_N2009-12-14T20:33:39Z2009-12-14T20:33:39ZEric Blake-----BEGIN PGP SIGNED MESSAGE-----
<br>Hash: SHA1
<br><br>According to Joel E. Denny on 12/14/2009 2:53 PM:
<div class='shrinkable-quote'><br>>>> I've been meaning to add a new macro to autotest to allow more flexibility in
<br>>>> generating files (no trailing newline, and/or allowing shell expansions) but it
<br>>>> will not be in time for autoconf 2.64. Feel free to ping the autoconf list, so
<br>>>> that I'll have a reminder email.
<br>>> Is it necessary to have a new macro for omitting the trailing newline?
<br>>> Couldn't AT_DATA detect whether a trailing newline is present in the
<br>>> expansion and then work whatever magic is necessary to print the last line
<br>>> correctly?
<br>>>
<br>>> I'll try to remember to ping again later.
<br>>
<br>> I was just clearing out some old emails and noticed that I never pinged
<br>> you about this. Have you had a chance to work on it? I don't see
<br>> anything in NEWS or the manual about it, so I assume you haven't. I'm
<br>> still interested, but I'm afraid I don't have time to help, so don't worry
<br>> about it if you don't have time either.
</div><br>Sorry, this still hasn't percolated to the top of my list (yes, I also
<br>have a backlog of old emails). But thanks for another ping.
<br><br>- --
<br>Don't work too hard, make some time for fun as well!
<br><br>Eric Blake <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26789366&i=0" target="_top" rel="nofollow">ebb9@...</a>
<br>-----BEGIN PGP SIGNATURE-----
<br>Version: GnuPG v1.4.9 (Cygwin)
<br>Comment: Public key at home.comcast.net/~ericblake/eblake.gpg
<br>Comment: Using GnuPG with Mozilla - <a href="http://enigmail.mozdev.org/" target="_top" rel="nofollow">http://enigmail.mozdev.org/</a><br><br>iEYEARECAAYFAksnEaMACgkQ84KuGfSFAYDpjQCg1dNSOnpifaBpwDDb47OUD1nJ
<br>5MwAni7md0OL/dxbW3Pel9cL6hQ/L6Cc
<br>=1IoA
<br>-----END PGP SIGNATURE-----
<br><br><br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Patches-f1505.html" embed="fixTarget[1505]" target="_top" >Gnu - Bison - Patches</a></p>tag:old.nabble.com,2006:post-26787090Re: Help with create_pipe_bidi2009-12-14T15:43:04Z2009-12-14T15:43:04ZJoel E. Denny-4On Mon, 7 Sep 2009, Joel E. Denny wrote:
<br><div class='shrinkable-quote'><br>> On Mon, 7 Sep 2009, Akim Demaille wrote:
<br>>
<br>> > I can understand the concerns of Paul wrt the space-footprint of assert,
<br>>
<br>> I'm probably showing my ignorance here, but are we talking about the size
<br>> of the bison executable? I didn't know what Paul meant by "heavyweight".
<br>>
<br>> On my system, src/bison increases from 988K to 1000K when I adjust
<br>> src/system.h as follows:
<br>>
<br>> -static inline void
<br>> -aver (bool assertion)
<br>> -{
<br>> - if (! assertion)
<br>> - abort ();
<br>> -}
<br>> -
<br>> +#include <assert.h>
<br>> +#define aver assert
<br>>
<br>> After stripping, it's 300K and 312K. Will anyone be bothered by 12K?
<br>>
<br>> > that's why I meant to twist the meaning of NDEBUG into "smaller foot-print".
<br>> > I would make --disable-asserts equivalent to NDEBUG, so it would move from
<br>> > classical assert to our current aver.
<br>> >
<br>> > We can also provide a three-state --disable-asserts:
<br>> >
<br>> > - yes: fully blown assert,
<br>> > - no: our current aver, same as NDEBUG,
<br>> > - totally: completely remove the assertions.
<br>>
<br>> During development, assert's output could prove helpful. Also, if a user
<br>> ever sees an assertion failure, it might be nice that he can report the
<br>> assert output to bug-bison. I can also understand the motivation to
<br>> suppress NDEBUG's effect. So, assert ignoring NDEBUG seems good to me.
<br>> I don't yet see much motivation for other options, but I'm not opposed to
<br>> bison providing them if you think people would find them useful.
</div><br>I want assert's output as it can facilitate development and bug reporting,
<br>so I pushed the following to master, branch-2.5, and branch-2.4.2. I
<br>decided to simply do `#define aver assert'. I also added gnulib's assert
<br>module. We can address the remaining concerns later once we've figured
<br>out what's really needed.
<br><br>Does anyone feel that these changes are NEWS-worthy?
<br><br>>From ca01f454f01645533622a59ef0df335ee07679e5 Mon Sep 17 00:00:00 2001
<br>From: Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26787090&i=0" target="_top" rel="nofollow">jdenny@...</a>>
<br>Date: Mon, 14 Dec 2009 18:04:57 -0500
<br>Subject: [PATCH] Enable assertion output and --disable-assert for configure.
<br><br>* bootstrap.conf (gnulib_modules): Add assert module.
<br>* src/system.h (aver): Define as assert, and summarize the
<br>discussion on this issue.
<br>---
<br> ChangeLog | 7 +++++++
<br> bootstrap.conf | 12 ++++++------
<br> m4/.cvsignore | 1 +
<br> m4/.gitignore | 1 +
<br> src/system.h | 25 +++++++++++++++++--------
<br> 5 files changed, 32 insertions(+), 14 deletions(-)
<br><br>diff --git a/ChangeLog b/ChangeLog
<br>index 549b084..e465744 100644
<br>--- a/ChangeLog
<br>+++ b/ChangeLog
<br>@@ -1,5 +1,12 @@
<br> 2009-12-14 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26787090&i=1" target="_top" rel="nofollow">jdenny@...</a>>
<br>
<br>+ Enable assertion output and --disable-assert for configure.
<br>+ * bootstrap.conf (gnulib_modules): Add assert module.
<br>+ * src/system.h (aver): Define as assert, and summarize the
<br>+ discussion on this issue.
<br>+
<br>+2009-12-14 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26787090&i=2" target="_top" rel="nofollow">jdenny@...</a>>
<br>+
<br> Expand GLR acronym in summary of Bison.
<br> Based on discussion with Akim Demaille starting at
<br> <<a href="http://lists.gnu.org/archive/html/bison-patches/2009-04/msg00087.html" target="_top" rel="nofollow">http://lists.gnu.org/archive/html/bison-patches/2009-04/msg00087.html</a>>.
<br>diff --git a/bootstrap.conf b/bootstrap.conf
<br>index 1cef2f0..dff8d80 100644
<br>--- a/bootstrap.conf
<br>+++ b/bootstrap.conf
<br>@@ -17,12 +17,12 @@
<br>
<br> # gnulib modules used by this package.
<br> gnulib_modules='
<br>- announce-gen argmatch config-h c-strcase configmake dirname error
<br>- extensions fopen-safer gendocs getopt-gnu gettext git-version-gen hash
<br>- inttypes javacomp-script javaexec-script maintainer-makefile malloc
<br>- mbswidth obstack quote quotearg stdbool stpcpy strerror strtoul
<br>- strverscmp unistd unistd-safer unlocked-io update-copyright unsetenv
<br>- verify warnings xalloc xalloc-die xstrndup
<br>+ announce-gen argmatch assert config-h c-strcase configmake dirname
<br>+ error extensions fopen-safer gendocs getopt-gnu gettext
<br>+ git-version-gen hash inttypes javacomp-script javaexec-script
<br>+ maintainer-makefile malloc mbswidth obstack quote quotearg stdbool
<br>+ stpcpy strerror strtoul strverscmp unistd unistd-safer unlocked-io
<br>+ update-copyright unsetenv verify warnings xalloc xalloc-die xstrndup
<br> '
<br>
<br> # Additional xgettext options to use. Use "\\\newline" to break lines.
<br>diff --git a/src/system.h b/src/system.h
<br>index 8ea5653..8d2358c 100644
<br>--- a/src/system.h
<br>+++ b/src/system.h
<br>@@ -123,17 +123,26 @@ typedef size_t uintptr_t;
<br> | Assertions. |
<br> `-------------*/
<br>
<br>-/* <assert.h>'s assertions are too heavyweight, and can be disabled
<br>- too easily, so use aver rather than assert. See discussions at
<br>+/* In the past, Bison defined aver to simply invoke abort in the case of
<br>+ a failed assertion. The rationale was that <assert.h>'s assertions
<br>+ were too heavyweight and could be disabled too easily. See
<br>+ discussions at
<br> <<a href="http://lists.gnu.org/archive/html/bison-patches/2006-01/msg00080.html" target="_top" rel="nofollow">http://lists.gnu.org/archive/html/bison-patches/2006-01/msg00080.html</a>>
<br> <<a href="http://lists.gnu.org/archive/html/bison-patches/2006-09/msg00111.html" target="_top" rel="nofollow">http://lists.gnu.org/archive/html/bison-patches/2006-09/msg00111.html</a>>.
<br>+
<br>+ However, normal assert output can be helpful during development and
<br>+ in bug reports from users. Moreover, it's not clear now that
<br>+ <assert.h>'s assertions are significantly heavyweight. Finally, if
<br>+ users want to experiment with disabling assertions, it's debatable
<br>+ whether it's our responsibility to stop them. See discussion
<br>+ starting at
<br>+ <<a href="http://lists.gnu.org/archive/html/bison-patches/2009-09/msg00013.html" target="_top" rel="nofollow">http://lists.gnu.org/archive/html/bison-patches/2009-09/msg00013.html</a>>.
<br>+
<br>+ For now, we use assert but we call it aver throughout Bison in case
<br>+ we later wish to try another scheme.
<br> */
<br>-static inline void
<br>-aver (bool assertion)
<br>-{
<br>- if (! assertion)
<br>- abort ();
<br>-}
<br>+#include <assert.h>
<br>+#define aver assert
<br>
<br>
<br> /*-----------.
<br>--
<br>1.5.4.3
<br><br><br><br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Patches-f1505.html" embed="fixTarget[1505]" target="_top" >Gnu - Bison - Patches</a></p>tag:old.nabble.com,2006:post-26785624Re: AT_DATA and AS_ECHO_N (was: Re: Warn about character literals not of length one.)2009-12-14T13:53:47Z2009-12-14T13:53:47ZJoel E. Denny-4Hi Eric,
<br><br>On Fri, 24 Jul 2009, Joel E. Denny wrote:
<br><div class='shrinkable-quote'><br>> > I've been meaning to add a new macro to autotest to allow more flexibility in
<br>> > generating files (no trailing newline, and/or allowing shell expansions) but it
<br>> > will not be in time for autoconf 2.64. Feel free to ping the autoconf list, so
<br>> > that I'll have a reminder email.
<br>>
<br>> Is it necessary to have a new macro for omitting the trailing newline?
<br>> Couldn't AT_DATA detect whether a trailing newline is present in the
<br>> expansion and then work whatever magic is necessary to print the last line
<br>> correctly?
<br>>
<br>> I'll try to remember to ping again later.
</div><br>I was just clearing out some old emails and noticed that I never pinged
<br>you about this. Have you had a chance to work on it? I don't see
<br>anything in NEWS or the manual about it, so I assume you haven't. I'm
<br>still interested, but I'm afraid I don't have time to help, so don't worry
<br>about it if you don't have time either.
<br><br><br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Patches-f1505.html" embed="fixTarget[1505]" target="_top" >Gnu - Bison - Patches</a></p>tag:old.nabble.com,2006:post-26785139Re: [PATCH] Document %define lr.type and lr.default_rules.2009-12-14T13:19:17Z2009-12-14T13:19:17ZJoel E. Denny-4On Wed, 6 May 2009, Joel E. Denny wrote:
<br><div class='shrinkable-quote'><br>> On Wed, 6 May 2009, Akim Demaille wrote:
<br>>
<br>> > > Generate a deterministic LR or GLR parser employing LALR(1),
<br>> > > IELR(1), or canonical LR(1) parser tables.
<br>>
<br>> > > What do you think?
<br>> >
<br>> > I regret that we don't explicit the "generalized" word, but I see your point
<br>> > too. Yet I would say that if the reader is not expected to understand
<br>> > "generalized", he will probably not understand "canonical" either.
<br>>
<br>> I'm not sure about that. It is not clear to me that a "generalized
<br>> parser" or even a "generalized shift/reduce parser" necessarily employs
<br>> the GLR algorithm. For example, backtracking is another way to generalize
<br>> LR. However, the meaning of "canonical" from English (the sense of
<br>> "classic" or "standard") alone seems sufficient to convey the meaning of
<br>> "canonical" in "canonical LR".
<br>>
<br>> > We may also escape the LR repetition with "generate a deterministic or
<br>> > generalized shift/reduce parser employing LALR(1), IELR(1) or LR(1) parser
<br>> > tables".
<br>>
<br>> Googling/grepping for "GLR" or "canonical LR" would not return a hit for
<br>> this sentence. That loss bothers me more than repetition of "LR". What
<br>> about the following?
<br>>
<br>> "Generate a GLR (generalized LR) or deterministic shift/reduce parser
<br>> employing LALR(1), IELR(1) or canonical LR(1) parser tables."
</div><br>It's been a long time coming, but I've finally done something about this.
<br><br>I think this keeps the most important keywords while remaining fairly
<br>readable:
<br><br> Generate a deterministic LR or generalized LR (GLR) parser employing
<br> LALR(1), IELR(1), or canonical LR(1) parser tables.
<br><br>I pushed the following to branch-2.5 and master. We can always revise
<br>again later if necessary.
<br><br>From 51c7ca01e2c9cfb5c7a274bed0e14899adbd005c Mon Sep 17 00:00:00 2001
<br>From: Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26785139&i=0" target="_top" rel="nofollow">jdenny@...</a>>
<br>Date: Mon, 14 Dec 2009 16:04:00 -0500
<br>Subject: [PATCH] Expand GLR acronym in summary of Bison.
<br><br>Based on discussion with Akim Demaille starting at
<br><<a href="http://lists.gnu.org/archive/html/bison-patches/2009-04/msg00087.html" target="_top" rel="nofollow">http://lists.gnu.org/archive/html/bison-patches/2009-04/msg00087.html</a>>.
<br>* doc/bison.texinfo (Introduction): Here.
<br>* src/getargs.c (usage): Here.
<br>---
<br> ChangeLog | 8 ++++++++
<br> doc/bison.texinfo | 7 ++++---
<br> src/getargs.c | 4 ++--
<br> 3 files changed, 14 insertions(+), 5 deletions(-)
<br><br>diff --git a/ChangeLog b/ChangeLog
<br>index 03660cf..de7790c 100644
<br>--- a/ChangeLog
<br>+++ b/ChangeLog
<br>@@ -1,3 +1,11 @@
<br>+2009-12-14 Joel E. Denny <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26785139&i=1" target="_top" rel="nofollow">jdenny@...</a>>
<br>+
<br>+ Expand GLR acronym in summary of Bison.
<br>+ Based on discussion with Akim Demaille starting at
<br>+ <<a href="http://lists.gnu.org/archive/html/bison-patches/2009-04/msg00087.html" target="_top" rel="nofollow">http://lists.gnu.org/archive/html/bison-patches/2009-04/msg00087.html</a>>.
<br>+ * doc/bison.texinfo (Introduction): Here.
<br>+ * src/getargs.c (usage): Here.
<br>+
<br> 2009-10-03 Alex Rozenman <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26785139&i=2" target="_top" rel="nofollow">rozenman@...</a>>
<br>
<br> Document named references.
<br>diff --git a/doc/bison.texinfo b/doc/bison.texinfo
<br>index b0f391b..1906cd9 100644
<br>--- a/doc/bison.texinfo
<br>+++ b/doc/bison.texinfo
<br>@@ -352,9 +352,10 @@ Copying This Manual
<br> @cindex introduction
<br>
<br> @dfn{Bison} is a general-purpose parser generator that converts an
<br>-annotated context-free grammar into a deterministic or @acronym{GLR}
<br>-parser employing @acronym{LALR}(1), @acronym{IELR}(1), or canonical
<br>-@acronym{LR}(1) parser tables.
<br>+annotated context-free grammar into a deterministic @acronym{LR} or
<br>+generalized @acronym{LR} (@acronym{GLR}) parser employing
<br>+@acronym{LALR}(1), @acronym{IELR}(1), or canonical @acronym{LR}(1)
<br>+parser tables.
<br> Once you are proficient with Bison, you can use it to develop a wide
<br> range of language parsers, from those used in simple desk calculators to
<br> complex programming languages.
<br>diff --git a/src/getargs.c b/src/getargs.c
<br>index 4231cfc..8f7e668 100644
<br>--- a/src/getargs.c
<br>+++ b/src/getargs.c
<br>@@ -257,8 +257,8 @@ usage (int status)
<br>
<br> printf (_("Usage: %s [OPTION]... FILE\n"), program_name);
<br> fputs (_("\
<br>-Generate a deterministic or GLR parser employing LALR(1), IELR(1), or\n\
<br>-canonical LR(1) parser tables.\n\
<br>+Generate a deterministic LR or generalized LR (GLR) parser employing\n\
<br>+LALR(1), IELR(1), or canonical LR(1) parser tables.\n\
<br> \n\
<br> "), stdout);
<br>
<br>--
<br>1.5.4.3
<br><br><br><br><p>From forum: <a href="http://old.nabble.com/Gnu---Bison---Patches-f1505.html" embed="fixTarget[1505]" target="_top" >Gnu - Bison - Patches</a></p>