nightly build

[Don Cohen-12]

ended with this:
    WARNING:
    LOAD: compiled file format.fas was created by an older CLISP version and needs to be recompiled
    ;; Loading file ../src/format.lisp ...
    ;; Loaded file ../src/format.lisp
    WARNING:
    LOAD: compiled file international.fas was created by an older CLISP version and needs to be recompiled
    ;; Loading file ../src/international.lisp ...
    ;; Loaded file ../src/international.lisp
    WARNING:
    LOAD: compiled file /root/clisp/build-dir/room.fas was created by an older CLISP version and needs to be recompiled
    *** - LOAD: compiled file /root/clisp/build-dir/room.fas was created by an older CLISP version and needs to be recompiled
    Bye.
    make: *** [interpreted.mem] Error 1

I guessed that a make clean would solve the problem and it did,
but again it would seem to make sense to provide some facility
that would automate whatever is needed to allow
 cvs up
 make
to work.

BTW, it's now a month since my 32 bit machine was last able to build MT.
Last successful build was Sept 1.

------------------------------------------------------------------------------
Come build with us! The BlackBerry® Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay
ahead of the curve. Join us from November 9-12, 2009. Register now!
http://p.sf.net/sfu/devconf
_______________________________________________
clisp-devel mailing list
clisp-devel@...
https://lists.sourceforge.net/lists/listinfo/clisp-devel

[Vladimir Tzankov]

On 9/30/09, Don Cohen <don-sourceforge-xxz@...> wrote:
> BTW, it's now a month since my 32 bit machine was last able to build MT.
> Last successful build was Sept 1.

I guess this is the problem when creating linking sets.
I am testing mostly on 32 bit machine and rarely the build fails but
'make clean-modules' has always "solved" the problem for me (this is
not to say this is real solution). Can you try it?.

------------------------------------------------------------------------------
Come build with us! The BlackBerry® Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay
ahead of the curve. Join us from November 9-12, 2009. Register now!
http://p.sf.net/sfu/devconf
_______________________________________________
clisp-devel mailing list
clisp-devel@...
https://lists.sourceforge.net/lists/listinfo/clisp-devel

[Don Cohen-12]

Vladimir Tzankov writes:
 > On 9/30/09, Don Cohen <don-sourceforge-xxz@...> wrote:
 > > BTW, it's now a month since my 32 bit machine was last able to build MT.
 > > Last successful build was Sept 1.
 >
 > I guess this is the problem when creating linking sets.
 > I am testing mostly on 32 bit machine and rarely the build fails but
 > 'make clean-modules' has always "solved" the problem for me (this is
 > not to say this is real solution). Can you try it?.
didn't help:

gcc -g -O2 -W -Wswitch -Wcomment -Wpointer-arith -Wimplicit -Wreturn-type -Wmissing-declarations -Wno-sign-compare -Wno-format-nonliteral -O2 -fexpensive-optimizations -falign-functions=4 -pthread -DUNICODE -DMULTITHREAD -DPOSIX_THREADS -DDYNAMIC_FFI -DDYNAMIC_MODULES -I. -fPIC -Wl,-export-dynamic modules.o readline.o -lreadline -lncurses regexi.o libgnu_rx.a calls.o libgnu_sc.a -lm -lcrypt gettext.o lisp.a -lreadline -lncurses -ldl /usr/local/lib/libavcall.a /usr/local/lib/libcallback.a -L/usr/local/lib -lsigsegv -lc libgnu_cl.a -o lisp.run
boot/lisp.run -B . -M boot/lispinit.mem -norc -q -i i18n/preload.lisp -i syscalls/preload.lisp -i regexp/preload.lisp -x (saveinitmem "base/lispinit.mem")
;; Loading file i18n/preload.lisp ...
;; Loaded file i18n/preload.lisp
;; Loading file syscalls/preload.lisp ...
;; Loaded file syscalls/preload.lisp
;; Loading file regexp/preload.lisp ...
;; Loaded file regexp/preload.lisp
;; Wrote the memory image into base/lispinit.mem (2,015,228 bytes)
Bytes permanently allocated:            107,256
Bytes currently in use:               1,903,624
Bytes available until next GC:          522,128
1903624 ;
522128 ;
107256 ;
1 ;
131632 ;
24001
base/lisp.run -B . -M base/lispinit.mem -norc -q -i i18n/i18n -i syscalls/posix -i regexp/regexp -i readline/readline -x (saveinitmem "base/lispinit.mem")
 
*** - handle_fault error2 ! address = 0x1 not in [0x21557004,0x216b3794) !
SIGSEGV cannot be cured. Fault address = 0x1.
GC count: 0
Space collected by GC: 0 0
Run time: 0 28001
Real time: 0 99221
GC time: 0 0
Permanently allocated: 110616 bytes.
Currently in use: 2072616 bytes.
Free space: 353136 bytes.
./clisp-link: line 97: 31699 Segmentation fault      "$@"
./clisp-link: failed in /root/clisp/build-mt
make: *** [base] Error 1

------------------------------------------------------------------------------
Come build with us! The BlackBerry® Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay
ahead of the curve. Join us from November 9-12, 2009. Register now!
http://p.sf.net/sfu/devconf
_______________________________________________
clisp-devel mailing list
clisp-devel@...
https://lists.sourceforge.net/lists/listinfo/clisp-devel

[Don Cohen-12]

Vladimir Tzankov writes:
 > 'make clean-modules' has always "solved" the problem for me
In hindsight I'm not surprised that this did not work.
Recall that I also did a fresh checkout and got the same error.
Since I did build successfully on Sept 1 and not on Sept 2, I'd
expect to find the problem in the updates made between those two
times.  Is there some cvs update around then that seems related?
Perhaps I should be doing a binary search to find the relevant
update?  If you want me to do that, some instructions would be
helpful.  I don't recall off hand how to check out as of some
particular time or, better yet, as of some particular update.

------------------------------------------------------------------------------
Come build with us! The BlackBerry® Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay
ahead of the curve. Join us from November 9-12, 2009. Register now!
http://p.sf.net/sfu/devconf
_______________________________________________
clisp-devel mailing list
clisp-devel@...
https://lists.sourceforge.net/lists/listinfo/clisp-devel

[Vladimir Tzankov]

On 10/1/09, Don Cohen <don-sourceforge-xxz@...> wrote:
> Is there some cvs update around then that seems related?
> Perhaps I should be doing a binary search to find the relevant
> update?  If you want me to do that, some instructions would be
> helpful.  I don't recall off hand how to check out as of some
> particular time or, better yet, as of some particular update.

you can checkout the source tree as it has been on Sep 1 like this:
cvs -z3 -d:pserver:anonymous@...:/cvsroot/clisp
co -D "09/01/2009 23:59" -P clisp

(or change -D option value)

I am not sure that the problem is related to particular change - I've
encountered this problem before. What gcc version is this?

Vladimir

------------------------------------------------------------------------------
Come build with us! The BlackBerry® Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay
ahead of the curve. Join us from November 9-12, 2009. Register now!
http://p.sf.net/sfu/devconf
_______________________________________________
clisp-devel mailing list
clisp-devel@...
https://lists.sourceforge.net/lists/listinfo/clisp-devel

[Sam Steingold]

Don Cohen wrote:
> Perhaps I should be doing a binary search to find the relevant
> update?  If you want me to do that, some instructions would be
> helpful.  I don't recall off hand how to check out as of some
> particular time or, better yet, as of some particular update.

cvs up -D date
or
cvs up -r revision
you will be using -D because -r is per-file.
note however that the bad change was not necessarily occurred between the dates
you think: the failure might have been triggered by an innocent change which
triggered a recompilation.
thus you might have to make a fresh checkout for each test.
i.e., "cvs co -D" and then configure --cbc in the fresh directory.
it might be helpful to keep two checkout: a working one and a broken one and
try to bring their dates close together.


------------------------------------------------------------------------------
Come build with us! The BlackBerry® Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay
ahead of the curve. Join us from November 9-12, 2009. Register now!
http://p.sf.net/sfu/devconf
_______________________________________________
clisp-devel mailing list
clisp-devel@...
https://lists.sourceforge.net/lists/listinfo/clisp-devel

[Don Cohen-12]

Vladimir Tzankov writes:

 > you can checkout the source tree as it has been on Sep 1 like this:
 > cvs -z3 -d:pserver:anonymous@...:/cvsroot/clisp
 > co -D "09/01/2009 23:59" -P clisp
 > (or change -D option value)
 > I am not sure that the problem is related to particular change - I've
 > encountered this problem before. What gcc version is this?
see below

Sam Steingold writes:
 > cvs up -D date
 > or
 > cvs up -r revision
 > you will be using -D because -r is per-file.
Is there a way to find out what -D arguments make sense?
That is, at what times were updates (I guess I should say "commits") done?

 > note however that the bad change was not necessarily occurred
 > between the dates you think: the failure might have been triggered
 > by an innocent change which triggered a recompilation.
 > thus you might have to make a fresh checkout for each test.
 > i.e., "cvs co -D" and then configure --cbc in the fresh directory.
That was my plan.
 > it might be helpful to keep two checkout: a working one and a
 > broken one and try to bring their dates close together.
In fact I was going to create a new directory for each -D argument.

$ gcc -v
Using built-in specs.
Target: i386-redhat-linux
Configured with: ../configure --prefix=/usr --mandir=/usr/share/man
--infodir=/usr/share/info --enable-shared --enable-threads=posix
--enable-checking=release --with-system-zlib --enable-__cxa_atexit
--disable-libunwind-exceptions --enable-libgcj-multifile
--enable-languages=c,c++,objc,java,f95,ada
--enable-java-awt=gtk --with-java-home=/usr/lib/jvm/java-1.4.2-gcj-1.4.2.0/jre
--host=i386-redhat-linux
Thread model: posix
gcc version 4.0.2 20051125 (Red Hat 4.0.2-8)
[2009-10-01 12:49:29 root@collabrium ~/clisp/build-dir/full]
$ uname -a
Linux collabrium.cs3-inc.com 2.6.17-1.2142_FC4 #1 Tue Jul 11 22:41:14
EDT 2006 i686 i686 i386 GNU/Linux
[2009-10-01 12:49:50 root@collabrium ~/clisp/build-dir/full]
$

I know this is pretty old but it's a huge pain to upgrade.

------------------------------------------------------------------------------
Come build with us! The BlackBerry® Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay
ahead of the curve. Join us from November 9-12, 2009. Register now!
http://p.sf.net/sfu/devconf
_______________________________________________
clisp-devel mailing list
clisp-devel@...
https://lists.sourceforge.net/lists/listinfo/clisp-devel

[Sam Steingold]

Don Cohen wrote:
> Is there a way to find out what -D arguments make sense?
> That is, at what times were updates (I guess I should say "commits") done?

try "cvs log src/ChangeLog"



------------------------------------------------------------------------------
Come build with us! The BlackBerry® Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay
ahead of the curve. Join us from November 9-12, 2009. Register now!
http://p.sf.net/sfu/devconf
_______________________________________________
clisp-devel mailing list
clisp-devel@...
https://lists.sourceforge.net/lists/listinfo/clisp-devel

[Don Cohen-12]

(sorry for the multiple copies last time - failed to edit
 from/to/cc appropriately)


==== build-clisp-mt-version ====
 #!/bin/bash -x

 # as of 2009-$1-$2

 mkdir ~/clisp-versions/2009-$1-$2-0001
 cd ~/clisp-versions/2009-$1-$2-0001
 cvs -z3 -d:pserver:anonymous@...:/cvsroot/clisp co -D "$1/$2/2009 00:01" -P clisp
 cd clisp
 ./configure --with-threads=POSIX_THREADS --with-module=rawsock
 build-mt
 cd build-mt
 make
====

09/01 failed, 08/15 failed, 08/01 succeeded,
08/10 failed, 08/06 succeeded
80/08 failed, 08/07 failed
now will start to change the times
(I hope these results are actually reliably reproducible!)

------------------------------------------------------------------------------
Come build with us! The BlackBerry® Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay
ahead of the curve. Join us from November 9-12, 2009. Register now!
http://p.sf.net/sfu/devconf
_______________________________________________
clisp-devel mailing list
clisp-devel@...
https://lists.sourceforge.net/lists/listinfo/clisp-devel

[Don Cohen-12]

new version of
==== build-clisp-mt-version ====
 #!/bin/bash -x

 # as of 2009-$1-$2 $3

 mkdir ~/clisp-versions/2009-$1-$2-$3
 cd ~/clisp-versions/2009-$1-$2-$3  
 cvs -z3 -d:pserver:anonymous@...:/cvsroot/clisp co -D "$1 /$2/2009 $3" -P clisp
 cd clisp
 ./configure --with-threads=POSIX_THREADS --with-module=rawsock build-mt
 cd build-mt
 make
====

  09/01 failed, 08/15 failed, 08/01 succeeded,
  08/10 failed, 08/06 succeeded
  80/08 failed, 08/07 failed
  now will start to change the times
  (I hope these results are actually reliably reproducible!)
====
  revision 1.7009
  date: 2009/08/06 00:03:32;  author: sds;  state: Exp;  lines: +13 -0
  use gnulib in syscalls (just string-time for now)
  * Makefile.devel (MAKEFILE_IN): add modules/syscalls/gllib/Makefile
  * modules/syscalls/Makefile.in (libgnu_sc.a): add target
  * modules/syscalls/calls.c (STRING-TIME): implement unconditionally
  * modules/syscalls/posix.lisp (STRING-TIME): export unconditionally
  * modules/syscalls/configure.in: call sc_gl_EARLY, sc_gl_INIT,
  gl_FUNC_MKTIME, gl_FUNC_GNU_STRFTIME, gl_FUNC_STRPTIME;
  do not check for strftime strptime mktime
  * modules/syscalls/link.sh.in (NEW_FILES): add libgnu_sc.a
  * modules/syscalls/gllib/, modules/syscalls/glm4/: add
  ----------------------------
  revision 1.7008
  date: 2009/08/05 23:17:22;  author: sds;  state: Exp;  lines: +3 -3

The state of the successful 8/6 build should be the same as that of
08 06 00:02, which I now try as a check, but it ends with:

  checking for struct lconv.int_n_sign_posn... yes
  configure: ** I18N (Output)
  configure: updating cache ../config.cache
  configure: creating ./config.status
  config.status: creating Makefile
  config.status: creating link.sh
  config.status: creating config.h
  configure: ** I18N (Done)
  /bin/sh: line 0: cd: i18n: No such file or directory
  make[1]: Entering directory
  `/root/clisp-versions/2009-08-06-00:02/clisp/build-mt/i18n'
  Makefile:23: *** target pattern contains no `%'.  Stop.
  make[1]: Leaving directory
  `/root/clisp-versions/2009-08-06-00:02/clisp/build-mt/i18n'
  make: *** [i18n] Error 2

I've ended up with this state in all builds I've done since the last
msg.  Any ideas what I'm doing wrong?

------------------------------------------------------------------------------
Come build with us! The BlackBerry® Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay
ahead of the curve. Join us from November 9-12, 2009. Register now!
http://p.sf.net/sfu/devconf
_______________________________________________
clisp-devel mailing list
clisp-devel@...
https://lists.sourceforge.net/lists/listinfo/clisp-devel

[Don Cohen-12]

I'm not sure, but it seems possible that the problem was times of the
form 12:01 rather than 1201.

Since I've been using times of form 1201 I have not had the builds
ending with
 make: *** [i18n] Error 2

The results are now almost but not quite consistent.
The lines containing xxx show builds ending with segfault,
the lines containing +++ show builds ending in success.
The other part is the time of the checkout, e.g., 2201 corresponds to
  cvs -z3 -d:pserver:anonymous@...:/cvsroot/clisp co -D '08/06/2009 2201' -P clisp


 revision 1.7019
 date: 2009/08/07 00:38:43;  author: sds;  state: Exp;  lines: +9 -0
 * makemake.in (DVIPDF): remove
 (PS2PDF): add
 (manual_print): rename from manual_dvi; remove clisp.dvi from it
 (clisp.dvi): remove target
 (clisp.pdf): generate from clisp.ps
 * configure.in: check for PS2PDF instead of DVIPDF

 xxx 2201

 ----------------------------
 revision 1.7018
 date: 2009/08/06 21:22:51;  author: sds;  state: Exp;  lines: +5 -0
 (clisp-module-distrib): distribute demos
 ----------------------------

 xxx 2100
 xxx 2102

 revision 1.7017
 date: 2009/08/06 20:59:16;  author: sds;  state: Exp;  lines: +8 -0
 (DISTRIBFILES): distribute README
 ----------------------------
 +++ 2040
 xxx 2045
 xxx 2050

Is it possible there was some change between 2040 and 2045 ?

 revision 1.7016
 date: 2009/08/06 20:38:55;  author: sds;  state: Exp;  lines: +7 -0
 fix bug #[ 1237776 ]: binary distribution does not include local libs
 (distrib): link local libs mentioned in base/makevars:LIBS into
 TOPDIR/base (same for full) and modify distmakefile accordingly
 ----------------------------
 +++ 2030
 +++ 2035

 revision 1.7015
 date: 2009/08/06 20:28:45;  author: sds;  state: Exp;  lines: +16 -0
 (DISTRIBFILES): include fas files for dynamic loading
 ----------------------------

 +++ 1702

 revision 1.7014
 date: 2009/08/06 16:42:12;  author: sds;  state: Exp;  lines: +6 -0
 (prepare_dynamic_modules): in the REQUIRE file,
 wrap LOADs in APPEASE-CERRORS because FFI modules might try to
 interface to functions absent on this platform
 ----------------------------
 revision 1.7013
 date: 2009/08/06 16:38:20;  author: sds;  state: Exp;  lines: +8 -1
 base modules are always linked statically
 * clisp-link.in (with_dynamic_modules): get from the environment
 with the default being @with_dynamic_modules@
 * makemake.in (base): pass with_dynamic_modules=no to clisp-link
 ----------------------------
 revision 1.7012
 date: 2009/08/06 16:35:17;  author: sds;  state: Exp;  lines: +5 -0
 (lncp): rename from link to separate from link.sh
 (lncp_some): rename from link_some to simplify search
 ----------------------------
 +++ 1201
 +++ 1301

 revision 1.7011
 date: 2009/08/06 04:44:26;  author: sds;  state: Exp;  lines: +4 -0
 (MAKEFILE_IN): depend on aclocal.m4

------------------------------------------------------------------------------
Come build with us! The BlackBerry® Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay
ahead of the curve. Join us from November 9-12, 2009. Register now!
http://p.sf.net/sfu/devconf
_______________________________________________
clisp-devel mailing list
clisp-devel@...
https://lists.sourceforge.net/lists/listinfo/clisp-devel

[Don Cohen-12]

  +++ 2040
  xxx 2045
  xxx 2050
 Is it possible there was some change between 2040 and 2045 ?

Diff shows that it is.
This looks to me like the prime suspect:

--- /root/clisp-versions/2009-08-06-2040/clisp/build-mt/cflags.h        2009-10-01 18:39:24.000000000 -0700
+++ /root/clisp-versions/2009-08-06-2045/clisp/build-mt/cflags.h        2009-10-01 18:54:25.000000000 -0700
@@ -1,10 +1,10 @@
 /* generated from Makefile */
 #define CC "gcc"
-#define CFLAGS "-g -O2 -W -Wswitch -Wcomment -Wpointer-arith -Wimplicit -Wreturn-type -Wmissing-declarations -Wno-sign-compare -Wno-format-nonliteral -O2 -fexpensive-optimizations -falign-functions=4 -pthread -DUNICODE -DMULTITHREAD -DPOSIX_THREADS -DDYNAMIC_FFI -I."
+#define CFLAGS "-g -O2 -W -Wswitch -Wcomment -Wpointer-arith -Wimplicit -Wreturn-type -Wmissing-declarations -Wno-sign-compare -Wno-format-nonliteral -O2 -fexpensive-optimizations -falign-functions=4 -pthread -DUNICODE -DMULTITHREAD -DPOSIX_THREADS -DDYNAMIC_FFI -DDYNAMIC_MODULES -I."


and many other similar things such as


--- /root/clisp-versions/2009-08-06-2040/clisp/build-mt/boot/makevars
 2009-10-01 18:41:08.000000000 -0700
+++ /root/clisp-versions/2009-08-06-2045/clisp/build-mt/boot/makevars
 2009-10-01 18:56:10.000000000 -0700
@@ -1,7 +1,7 @@
 CC='gcc'
-CPPFLAGS='-I/root/clisp-versions/2009-08-06-2040/clisp/build-mt/gllib'
-CFLAGS='-g -O2 -W -Wswitch -Wcomment -Wpointer-arith -Wimplicit
-Wreturn-type -Wmissing-declarations -Wno-sign-compare
-Wno-format-nonliteral -O2 -fexpensive-opt\
imizations -falign-functions=4 -pthread -DUNICODE -DMULTITHREAD
-DPOSIX_THREADS -DDYNAMIC_FFI -I.'
-CLFLAGS=''
+CPPFLAGS='-I/root/clisp-versions/2009-08-06-2045/clisp/build-mt/gllib'

+CFLAGS='-g -O2 -W -Wswitch -Wcomment -Wpointer-arith -Wimplicit
-Wreturn-type -Wmissing-declarations -Wno-sign-compare
-Wno-format-nonliteral -O2 -fexpensive-opt\
imizations -falign-functions=4 -pthread -DUNICODE -DMULTITHREAD
-DPOSIX_THREADS -DDYNAMIC_FFI -DDYNAMIC_MODULES -I. -fPIC'
+CLFLAGS=' -Wl,-export-dynamic'


--- /root/clisp-versions/2009-08-06-2040/clisp/build-mt/clisp.h
2009-10-01 18:39:26.000000000 -0700
+++ /root/clisp-versions/2009-08-06-2045/clisp/build-mt/clisp.h
2009-10-01 18:54:27.000000000 -0700
@@ -166,7 +166,7 @@
 #define INET_ADDR_SUFFIX /**/
 #define IOCTL_DOTS /**/
 #define IOCTL_REQUEST_T unsigned long
-#define MALLOC_ADDRESS_RANGE 0x09000000UL
+#define MALLOC_ADDRESS_RANGE 0x08000000UL
 #define NEED_SYS_IOCTL_H /**/
 #define PACKAGE "clisp"
 #define PACKAGE_BUGREPORT "http://clisp.cons.org/"
@@ -1308,6 +1308,7 @@
   subr_t D_shell;
   subr_t D_launch;
   subr_t D_savemem;
+  subr_t D_dynload_modules;
   subr_t D_program_name;
   subr_t D_lib_directory;
   subr_t D_set_lib_directory;
@@ -2383,6 +2384,7 @@
   symbol_ S_shell;
   symbol_ S_launch;
   symbol_ S_savemem;
+  symbol_ S_dynload_modules;
   symbol_ S_program_name;
   symbol_ S_lib_directory;
   symbol_ S_set_lib_directory;

------------------------------------------------------------------------------
Come build with us! The BlackBerry® Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay
ahead of the curve. Join us from November 9-12, 2009. Register now!
http://p.sf.net/sfu/devconf
_______________________________________________
clisp-devel mailing list
clisp-devel@...
https://lists.sourceforge.net/lists/listinfo/clisp-devel

[Sam Steingold]

Don Cohen wrote:

>   revision 1.7009
>   date: 2009/08/06 00:03:32;  author: sds;  state: Exp;  lines: +13 -0
>   use gnulib in syscalls (just string-time for now)
>   * Makefile.devel (MAKEFILE_IN): add modules/syscalls/gllib/Makefile
>   * modules/syscalls/Makefile.in (libgnu_sc.a): add target
>   * modules/syscalls/calls.c (STRING-TIME): implement unconditionally
>   * modules/syscalls/posix.lisp (STRING-TIME): export unconditionally
>   * modules/syscalls/configure.in: call sc_gl_EARLY, sc_gl_INIT,
>   gl_FUNC_MKTIME, gl_FUNC_GNU_STRFTIME, gl_FUNC_STRPTIME;
>   do not check for strftime strptime mktime
>   * modules/syscalls/link.sh.in (NEW_FILES): add libgnu_sc.a
>   * modules/syscalls/gllib/, modules/syscalls/glm4/: add

this is a "self contained patch" in the sense that nothing that comes later
relies on it.
this means that if you suspect that it causes you grief, you should take cvs
head and revert it and see if the problem goes away.

>   configure: ** I18N (Output)
>   configure: updating cache ../config.cache
>   configure: creating ./config.status
>   config.status: creating Makefile
>   config.status: creating link.sh
>   config.status: creating config.h
>   configure: ** I18N (Done)
>   /bin/sh: line 0: cd: i18n: No such file or directory

is this from the top-level configure?
try running it as "bash -v -x configure ..."

>   make[1]: Entering directory
>   `/root/clisp-versions/2009-08-06-00:02/clisp/build-mt/i18n'
>   Makefile:23: *** target pattern contains no `%'.  Stop.
>   make[1]: Leaving directory
>   `/root/clisp-versions/2009-08-06-00:02/clisp/build-mt/i18n'
>   make: *** [i18n] Error 2

so, what does build-mt/i18n/Makefile:23 look like?

Sam

------------------------------------------------------------------------------
Come build with us! The BlackBerry® Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay
ahead of the curve. Join us from November 9-12, 2009. Register now!
http://p.sf.net/sfu/devconf
_______________________________________________
clisp-devel mailing list
clisp-devel@...
https://lists.sourceforge.net/lists/listinfo/clisp-devel

[Sam Steingold]

Don Cohen wrote:
>   +++ 2040
>   xxx 2045
>   xxx 2050
>  Is it possible there was some change between 2040 and 2045 ?
>
> Diff shows that it is.

the changes below point to

2009-08-06  Sam Steingold  <sds@...>

        * configure, makemake.in (with_dynamic_modules):
        "yes" is now the default

to mitigate the effects of this change, please pass an explicit
--with-dynamic-modules=yes (or --with-dynamic-modules=no) to the top-level
configure.

> This looks to me like the prime suspect:
>
> --- /root/clisp-versions/2009-08-06-2040/clisp/build-mt/cflags.h        2009-10-01 18:39:24.000000000 -0700
> +++ /root/clisp-versions/2009-08-06-2045/clisp/build-mt/cflags.h        2009-10-01 18:54:25.000000000 -0700
> @@ -1,10 +1,10 @@
>  /* generated from Makefile */
>  #define CC "gcc"
> -#define CFLAGS "-g -O2 -W -Wswitch -Wcomment -Wpointer-arith -Wimplicit -Wreturn-type -Wmissing-declarations -Wno-sign-compare -Wno-format-nonliteral -O2 -fexpensive-optimizations -falign-functions=4 -pthread -DUNICODE -DMULTITHREAD -DPOSIX_THREADS -DDYNAMIC_FFI -I."
> +#define CFLAGS "-g -O2 -W -Wswitch -Wcomment -Wpointer-arith -Wimplicit -Wreturn-type -Wmissing-declarations -Wno-sign-compare -Wno-format-nonliteral -O2 -fexpensive-optimizations -falign-functions=4 -pthread -DUNICODE -DMULTITHREAD -DPOSIX_THREADS -DDYNAMIC_FFI -DDYNAMIC_MODULES -I."
>
>
> and many other similar things such as
>
>
> --- /root/clisp-versions/2009-08-06-2040/clisp/build-mt/boot/makevars
>  2009-10-01 18:41:08.000000000 -0700
> +++ /root/clisp-versions/2009-08-06-2045/clisp/build-mt/boot/makevars
>  2009-10-01 18:56:10.000000000 -0700
> @@ -1,7 +1,7 @@
>  CC='gcc'
> -CPPFLAGS='-I/root/clisp-versions/2009-08-06-2040/clisp/build-mt/gllib'
> -CFLAGS='-g -O2 -W -Wswitch -Wcomment -Wpointer-arith -Wimplicit
> -Wreturn-type -Wmissing-declarations -Wno-sign-compare
> -Wno-format-nonliteral -O2 -fexpensive-opt\
> imizations -falign-functions=4 -pthread -DUNICODE -DMULTITHREAD
> -DPOSIX_THREADS -DDYNAMIC_FFI -I.'
> -CLFLAGS=''
> +CPPFLAGS='-I/root/clisp-versions/2009-08-06-2045/clisp/build-mt/gllib'
>
> +CFLAGS='-g -O2 -W -Wswitch -Wcomment -Wpointer-arith -Wimplicit
> -Wreturn-type -Wmissing-declarations -Wno-sign-compare
> -Wno-format-nonliteral -O2 -fexpensive-opt\
> imizations -falign-functions=4 -pthread -DUNICODE -DMULTITHREAD
> -DPOSIX_THREADS -DDYNAMIC_FFI -DDYNAMIC_MODULES -I. -fPIC'
> +CLFLAGS=' -Wl,-export-dynamic'
>
>
> --- /root/clisp-versions/2009-08-06-2040/clisp/build-mt/clisp.h
> 2009-10-01 18:39:26.000000000 -0700
> +++ /root/clisp-versions/2009-08-06-2045/clisp/build-mt/clisp.h
> 2009-10-01 18:54:27.000000000 -0700
> @@ -166,7 +166,7 @@
>  #define INET_ADDR_SUFFIX /**/
>  #define IOCTL_DOTS /**/
>  #define IOCTL_REQUEST_T unsigned long
> -#define MALLOC_ADDRESS_RANGE 0x09000000UL
> +#define MALLOC_ADDRESS_RANGE 0x08000000UL
>  #define NEED_SYS_IOCTL_H /**/
>  #define PACKAGE "clisp"
>  #define PACKAGE_BUGREPORT "http://clisp.cons.org/"
> @@ -1308,6 +1308,7 @@
>    subr_t D_shell;
>    subr_t D_launch;
>    subr_t D_savemem;
> +  subr_t D_dynload_modules;
>    subr_t D_program_name;
>    subr_t D_lib_directory;
>    subr_t D_set_lib_directory;
> @@ -2383,6 +2384,7 @@
>    symbol_ S_shell;
>    symbol_ S_launch;
>    symbol_ S_savemem;
> +  symbol_ S_dynload_modules;
>    symbol_ S_program_name;
>    symbol_ S_lib_directory;
>    symbol_ S_set_lib_directory;

------------------------------------------------------------------------------
Come build with us! The BlackBerry® Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay
ahead of the curve. Join us from November 9-12, 2009. Register now!
http://p.sf.net/sfu/devconf
_______________________________________________
clisp-devel mailing list
clisp-devel@...
https://lists.sourceforge.net/lists/listinfo/clisp-devel

[Don Cohen-12]

Sam Steingold writes:
 > Don Cohen wrote:
 > >   revision 1.7009
 > >   date: 2009/08/06 00:03:32;  author: sds;  state: Exp;  lines: +13 -0
 ...
 > this is a "self contained patch" in the sense that nothing that comes later
 > relies on it.
 > this means that if you suspect that it causes you grief, you should
 > take cvs head and revert it and see if the problem goes away.

To me it appears there are two different (I imagine unrelated) issues.
One is the I18N error and the other is the segfault.
So far I am mystified by both, but the segfault seems more serious
since I18N seems to occur only when I check out with a time containing
a colon.  Is this remotely plausible?  Shall I collect and send the
empirical evidence?

 > >   configure: ** I18N (Output)
 > >   configure: updating cache ../config.cache
 > >   configure: creating ./config.status
 > >   config.status: creating Makefile
 > >   config.status: creating link.sh
 > >   config.status: creating config.h
 > >   configure: ** I18N (Done)
 > >   /bin/sh: line 0: cd: i18n: No such file or directory
 >
 > is this from the top-level configure?
 > try running it as "bash -v -x configure ..."
I guess you mean was it running configure when the error occurred.
No.  The shell command that started the build was this:
 ./build-clisp-mt-version 08 06 00:02 >> transc08060002 2>> transc08060002
where build-clisp-mt-version is the second script I sent.
The last line of that file, which was executing when the output above
was generated was
 make
The previous line was
 ./configure --with-threads=POSIX_THREADS --with-module=rawsock build-mt

 > >   make[1]: Entering directory
 > >   `/root/clisp-versions/2009-08-06-00:02/clisp/build-mt/i18n'
 > >   Makefile:23: *** target pattern contains no `%'.  Stop.
 > >   make[1]: Leaving directory
 > >   `/root/clisp-versions/2009-08-06-00:02/clisp/build-mt/i18n'
 > >   make: *** [i18n] Error 2
 >
 > so, what does build-mt/i18n/Makefile:23 look like?

 %.fas: $(srcdir)/%.lisp

(Not that I know how to interpret the error message but it sure looks
to me like the target pattern contains a %.  I'm guessing that the
target is either what comes before or after the colon.)


------------------------------------------------------------------------------
Come build with us! The BlackBerry® Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay
ahead of the curve. Join us from November 9-12, 2009. Register now!
http://p.sf.net/sfu/devconf
_______________________________________________
clisp-devel mailing list
clisp-devel@...
https://lists.sourceforge.net/lists/listinfo/clisp-devel

[Don Cohen-12]

Sam Steingold writes:
 > to mitigate the effects of this change, please pass an explicit
 > --with-dynamic-modules=yes (or --with-dynamic-modules=no) to the top-level
 > configure.

yes produces the same error as before, no works.

What's the next step?  I guess we want to make yes work?
If this is going to take a while then perhaps I should
do nightly builds of both yes and no.

------------------------------------------------------------------------------
Come build with us! The BlackBerry® Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay
ahead of the curve. Join us from November 9-12, 2009. Register now!
http://p.sf.net/sfu/devconf
_______________________________________________
clisp-devel mailing list
clisp-devel@...
https://lists.sourceforge.net/lists/listinfo/clisp-devel

[Sam Steingold]

Don Cohen wrote:
> Sam Steingold writes:
>  > to mitigate the effects of this change, please pass an explicit
>  > --with-dynamic-modules=yes (or --with-dynamic-modules=no) to the top-level
>  > configure.
>
> yes produces the same error as before, no works.

so, are you saying that mt+dynamic modules never worked?

> What's the next step?  I guess we want to make yes work?

yes.

> If this is going to take a while then perhaps I should
> do nightly builds of both yes and no.

that would be nice.


------------------------------------------------------------------------------
Come build with us! The BlackBerry® Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay
ahead of the curve. Join us from November 9-12, 2009. Register now!
http://p.sf.net/sfu/devconf
_______________________________________________
clisp-devel mailing list
clisp-devel@...
https://lists.sourceforge.net/lists/listinfo/clisp-devel

[Sam Steingold]

Don Cohen wrote:
>
> To me it appears there are two different (I imagine unrelated) issues.
> One is the I18N error and the other is the segfault.
> So far I am mystified by both, but the segfault seems more serious
> since I18N seems to occur only when I check out with a time containing
> a colon.  Is this remotely plausible?  Shall I collect and send the
> empirical evidence?

are you saying that the outcome depends on the time format?
it is certainly worth investigating.
specifically, check out two trees with the same date (with and without the
colon), configure and observe the error.
then compare the directory trees to figure out how they differ
(either use emacs or a shell tool, google for compare directories).



------------------------------------------------------------------------------
Come build with us! The BlackBerry® Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay
ahead of the curve. Join us from November 9-12, 2009. Register now!
http://p.sf.net/sfu/devconf
_______________________________________________
clisp-devel mailing list
clisp-devel@...
https://lists.sourceforge.net/lists/listinfo/clisp-devel

[Vladimir Tzankov]

On 10/5/09, Sam Steingold <sds@...> wrote:

> Don Cohen wrote:
>> Sam Steingold writes:
>>  > to mitigate the effects of this change, please pass an explicit
>>  > --with-dynamic-modules=yes (or --with-dynamic-modules=no) to the
>> top-level
>>  > configure.
>>
>> yes produces the same error as before, no works.
>
> so, are you saying that mt+dynamic modules never worked?

They work fine.
I guess the problem is the same or related to this one:
http://thread.gmane.org/gmane.lisp.clisp.devel/20732/

Currently I cannot reproduce it but I remember that when inspecting
the C stack there was extra value pushed when vectorof was called.

Vladimir

------------------------------------------------------------------------------
Come build with us! The BlackBerry® Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay
ahead of the curve. Join us from November 9-12, 2009. Register now!
http://p.sf.net/sfu/devconf
_______________________________________________
clisp-devel mailing list
clisp-devel@...
https://lists.sourceforge.net/lists/listinfo/clisp-devel

[Don Cohen-12]

Vladimir Tzankov writes:
 > On 10/5/09, Sam Steingold <sds@...> wrote:
 > > Don Cohen wrote:
 > >> Sam Steingold writes:
 > >>  > to mitigate the effects of this change, please pass an explicit
 > >>  > --with-dynamic-modules=yes (or --with-dynamic-modules=no) to the
 > >> top-level
 > >>  > configure.
 > >>
 > >> yes produces the same error as before, no works.
 > >
 > > so, are you saying that mt+dynamic modules never worked?
It seems entirely plausible that it never worked on this particular
machine.  Recall I have another machine on which it has been working
all along.
The one on which it fails is running Fedora Core 4 (2.6.17-1.2142_FC4)
Of course the problem could be related to all sorts of other things,
like
 $ gcc --version
 gcc (GCC) 4.0.2 20051125 (Red Hat 4.0.2-8)
etc. (Let me know if you want more such details.)

 > They work fine.
 > I guess the problem is the same or related to this one:
 > http://thread.gmane.org/gmane.lisp.clisp.devel/20732/
 >
 > Currently I cannot reproduce it but I remember that when inspecting
 > the C stack there was extra value pushed when vectorof was called.

I'm willing to do experiments but you'll have to describe in some
detail what you want me to try.

------------------------------------------------------------------------------
Come build with us! The BlackBerry® Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay
ahead of the curve. Join us from November 9-12, 2009. Register now!
http://p.sf.net/sfu/devconf
_______________________________________________
clisp-devel mailing list
clisp-devel@...
https://lists.sourceforge.net/lists/listinfo/clisp-devel