tag:old.nabble.com,2006:forum-1648
Nabble - Gnu - Global
2009-11-24T20:40:56Z
GNU GLOBAL is a source code tag system that works the same way across diverse environments. You can locate a specified object in the source files and move there easily. It is useful for hacking a large project containing many subdirectories, many #ifdef and many main() functions. It is similar to ctags or etags but is different from them at the point of independence of any editor. Gnu - Global home is <a href="http://www.gnu.org/software/global/global.html" target="_top" rel="nofollow">here</a>.
tag:old.nabble.com,2006:post-26507476
Re: Binary recognition is to narrow [new suggestion]
2009-11-24T20:40:56Z
2009-11-24T20:40:56Z
Shigio YAMAGUCHI-2
> > I would like to make the 512 a customizable variable too.
<br>>
<br>> I examined the performance using attached test program.
<br>> And I confirmed that execute time did not have a significant difference
<br>> at size smaller than 512. I think that 512 is an appropriate value.
<br><br>I understood. Fixed value 512 seems to be the best.
<br>--
<br>Shigio YAMAGUCHI <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26507476&i=0" target="_top" rel="nofollow">shigio@...</a>>
<br>PGP fingerprint: D1CB 0B89 B346 4AB6 5663 C4B6 3CA5 BBB3 57BE DDA3
<br><br><br>_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26507476&i=1" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>
tag:old.nabble.com,2006:post-26495846
Re: Binary recognition is to narrow [new suggestion]
2009-11-24T05:24:59Z
2009-11-24T05:24:59Z
Hideki IWAMOTO
Hi.
<br>> I would like to make the 512 a customizable variable too.
<br><br>I examined the performance using attached test program.
<br>And I confirmed that execute time did not have a significant difference
<br>at size smaller than 512. I think that 512 is an appropriate value.
<br><br><br>$ foreach a ( 32 64 128 256 512 1024 2048 4096 8192 )
<br>foreach? rm -fr linux-2.6.31; tar xfj ~/download/linux/linux-2.6.31.tar.bz2; sync
<br>foreach? time sh -c 'find linux-2.6.31 -type f | test_isbinary '$a' > /dev/null'
<br>foreach? end
<br>0.060u 0.348s 0:00.30 133.3% 0+0k 0+0io 0pf+0w
<br>0.048u 0.344s 0:00.31 122.5% 0+0k 0+0io 0pf+0w
<br>0.088u 0.340s 0:00.32 131.2% 0+0k 0+0io 0pf+0w
<br>0.076u 0.364s 0:00.32 134.3% 0+0k 0+0io 0pf+0w
<br>0.084u 0.372s 0:00.34 132.3% 0+0k 0+0io 0pf+0w
<br>0.112u 0.368s 0:00.37 127.0% 0+0k 0+0io 0pf+0w
<br>0.152u 0.368s 0:00.42 121.4% 0+0k 0+0io 0pf+0w
<br>0.260u 0.368s 0:00.51 121.5% 0+0k 0+0io 0pf+0w
<br>0.388u 0.368s 0:00.75 98.6% 0+0k 0+0io 0pf+0w
<br><br><br>On Sat, 21 Nov 2009 15:42:11 +0900, Shigio YAMAGUCHI wrote...
<div class='shrinkable-quote'><div class='shrinkable-quote'><br>> > Instead of counting characters over 127 the only test is that the first
<br>> > 511 bytes don't contain any of the controll characters 0-8, 14-31. No
<br>> > normal textfile would contain these.
<br>> >
<br>> > Assuming that binary data is random the probability of a incorrectly
<br>> > tagged binary would be
<br>> >
<br>> > ((256-8-18)/256)^511=.00000000000000000000000170726
<br>> >
<br>> > just testing 127 bits would be a bit to little
<br>> >
<br>> > ((256-8-18)/256)^127=.00000123868
<br>>
<br>> This is a very interesting idea.
<br>>
<br>> > One of the benefits is that this will correctly tag files in uni-code as
<br>> > text as well. Since those control characters never appears in uni-code
<br>> > either.
<br>>
<br>> This is a big merit.
<br>> Most other multi-byte character set are sure to be designed like that,
<br>>
<br>> I would like to make the 512 a customizable variable too.
<br>>
<br>> $ gtags ... use conventional test
<br>>
<br>> [File gtags.conf]
<br>> +----------------------------
<br>> |...
<br>> | :binarytest_size=512:... ----------------------------------+
<br>> | |
<br>> v
<br>> $ gtags ... use new test using the first n=512 bytes
<br>>
<br>> After testing for a while, we can decide what we should do.
<br>> Thank you for your profitable consideration.
<br>> --
<br>> Shigio YAMAGUCHI <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26495846&i=0" target="_top" rel="nofollow">shigio@...</a>>
<br>> PGP fingerprint: D1CB 0B89 B346 4AB6 5663 C4B6 3CA5 BBB3 57BE DDA3
<br>>
<br>>
<br>> _______________________________________________
<br>> Bug-global mailing list
<br>> <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26495846&i=1" target="_top" rel="nofollow">Bug-global@...</a>
<br>> <a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a></div></div>----
<br>Hideki IWAMOTO <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26495846&i=2" target="_top" rel="nofollow">h-iwamoto@...</a>
<br><br /> <br />_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26495846&i=3" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><div class="small"><br/><img src="http://old.nabble.com/images/icon_attachment.gif" > <strong>20091124-test_isbinary.patch</strong> (6K) <a href="http://old.nabble.com/attachment/26495846/0/20091124-test_isbinary.patch" target="_top">Download Attachment</a></div><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>
tag:old.nabble.com,2006:post-26489677
Re: [PATCH] Add 2-pass mode to gtags
2009-11-23T18:41:38Z
2009-11-23T18:41:38Z
Shigio YAMAGUCHI-2
> When bree cache is small, current implementation of 2-pass mode
<br>> is slower than default mode.
<br>...
<br>> It seems that moving use of defined() from gtags-parser to gtags is bad idea.
<br>> I will reimplement 2-pass mode in another way.
<br>> Since new 2-pass mode is implemented by adding tag type to output of gtags-parser,
<br>> the specifications of parser will be compatible with default mode.
<br><br>Since we enhanced the default cash size of GLOBAL to 50MB for about one year ago,
<br>there might particularly not be necessity for sticking to a small cash size.
<br>It is unavoidable that a new improvement consumes the resource more.
<br>Current 2-pass mode works faster than the default mode even when cash size == 33MB.
<br>It has domination enough, I think.
<br><br>Of course, I don't oppose the change to a better method.
<br>However, in one side, I hope not to change command interface if possible.
<br><br>Anyway, it is not necessary to hurry up. :)
<br>--
<br>Shigio YAMAGUCHI <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26489677&i=0" target="_top" rel="nofollow">shigio@...</a>>
<br>PGP fingerprint: D1CB 0B89 B346 4AB6 5663 C4B6 3CA5 BBB3 57BE DDA3
<br><br><br>_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26489677&i=1" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>
tag:old.nabble.com,2006:post-26478352
Re: [PATCH] Add 2-pass mode to gtags
2009-11-23T05:49:05Z
2009-11-23T05:49:05Z
Hideki IWAMOTO
Hi.
<br><br>> Since there are a lot of changes in the parser, the output should be seen by
<br>> a lot of people before the formal release, I think.
<br><br>When bree cache is small, current implementation of 2-pass mode is slower than default mode.
<br><br>$ foreach cpumask ( 1 3 )
<br>foreach? foreach label ( default 2pass )
<br>foreach? echo ==== cpumask:$cpumask label:$label ====
<br>foreach? foreach cachesize ( 8388608 16777216 33554432 67108864 )
<br>foreach? rm -fr linux-2.6.31; tar xfj ~/download/linux/linux-2.6.31.tar.bz2; sync
<br>foreach? (cd linux-2.6.31; time taskset $cpumask env GTAGSLABEL=$label GTAGSCACHE=$cachesize gtags)
<br>foreach? end
<br>foreach? end
<br>foreach? end
<br>==== cpumask:1 label:default ====
<br>114.267u 24.085s 3:09.48 73.0% 0+0k 0+0io 1pf+0w
<br>115.367u 20.521s 2:55.73 77.3% 0+0k 0+0io 0pf+0w
<br>116.227u 18.081s 2:32.98 87.7% 0+0k 0+0io 0pf+0w
<br>116.671u 16.093s 2:24.36 91.9% 0+0k 0+0io 0pf+0w
<br>==== cpumask:1 label:2pass ====
<br>91.353u 19.205s 3:48.79 48.3% 0+0k 0+0io 0pf+0w
<br>92.317u 16.501s 3:05.40 58.6% 0+0k 0+0io 0pf+0w
<br>92.797u 13.580s 2:08.52 82.7% 0+0k 0+0io 0pf+0w
<br>93.929u 11.080s 1:56.03 90.4% 0+0k 0+0io 0pf+0w
<br>==== cpumask:3 label:default ====
<br>112.507u 23.593s 2:48.80 80.6% 0+0k 0+0io 1pf+0w
<br>112.991u 18.965s 2:30.06 87.9% 0+0k 0+0io 0pf+0w
<br>113.591u 17.305s 2:13.30 98.1% 0+0k 0+0io 0pf+0w
<br>114.831u 16.113s 2:11.92 99.2% 0+0k 0+0io 0pf+0w
<br>==== cpumask:3 label:2pass ====
<br>88.773u 18.929s 3:17.01 54.6% 0+0k 0+0io 0pf+0w
<br>90.085u 15.108s 2:35.38 67.6% 0+0k 0+0io 0pf+0w
<br>90.877u 12.204s 1:44.62 98.5% 0+0k 0+0io 0pf+0w
<br>91.217u 10.540s 1:29.74 113.3% 0+0k 0+0io 0pf+0w
<br><br>It seems that moving use of defined() from gtags-parser to gtags is bad idea.
<br>I will reimplement 2-pass mode in another way.
<br>Since new 2-pass mode is implemented by adding tag type to output of gtags-parser,
<br>the specifications of parser will be compatible with default mode.
<br><br><br>On Thu, 22 Oct 2009 15:44:28 +0900, Shigio YAMAGUCHI wrote...
<div class='shrinkable-quote'><br>> Hi Hideki,
<br>>
<br>> > To improve the performance of gtags, this patch adds 2-pass mode.
<br>> >
<br>> > The following three points are improved by moving use of function defined()
<br>> > from gtags-parser to gtags.
<br>> > 1. Decreases the frequency of parsing source file from three times to two times.
<br>> > 2. Decreases reading of GTAGS by keeping cache of GTAGS
<br>> > beyond the duration of parser process.
<br>> > 3. Increases parallelism of gtags-parser and gtags on multiprocessor system.
<br>>
<br>> Great!!! I named your code 'Giant steps'.
<br>>
<br>> Since there are a lot of changes in the parser, the output should be seen by
<br>> a lot of people before the formal release, I think.
<br>> How about releasing twice like follows?
<br>>
<br>> include Giant steps? enable Giant steps by default?
<br>> ------------------------------------------------------------------------------
<br>> GLOBAL-5.7.7(beta) Yes No
<br>> GLOBAL-5.8(formal) Yes Yes
<br>>
<br>> Thank you for your wonderful 'Giant steps'!
<br>> --
<br>> Shigio YAMAGUCHI <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26478352&i=0" target="_top" rel="nofollow">shigio@...</a>>
<br>> PGP fingerprint: D1CB 0B89 B346 4AB6 5663 C4B6 3CA5 BBB3 57BE DDA3
</div><br>----
<br>Hideki IWAMOTO <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26478352&i=1" target="_top" rel="nofollow">h-iwamoto@...</a>
<br><br><br>_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26478352&i=2" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>
tag:old.nabble.com,2006:post-26471324
Re: GLOBAL-5.7.5 released.
2009-11-22T16:59:06Z
2009-11-22T16:59:06Z
Shigio YAMAGUCHI-2
Hi
<br>> --from-here option seems not to support compact format GTAGS.
<br>> Is this a bug?
<br><br>You are right. This is a bug.
<br>I have fixed it.
<br><br>I will commit it after your commiting your work.
<br>If I should do it right now, please say so.
<br>Thank you.
<br>--
<br>Shigio YAMAGUCHI <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26471324&i=0" target="_top" rel="nofollow">shigio@...</a>>
<br>PGP fingerprint: D1CB 0B89 B346 4AB6 5663 C4B6 3CA5 BBB3 57BE DDA3
<br><br><br>_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26471324&i=1" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>
tag:old.nabble.com,2006:post-26465483
Re: GLOBAL-5.7.5 released.
2009-11-22T05:53:10Z
2009-11-22T05:53:10Z
Hideki IWAMOTO
Hi.
<br><br>> o Revived the -c option.
<br>> If this option is specified, gtags uses compact format for GTAGS.
<br>> This option does not influence GRTAGS and GSYMS, because they are always
<br>> made using compact format.
<br>> If you think that GTAGS is too big, please use this option.
<br><br>--from-here option seems not to support compact format GTAGS.
<br>Is this a bug?
<br><br>On Sat, 14 Mar 2009 09:39:04 +0900, Shigio YAMAGUCHI wrote...
<div class='shrinkable-quote'><br>> -----BEGIN PGP SIGNED MESSAGE-----
<br>> Hash: SHA1
<br>>
<br>> Hello hackers,
<br>>
<br>> It is my great pleasure to announce the release of GLOBAL-5.7.5.
<br>> It's a small update.
<br>>
<br>> [CHANGES]
<br>> Change of gtags:
<br>> o Revived the -c option.
<br>> If this option is specified, gtags uses compact format for GTAGS.
<br>> This option does not influence GRTAGS and GSYMS, because they are always
<br>> made using compact format.
<br>> If you think that GTAGS is too big, please use this option.
<br>> o Removed the limitation of the depth of the directory hierarchy.
<br>> o Changed not to do useless inspection for the files in the directories
<br>> which are listed in the skip list.
<br>>
<br>> You can download it from
<br>> <a href="http://www.gnu.org/software/global/download.html" target="_top" rel="nofollow">http://www.gnu.org/software/global/download.html</a><br>>
<br>> Enjoy!
<br>> -----BEGIN PGP SIGNATURE-----
<br>> Version: GnuPG v1.4.2.2 (FreeBSD)
<br>> Comment: For info see <a href="http://www.gnupg.org" target="_top" rel="nofollow">http://www.gnupg.org</a><br>>
<br>> iD8DBQFJuvsxPKW7s1e+3aMRAmIdAJoDhWQFv+Qqh4QE2p/O/fgO899sywCgncdP
<br>> suGvXs0Y1N3bBYGuW3mBEPU=
<br>> =urxa
<br>> -----END PGP SIGNATURE-----
<br>> --
<br>> Shigio YAMAGUCHI <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26465483&i=0" target="_top" rel="nofollow">shigio@...</a>>
<br>> PGP fingerprint: D1CB 0B89 B346 4AB6 5663 C4B6 3CA5 BBB3 57BE DDA3
<br>>
<br>>
<br>> _______________________________________________
<br>> Info-global mailing list
<br>> <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26465483&i=1" target="_top" rel="nofollow">Info-global@...</a>
<br>> <a href="http://lists.gnu.org/mailman/listinfo/info-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/info-global</a></div><br>----
<br>Hideki IWAMOTO <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26465483&i=2" target="_top" rel="nofollow">h-iwamoto@...</a>
<br><br><br>_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26465483&i=3" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>
tag:old.nabble.com,2006:post-26463640
Re: [PATCH] Allow specifying both of -r and -s for -f on 2-pass mode
2009-11-22T00:57:36Z
2009-11-22T00:57:36Z
Hideki IWAMOTO
Hi.
<br><br>> By the way, how about separating between the tag type and name
<br>> by single blank?
<br>> And if possible, would you please write a simple comment about
<br>> the change in ChangeLog.
<br><br>OK. This is revised version.
<br>I will commit this patch on the next weekend.
<br><br> ChangeLog | 10 +++++
<br> global/global.c | 87 +++++++++++++++++++++++++++++---------------------
<br> gtags/gtags.c | 2 -
<br> htags/anchor.c | 52 +++++++++++++++++------------
<br> htags/anchor.h | 8 ++--
<br> htags/htags.c | 3 +
<br> htags/htags.h | 1
<br> libutil/format.h | 29 ++++++++++------
<br> libutil/pathconvert.c | 18 +++++++++-
<br> libutil/pathconvert.h | 7 +---
<br> 10 files changed, 139 insertions(+), 78 deletions(-)
<br><br>On Sun, 22 Nov 2009 10:24:44 +0900, Shigio YAMAGUCHI wrote...
<div class='shrinkable-quote'><div class='shrinkable-quote'><br>> Hi
<br>> > The number of executions of "global -f" from htags has decreased
<br>> > by this change and adding new format to --result option.
<br>> > The CPU time of 2-pass mode is shorter than that of the default mode now.
<br>>
<br>> Good job. The speed decrease of htags was completely canceled,
<br>> and it has been improved further.
<br>>
<br>> By the way, how about separating between the tag type and name
<br>> by single blank?
<br>>
<br>> Dmain 219 global/global.c main(int argc, char **argv)
<br>> |
<br>> v
<br>> D main 219 global/global.c main(int argc, char **argv)
<br>>
<br>> Limiting tag type to single character will become a limitation
<br>> for some enhancing in the future.
<br>>
<br>> And if possible, would you please write a simple comment about
<br>> the change in ChangeLog.
<br>> Thank you.
<br>> --
<br>> Shigio YAMAGUCHI <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26463640&i=0" target="_top" rel="nofollow">shigio@...</a>>
<br>> PGP fingerprint: D1CB 0B89 B346 4AB6 5663 C4B6 3CA5 BBB3 57BE DDA3
</div></div>----
<br>Hideki IWAMOTO <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26463640&i=1" target="_top" rel="nofollow">h-iwamoto@...</a>
<br><br /> <br />_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26463640&i=2" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><div class="small"><br/><img src="http://old.nabble.com/images/icon_attachment.gif" > <strong>20091122-global-frs+ctags-xtype.patch</strong> (27K) <a href="http://old.nabble.com/attachment/26463640/0/20091122-global-frs%2Bctags-xtype.patch" target="_top">Download Attachment</a></div><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>
tag:old.nabble.com,2006:post-26462043
Re: [PATCH] Allow specifying both of -r and -s for -f on 2-pass mode
2009-11-21T17:14:26Z
2009-11-21T17:14:26Z
Shigio YAMAGUCHI-2
Hi
<br>> The number of executions of "global -f" from htags has decreased
<br>> by this change and adding new format to --result option.
<br>> The CPU time of 2-pass mode is shorter than that of the default mode now.
<br><br>Good job. The speed decrease of htags was completely canceled,
<br>and it has been improved further.
<br><br>By the way, how about separating between the tag type and name
<br>by single blank?
<br><br>Dmain 219 global/global.c main(int argc, char **argv)
<br>|
<br>v
<br>D main 219 global/global.c main(int argc, char **argv)
<br><br>Limiting tag type to single character will become a limitation
<br>for some enhancing in the future.
<br><br>And if possible, would you please write a simple comment about
<br>the change in ChangeLog.
<br>Thank you.
<br>--
<br>Shigio YAMAGUCHI <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26462043&i=0" target="_top" rel="nofollow">shigio@...</a>>
<br>PGP fingerprint: D1CB 0B89 B346 4AB6 5663 C4B6 3CA5 BBB3 57BE DDA3
<br><br><br>_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26462043&i=1" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>
tag:old.nabble.com,2006:post-26458467
Re: [PATCH] Allow specifying both of -r and -s for -f on 2-pass mode
2009-11-21T09:18:28Z
2009-11-21T09:18:28Z
Hideki IWAMOTO
Hi.
<br><br>> This patch allows specifying both of -r and -s for "global -f" command in 2-pass mode.
<br><br>The number of executions of "global -f" from htags has decreased
<br>by this change and adding new format to --result option.
<br>The CPU time of 2-pass mode is shorter than that of the default mode now.
<br><br>$ foreach label ( default 2pass )
<br>foreach? echo ==== label:$label ====
<br>foreach? rm -fr linux-2.6.31; tar xfj ~/download/linux/linux-2.6.31.tar.bz2; sync
<br>foreach? (cd linux-2.6.31; setenv GTAGSLABEL $label; gtags; strace -c -f -F -q -S calls htags)
<br>foreach? end
<br>==== label:default ====
<br>% time seconds usecs/call calls errors syscall
<br>------ ----------- ----------- --------- --------- ----------------
<br> 34.09 121.072744 80 1508837 76 write
<br> 43.94 156.029997 107 1453270 read
<br> 8.53 30.283076 82 367068 pread
<br> 2.56 9.091694 51 179733 lseek
<br> 1.82 6.463504 38 169023 mmap
<br> 1.67 5.915556 35 167323 close
<br> 1.81 6.413643 38 167047 fstat
<br> 2.86 10.138799 61 166820 open
<br> 1.86 6.613891 40 166056 munmap
<br> 0.39 1.368624 13 109252 1325 stat
<br> 0.32 1.131337 47 24007 brk
<br> 0.04 0.125198 9 13329 mremap
<br> 0.01 0.040300 25 1632 rt_sigaction
<br> 0.01 0.039444 30 1335 mprotect
<br> 0.01 0.038355 43 901 297 access
<br> 0.01 0.021350 24 888 rt_sigprocmask
<br> 0.00 0.011483 26 445 uname
<br> 0.05 0.187592 632 297 execve
<br> 0.00 0.006622 22 297 arch_prctl
<br> 0.00 0.003066 20 152 getpid
<br> 0.00 0.007155 48 148 pipe
<br> 0.00 0.004293 29 148 dup2
<br> 0.02 0.063565 429 148 clone
<br> 0.00 0.008626 58 148 wait4
<br> 0.00 0.004727 32 148 getuid
<br> 0.00 0.003798 26 148 getgid
<br> 0.00 0.007550 51 148 geteuid
<br> 0.00 0.004409 30 148 getegid
<br> 0.00 0.003346 23 148 getppid
<br> 0.00 0.005175 35 148 getpgrp
<br> 0.00 0.001957 16 122 fcntl
<br> 0.00 0.000308 10 30 lstat
<br> 0.00 0.000113 5 24 getrusage
<br> 0.00 0.000156 14 11 getcwd
<br> 0.00 0.000277 35 8 mkdir
<br> 0.00 0.000206 26 8 unlink
<br> 0.00 0.000210 30 7 chdir
<br> 0.00 0.000044 15 3 chmod
<br>------ ----------- ----------- --------- --------- ----------------
<br>100.00 355.112190 4499405 1698 total
<br>==== label:2pass ====
<br>% time seconds usecs/call calls errors syscall
<br>------ ----------- ----------- --------- --------- ----------------
<br> 47.37 131.383273 87 1508223 47 write
<br> 39.64 109.940091 82 1337643 read
<br> 1.58 4.381549 31 142934 mmap
<br> 1.37 3.804107 27 141797 close
<br> 1.55 4.287069 30 141481 fstat
<br> 2.69 7.466272 53 141393 open
<br> 1.57 4.343462 31 140907 munmap
<br> 1.56 4.317122 36 119833 lseek
<br> 0.32 0.896082 11 82572 904 stat
<br> 2.06 5.719395 92 62208 pread
<br> 0.04 0.105069 8 13297 mremap
<br> 0.05 0.133677 25 5324 brk
<br> 0.01 0.017807 16 1115 rt_sigaction
<br> 0.01 0.019009 21 912 mprotect
<br> 0.01 0.014424 23 618 203 access
<br> 0.00 0.009484 16 606 rt_sigprocmask
<br> 0.00 0.004613 15 304 uname
<br> 0.11 0.299395 1475 203 execve
<br> 0.00 0.003417 17 203 arch_prctl
<br> 0.00 0.001624 16 104 getpid
<br> 0.00 0.004607 46 101 pipe
<br> 0.00 0.003418 34 101 dup2
<br> 0.07 0.182768 1810 101 clone
<br> 0.00 0.004143 41 101 wait4
<br> 0.00 0.001716 17 101 getuid
<br> 0.00 0.001417 14 101 getgid
<br> 0.00 0.001835 18 101 geteuid
<br> 0.00 0.001522 15 101 getegid
<br> 0.00 0.001358 13 101 getppid
<br> 0.00 0.001555 15 101 getpgrp
<br> 0.00 0.000110 4 29 fcntl
<br> 0.00 0.000519 19 27 lstat
<br> 0.00 0.000083 3 24 getrusage
<br> 0.00 0.000105 11 10 getcwd
<br> 0.00 0.004208 526 8 mkdir
<br> 0.00 0.004116 588 7 unlink
<br> 0.00 0.000193 39 5 chdir
<br> 0.00 0.000042 14 3 chmod
<br>------ ----------- ----------- --------- --------- ----------------
<br>100.00 277.360656 3842800 1154 total
<br><br><br>$ foreach cpumask ( 1 3 )
<br>foreach? foreach label ( default 2pass )
<br>foreach? echo ==== cpumask:$cpumask label:$label ====
<br>foreach? rm -fr linux-2.6.31; tar xfj ~/download/linux/linux-2.6.31.tar.bz2; sync
<br>foreach? (cd linux-2.6.31; setenv GTAGSLABEL $label; gtags; taskset $cpumask htags --statistics)
<br>foreach? end
<br>foreach? end
<br>==== cpumask:1 label:default ====
<br>period user[sec] system[sec] elapsed[sec] %CPU
<br>--------------------------------- --------- ----------- ------------ ----
<br>Time of making duplicate entries 43.791 2.732 54.398 85.5
<br>Time of making function index 4.544 0.620 5.628 91.8
<br>Time of making file index 0.388 0.336 2.189 33.1
<br>Time of making include file index 12.485 2.984 18.859 82.0
<br>Time of making hypertext 222.534 30.930 376.875 67.3
<br>--------------------------------- --------- ----------- ------------ ----
<br>The entire time 283.762 37.638 458.304 70.1
<br>==== cpumask:1 label:2pass ====
<br>period user[sec] system[sec] elapsed[sec] %CPU
<br>--------------------------------- --------- ----------- ------------ ----
<br>Time of making duplicate entries 42.275 2.500 46.125 97.1
<br>Time of making function index 4.460 0.684 5.189 99.1
<br>Time of making file index 0.404 0.328 2.719 26.9
<br>Time of making include file index 12.421 3.016 19.498 79.2
<br>Time of making hypertext 201.149 25.862 365.167 62.2
<br>--------------------------------- --------- ----------- ------------ ----
<br>The entire time 260.728 32.418 439.295 66.7
<br>==== cpumask:3 label:default ====
<br>period user[sec] system[sec] elapsed[sec] %CPU
<br>--------------------------------- --------- ----------- ------------ -----
<br>Time of making duplicate entries 43.083 3.124 63.430 72.8
<br>Time of making function index 4.448 0.628 4.612 110.1
<br>Time of making file index 0.396 0.344 1.092 67.8
<br>Time of making include file index 12.065 3.048 18.916 79.9
<br>Time of making hypertext 217.850 30.350 284.112 87.4
<br>--------------------------------- --------- ----------- ------------ -----
<br>The entire time 277.865 37.510 372.440 84.7
<br>==== cpumask:3 label:2pass ====
<br>period user[sec] system[sec] elapsed[sec] %CPU
<br>--------------------------------- --------- ----------- ------------ -----
<br>Time of making duplicate entries 42.403 2.476 45.594 98.4
<br>Time of making function index 4.556 0.568 4.756 107.7
<br>Time of making file index 0.404 0.344 1.160 64.5
<br>Time of making include file index 12.229 3.128 19.792 77.6
<br>Time of making hypertext 197.568 25.990 297.968 75.0
<br>--------------------------------- --------- ----------- ------------ -----
<br>The entire time 257.192 32.526 369.692 78.4
<br><br>Diffstat:
<br> global/global.c | 87 +++++++++++++++++++++++++++++---------------------
<br> gtags/gtags.c | 2 -
<br> htags/anchor.c | 37 ++++++++++++++-------
<br> htags/anchor.h | 8 ++--
<br> htags/htags.c | 3 +
<br> htags/htags.h | 1
<br> libutil/format.h | 19 +++++++---
<br> libutil/pathconvert.c | 14 ++++++--
<br> libutil/pathconvert.h | 7 +---
<br> 9 files changed, 115 insertions(+), 63 deletions(-)
<br><br>On Sat, 21 Nov 2009 01:23:00 +0900, Hideki IWAMOTO wrote...
<div class='shrinkable-quote'><div class='shrinkable-quote'><br>> Hi.
<br>>
<br>> This patch allows specifying both of -r and -s for "global -f" command in 2-pass mode.
<br>>
<br>>
<br>> Example:
<br>> $ env GTAGSLABEL=gtags global -fr test.c
<br>> g 1 test.c void f(void){g();}
<br>> $ env GTAGSLABEL=gtags global -fs test.c
<br>> h 2 test.c void g(void){h();}
<br>> $ env GTAGSLABEL=gtags global -frs test.c
<br>> global: both of -s and -r are not allowed.
<br>> $ env GTAGSLABEL=gtags-2pass global -fr test.c
<br>> g 1 test.c void f(void){g();}
<br>> $ env GTAGSLABEL=gtags-2pass global -fs test.c
<br>> h 2 test.c void g(void){h();}
<br>> $ env GTAGSLABEL=gtags-2pass global -frs test.c
<br>> g 1 test.c void f(void){g();}
<br>> h 2 test.c void g(void){h();}
<br>>
<br>> ----
<br>> Hideki IWAMOTO <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26458467&i=0" target="_top" rel="nofollow">h-iwamoto@...</a>
<br>> ______________________________________________________________________
<br>>
<br>> _______________________________________________
<br>> Bug-global mailing list
<br>> <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26458467&i=1" target="_top" rel="nofollow">Bug-global@...</a>
<br>> <a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a></div></div>----
<br>Hideki IWAMOTO <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26458467&i=2" target="_top" rel="nofollow">h-iwamoto@...</a>
<br><br /> <br />_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26458467&i=3" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><div class="small"><br/><img src="http://old.nabble.com/images/icon_attachment.gif" > <strong>20091121-global-frs+ctags-xtype.patch</strong> (23K) <a href="http://old.nabble.com/attachment/26458467/0/20091121-global-frs%2Bctags-xtype.patch" target="_top">Download Attachment</a></div><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>
tag:old.nabble.com,2006:post-26455226
Re: Binary recognition is to narrow [new suggestion]
2009-11-21T02:21:25Z
2009-11-21T02:21:25Z
Hideki IWAMOTO
Hi.
<br><br>> if (c <= 8)
<br>> return 1;
<br>> if (c >= 14 && c < 32)
<br>> return 1;
<br><br>You had better use table look-up like attached patch.
<br><br>On Fri, 20 Nov 2009 17:33:01 +0100 (CET), Erik Jonsson wrote...
<div class='shrinkable-quote'><div class='shrinkable-quote'><br>> Hi again,
<br>>
<br>> I have done some more testing and calculations now. The probability that a
<br>> binary file will pass as a text-file is quite high if one only tests the
<br>> first 32 bytes. I have therefore tested the performance if one where to
<br>> use the first 512 bytes. What I found was that the performance hit was
<br>> minimal however the benefits are several.
<br>>
<br>> Instead of counting characters over 127 the only test is that the first
<br>> 511 bytes don't contain any of the controll characters 0-8, 14-31. No
<br>> normal textfile would contain these.
<br>>
<br>> Assuming that binary data is random the probability of a incorrectly
<br>> tagged binary would be
<br>>
<br>> ((256-8-18)/256)^511=.00000000000000000000000170726
<br>>
<br>> just testing 127 bits would be a bit to little
<br>>
<br>> ((256-8-18)/256)^127=.00000123868
<br>>
<br>> One of the benefits is that this will correctly tag files in uni-code as
<br>> text as well. Since those control characters never appears in uni-code
<br>> either.
<br>>
<br>> The performance hit seems minimal on my computer.
<br>>
<br>> 511 byte version
<br>> erikj@deer:~/source/dps/src$ time ~/install/global-5.7.6/gtags/gtags
<br>> real 0m34.425s
<br>> user 0m8.337s
<br>> sys 0m3.080s
<br>>
<br>> 32 byte version
<br>> erikj@deer:~/source/dps/src$ time gtags
<br>> real 0m32.120s
<br>> user 0m8.361s
<br>> sys 0m2.820s
<br>>
<br>>
<br>> I have tried to clear the cache as good as possible between the runs.
<br>>
<br>> Here is the 511 byte is_binary that I'm using.
<br>>
<br>> static int
<br>> is_binary(const char *path)
<br>> {
<br>> int ip;
<br>> char buf[512];
<br>> char *cp;
<br>> int i, c, size;
<br>>
<br>> ip = open(path, O_RDONLY);
<br>> if (ip < 0)
<br>> die("cannot open file '%s' in read mode.", path);
<br>> size = read(ip, buf, sizeof(buf)-1);
<br>> close(ip);
<br>>
<br>> buf[size] = 0; //Terminate the data
<br>>
<br>> if (size <= 0)
<br>> return 1;
<br>> if (size >= 7 && locatestring(buf, "!<arch>", MATCH_AT_FIRST))
<br>> return 1;
<br>> cp = buf;
<br>> while ((c = (unsigned char) *cp)) {
<br>> if (c <= 8)
<br>> return 1;
<br>> if (c >= 14 && c < 32)
<br>> return 1;
<br>> cp++;
<br>> }
<br>>
<br>> return cp != buf+size;
<br>> }
<br>>
<br>> feel free to use the code as you like.
<br>>
<br>> /Erik J.
<br>>
<br>>
<br>>
<br>>
<br>>
<br>> _______________________________________________
<br>> Bug-global mailing list
<br>> <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26455226&i=0" target="_top" rel="nofollow">Bug-global@...</a>
<br>> <a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a></div></div>----
<br>Hideki IWAMOTO <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26455226&i=1" target="_top" rel="nofollow">h-iwamoto@...</a>
<br><br /> <br />_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26455226&i=2" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><div class="small"><br/><img src="http://old.nabble.com/images/icon_attachment.gif" > <strong>20091121-binarychar.patch</strong> (3K) <a href="http://old.nabble.com/attachment/26455226/0/20091121-binarychar.patch" target="_top">Download Attachment</a></div><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>
tag:old.nabble.com,2006:post-26454162
Re: Binary recognition is to narrow [new suggestion]
2009-11-20T22:42:11Z
2009-11-20T22:42:11Z
Shigio YAMAGUCHI-2
<div class='shrinkable-quote'>> Instead of counting characters over 127 the only test is that the first
<br>> 511 bytes don't contain any of the controll characters 0-8, 14-31. No
<br>> normal textfile would contain these.
<br>>
<br>> Assuming that binary data is random the probability of a incorrectly
<br>> tagged binary would be
<br>>
<br>> ((256-8-18)/256)^511=.00000000000000000000000170726
<br>>
<br>> just testing 127 bits would be a bit to little
<br>>
<br>> ((256-8-18)/256)^127=.00000123868
</div><br>This is a very interesting idea.
<br><br>> One of the benefits is that this will correctly tag files in uni-code as
<br>> text as well. Since those control characters never appears in uni-code
<br>> either.
<br><br>This is a big merit.
<br>Most other multi-byte character set are sure to be designed like that,
<br><br>I would like to make the 512 a customizable variable too.
<br><br>$ gtags ... use conventional test
<br><br>[File gtags.conf]
<br>+----------------------------
<br>|...
<br>| :binarytest_size=512:... ----------------------------------+
<br>| |
<br> v
<br>$ gtags ... use new test using the first n=512 bytes
<br><br>After testing for a while, we can decide what we should do.
<br>Thank you for your profitable consideration.
<br>--
<br>Shigio YAMAGUCHI <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26454162&i=0" target="_top" rel="nofollow">shigio@...</a>>
<br>PGP fingerprint: D1CB 0B89 B346 4AB6 5663 C4B6 3CA5 BBB3 57BE DDA3
<br><br><br>_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26454162&i=1" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>
tag:old.nabble.com,2006:post-26454519
Re: Binary recognition is to narrow [new suggestion]
2009-11-20T22:33:53Z
2009-11-20T22:33:53Z
Jason Hood
Erik Jonsson wrote:
<br>> Instead of counting characters over 127 the only test is that the first
<br>> 511 bytes don't contain any of the controll characters 0-8, 14-31. No
<br>> normal textfile would contain these.
<br><br>No normal source file, but if you want to generalise to
<br>text files, 8 (backspace) and 27 (escape) could probably
<br>occur (man files being a prime example).
<br><br>> One of the benefits is that this will correctly tag files in uni-code as
<br>> text as well. Since those control characters never appears in uni-code
<br>> either.
<br><br>I guess you mean UTF8, since UTF16/32 would most likely have
<br>a few 0s.
<br><br>Jason.
<br><br><br><br>_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26454519&i=0" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>
tag:old.nabble.com,2006:post-26453802
Binary recognition is to narrow [new suggestion]
2009-11-20T08:33:01Z
2009-11-20T08:33:01Z
Erik Jonsson-2
Hi again,
<br><br>I have done some more testing and calculations now. The probability that a
<br>binary file will pass as a text-file is quite high if one only tests the
<br>first 32 bytes. I have therefore tested the performance if one where to
<br>use the first 512 bytes. What I found was that the performance hit was
<br>minimal however the benefits are several.
<br><br>Instead of counting characters over 127 the only test is that the first
<br>511 bytes don't contain any of the controll characters 0-8, 14-31. No
<br>normal textfile would contain these.
<br><br>Assuming that binary data is random the probability of a incorrectly
<br>tagged binary would be
<br><br>((256-8-18)/256)^511=.00000000000000000000000170726
<br><br>just testing 127 bits would be a bit to little
<br><br>((256-8-18)/256)^127=.00000123868
<br><br>One of the benefits is that this will correctly tag files in uni-code as
<br>text as well. Since those control characters never appears in uni-code
<br>either.
<br><br>The performance hit seems minimal on my computer.
<br><br>511 byte version
<br>erikj@deer:~/source/dps/src$ time ~/install/global-5.7.6/gtags/gtags
<br>real 0m34.425s
<br>user 0m8.337s
<br>sys 0m3.080s
<br><br>32 byte version
<br>erikj@deer:~/source/dps/src$ time gtags
<br>real 0m32.120s
<br>user 0m8.361s
<br>sys 0m2.820s
<br><br><br>I have tried to clear the cache as good as possible between the runs.
<br><br>Here is the 511 byte is_binary that I'm using.
<br><br>static int
<br>is_binary(const char *path)
<br>{
<br> int ip;
<br> char buf[512];
<br> char *cp;
<br> int i, c, size;
<br><br> ip = open(path, O_RDONLY);
<br> if (ip < 0)
<br> die("cannot open file '%s' in read mode.", path);
<br> size = read(ip, buf, sizeof(buf)-1);
<br> close(ip);
<br><br> buf[size] = 0; //Terminate the data
<br><br> if (size <= 0)
<br> return 1;
<br> if (size >= 7 && locatestring(buf, "!<arch>", MATCH_AT_FIRST))
<br> return 1;
<br> cp = buf;
<br> while ((c = (unsigned char) *cp)) {
<br> if (c <= 8)
<br> return 1;
<br> if (c >= 14 && c < 32)
<br> return 1;
<br> cp++;
<br> }
<br><br> return cp != buf+size;
<br>}
<br><br>feel free to use the code as you like.
<br><br>/Erik J.
<br><br><br><br><br><br>_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26453802&i=0" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>
tag:old.nabble.com,2006:post-26446145
[PATCH] Allow specifying both of -r and -s for -f on 2-pass mode
2009-11-20T08:23:00Z
2009-11-20T08:23:00Z
Hideki IWAMOTO
Hi.
<br><br>This patch allows specifying both of -r and -s for "global -f" command in 2-pass mode.
<br><br><br>Example:
<br>$ env GTAGSLABEL=gtags global -fr test.c
<br>g 1 test.c void f(void){g();}
<br>$ env GTAGSLABEL=gtags global -fs test.c
<br>h 2 test.c void g(void){h();}
<br>$ env GTAGSLABEL=gtags global -frs test.c
<br>global: both of -s and -r are not allowed.
<br>$ env GTAGSLABEL=gtags-2pass global -fr test.c
<br>g 1 test.c void f(void){g();}
<br>$ env GTAGSLABEL=gtags-2pass global -fs test.c
<br>h 2 test.c void g(void){h();}
<br>$ env GTAGSLABEL=gtags-2pass global -frs test.c
<br>g 1 test.c void f(void){g();}
<br>h 2 test.c void g(void){h();}
<br><br>----
<br>Hideki IWAMOTO <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26446145&i=0" target="_top" rel="nofollow">h-iwamoto@...</a>
<br><br /> <br />_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26446145&i=1" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><div class="small"><br/><img src="http://old.nabble.com/images/icon_attachment.gif" > <strong>20091120-global-frs.patch</strong> (3K) <a href="http://old.nabble.com/attachment/26446145/0/20091120-global-frs.patch" target="_top">Download Attachment</a></div><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>
tag:old.nabble.com,2006:post-26425530
Re: Binary recognition is to narrow.
2009-11-19T04:40:12Z
2009-11-19T04:40:12Z
Hideki IWAMOTO
Hi.
<br><br>> if (((float)strange)/size > 0.3f)
<br><br>When an empty file is tested on old systems, SIGFPE may occur here.
<br>The calculation of 0.0/0.0 should be avoided for portability.
<br><br><br>On Wed, 18 Nov 2009 17:46:01 +0100 (CET), Erik Jonsson wrote...
<div class='shrinkable-quote'><br>> Hi all,
<br>>
<br>> The function is_binary is a bit naive and therefore tags to much as binary
<br>> content. It is quite common with names in the first 32 bytes and it's also
<br>> common with strange characters in names. Strange as in char > 127.
<br>>
<br>> I have rewritten the function a bit and think you probably should
<br>> incorporate this fix or one of your own in a future release.
<br>>
<br>> Here is my version...
<br>>
<br>> static int
<br>> is_binary(const char *path)
<br>> {
<br>> int ip;
<br>> char buf[32];
<br>> int i, c, size;
<br>> int strange = 0;
<br>>
<br>> ip = open(path, O_RDONLY);
<br>> if (ip < 0)
<br>> die("cannot open file '%s' in read mode.", path);
<br>> size = read(ip, buf, sizeof(buf));
<br>> close(ip);
<br>> if (size < 0)
<br>> return 1;
<br>> if (size >= 7 && locatestring(buf, "!<arch>", MATCH_AT_FIRST))
<br>> return 1;
<br>> for (i = 0; i < size; i++) {
<br>> c = (unsigned char)buf[i];
<br>> if (c <= 8)
<br>> return 1;
<br>> if (c >= 14 && c < 32)
<br>> return 1;
<br>> if (c > 128)
<br>> strange++;
<br>>
<br>> }
<br>> if (((float)strange)/size > 0.3f)
<br>> return 1;
<br>>
<br>> return 0;
<br>> }
<br>>
<br>>
<br>>
<br>>
<br>> _______________________________________________
<br>> Bug-global mailing list
<br>> <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26425530&i=0" target="_top" rel="nofollow">Bug-global@...</a>
<br>> <a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a></div><br>----
<br>Hideki IWAMOTO <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26425530&i=1" target="_top" rel="nofollow">h-iwamoto@...</a>
<br><br><br>_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26425530&i=2" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>
tag:old.nabble.com,2006:post-26421607
Re: Binary recognition is to narrow.
2009-11-18T23:18:15Z
2009-11-18T23:18:15Z
Shigio YAMAGUCHI-2
Hi
<br>> The function is_binary is a bit naive and therefore tags to much as binary
<br>> content. It is quite common with names in the first 32 bytes and it's also
<br>> common with strange characters in names. Strange as in char > 127.
<br>>
<br>> I have rewritten the function a bit and think you probably should
<br>> incorporate this fix or one of your own in a future release.
<br>>
<br>> Here is my version...
<br><br>I agree. It's a good idea.
<br><br>> if (((float)strange)/size > 0.3f)
<br>> return 1;
<br><br>I would like to make the 0.3 a customizable variable like follows:
<br><br>[File gtags.conf]
<br>+----------------------------
<br>|...
<br>| :binary_boundary=30:... ---+ (0-100%)
<br>| |
<br> |
<br>[Initial processing] |
<br> |
<br>Load binary_boundary |
<br> |
<br> binary_boundary = 0.3; <---+
<br><br>[Function is_binary()]
<br><br> if (((float)strange)/size > binary_boundary)
<br> return 1;
<br><br>The default of binary_boundary might be 0 (same as the current spec).
<br>Any comment?
<br>--
<br>Shigio YAMAGUCHI <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26421607&i=0" target="_top" rel="nofollow">shigio@...</a>>
<br>PGP fingerprint: D1CB 0B89 B346 4AB6 5663 C4B6 3CA5 BBB3 57BE DDA3
<br><br><br>_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26421607&i=1" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>
tag:old.nabble.com,2006:post-26419616
Binary recognition is to narrow.
2009-11-18T08:46:01Z
2009-11-18T08:46:01Z
Erik Jonsson-2
Hi all,
<br><br>The function is_binary is a bit naive and therefore tags to much as binary
<br>content. It is quite common with names in the first 32 bytes and it's also
<br>common with strange characters in names. Strange as in char > 127.
<br><br>I have rewritten the function a bit and think you probably should
<br>incorporate this fix or one of your own in a future release.
<br><br>Here is my version...
<br><br>static int
<br>is_binary(const char *path)
<br>{
<br> int ip;
<br> char buf[32];
<br> int i, c, size;
<br> int strange = 0;
<br><br> ip = open(path, O_RDONLY);
<br> if (ip < 0)
<br> die("cannot open file '%s' in read mode.", path);
<br> size = read(ip, buf, sizeof(buf));
<br> close(ip);
<br> if (size < 0)
<br> return 1;
<br> if (size >= 7 && locatestring(buf, "!<arch>", MATCH_AT_FIRST))
<br> return 1;
<br> for (i = 0; i < size; i++) {
<br> c = (unsigned char)buf[i];
<br> if (c <= 8)
<br> return 1;
<br> if (c >= 14 && c < 32)
<br> return 1;
<br> if (c > 128)
<br> strange++;
<br><br> }
<br> if (((float)strange)/size > 0.3f)
<br> return 1;
<br><br> return 0;
<br>}
<br><br><br><br><br>_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26419616&i=0" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>
tag:old.nabble.com,2006:post-26409511
Re: [PATCH] Add --statistics option to gtags
2009-11-18T07:10:37Z
2009-11-18T07:10:37Z
Shigio YAMAGUCHI-2
Hi
<br>> This patch add --statistics option to gtags.
<br><br>It's very convenient. Thank you.
<br>--
<br>Shigio YAMAGUCHI <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26409511&i=0" target="_top" rel="nofollow">shigio@...</a>>
<br>PGP fingerprint: D1CB 0B89 B346 4AB6 5663 C4B6 3CA5 BBB3 57BE DDA3
<br><br><br>_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26409511&i=1" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>
tag:old.nabble.com,2006:post-26405466
[PATCH] Add --statistics option to gtags
2009-11-18T02:18:02Z
2009-11-18T02:18:02Z
Hideki IWAMOTO
Hi.
<br><br>This patch add --statistics option to gtags.
<br>The output is as follows.
<br><br>$ foreach cpumask ( 1 3 )
<br>foreach? foreach label ( gtags gtags-2pass )
<br>foreach? echo ==== cpumask:$cpumask label:$label ====
<br>foreach? rm -fr linux-2.6.31; tar xfj ~/download/linux/linux-2.6.31.tar.bz2; sync
<br>foreach? (cd linux-2.6.31; env GTAGSLABEL=$label taskset $cpumask gtags --statistics)
<br>foreach? end
<br>foreach? end
<br>==== cpumask:1 label:gtags ====
<br>period user[sec] system[sec] elapsed[sec] %CPU
<br>----------------------- --------- ----------- ------------ ----
<br>Time of creating GTAGS 17.237 2.876 25.743 78.1
<br>Time of creating GRTAGS 52.967 6.900 68.607 87.3
<br>Time of creating GSYMS 46.519 6.328 57.816 91.4
<br>----------------------- --------- ----------- ------------ ----
<br>The entire time 116.723 16.105 152.166 87.3
<br>==== cpumask:1 label:gtags-2pass ====
<br>period user[sec] system[sec] elapsed[sec] %CPU
<br>--------------------------------- --------- ----------- ------------ ----
<br>Time of creating GTAGS 17.349 2.836 25.707 78.5
<br>Time of creating GRTAGS and GSYMS 75.681 8.137 92.310 90.8
<br>--------------------------------- --------- ----------- ------------ ----
<br>The entire time 93.030 10.973 118.018 88.1
<br>==== cpumask:3 label:gtags ====
<br>period user[sec] system[sec] elapsed[sec] %CPU
<br>----------------------- --------- ----------- ------------ -----
<br>Time of creating GTAGS 16.909 2.868 24.126 82.0
<br>Time of creating GRTAGS 51.471 6.768 58.139 100.2
<br>Time of creating GSYMS 45.807 6.076 51.598 100.6
<br>----------------------- --------- ----------- ------------ -----
<br>The entire time 114.187 15.713 133.863 97.0
<br>==== cpumask:3 label:gtags-2pass ====
<br>period user[sec] system[sec] elapsed[sec] %CPU
<br>--------------------------------- --------- ----------- ------------ -----
<br>Time of creating GTAGS 17.217 2.804 24.136 83.0
<br>Time of creating GRTAGS and GSYMS 73.109 7.468 66.939 120.4
<br>--------------------------------- --------- ----------- ------------ -----
<br>The entire time 90.326 10.273 91.074 110.5
<br><br><br>Diffstat:
<br> gtags/gtags.c | 12 +++++++++++-
<br> gtags/manual.in | 3 +++
<br> 2 files changed, 14 insertions(+), 1 deletion(-)
<br><br><br><br><br>----
<br>Hideki IWAMOTO <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26405466&i=0" target="_top" rel="nofollow">h-iwamoto@...</a>
<br><br /> <br />_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26405466&i=1" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><div class="small"><br/><img src="http://old.nabble.com/images/icon_attachment.gif" > <strong>20091116-gtags-statistics.patch</strong> (4K) <a href="http://old.nabble.com/attachment/26405466/0/20091116-gtags-statistics.patch" target="_top">Download Attachment</a></div><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>
tag:old.nabble.com,2006:post-26399509
Re: [PATCH] Add new command --file-list (-F) to global, and use it in htags
2009-11-17T15:07:41Z
2009-11-17T15:07:41Z
Hideki IWAMOTO
> How about adding --args-file option instead of the -F command?
<br><br>OK. This is revised edition.
<br><br> global/global.c | 87 ++++++++++++++++++++++++++++++++++++++++++++++---------
<br> global/manual.in | 5 +++
<br> gtags.conf.in | 2 +
<br> htags/anchor.c | 63 ++++++++++++++++++++++++++++++---------
<br> htags/anchor.h | 3 +
<br> htags/htags.c | 16 +---------
<br> 6 files changed, 133 insertions(+), 43 deletions(-)
<br><br>On Tue, 17 Nov 2009 22:31:33 +0900, Shigio YAMAGUCHI wrote...
<div class='shrinkable-quote'><div class='shrinkable-quote'><br>> Hello again,
<br>> > -f, --file files
<br>> > Print all tags in the files.
<br>> > This option implies the -x option.
<br>> > -F, --file-list file
<br>> > Read from file a list of file names and print all tags
<br>> > in each file in the list.
<br>> > If file is -, read from standard input.
<br>> > File names must be separated by newline.
<br>> > This option implies the -x option.
<br>>
<br>> How about adding --args-file option instead of the -F command?
<br>>
<br>> Commands:
<br>> -f, --file files
<br>> Print all tags in the files.
<br>> This option implies the -x option.
<br>> If the --args-file options is specified, [files] on the command line
<br>> is ignored.
<br>>
<br>> Options:
<br>> --args-file file
<br>> Read from file a list of arguments.
<br>> If file is -, read from standard input.
<br>> Arguments must be separated by newline.
<br>> This option can be used only with the -f(--file-list) command now.
<br>>
<br>>
<br>> I don't think that a usual user uses the -F command. So, it is not necessary
<br>> to offer it as a command. The --args-file=file option can be used with other
<br>> commands in the future.
<br>>
<br>> What do you think?
<br>> --
<br>> Shigio YAMAGUCHI <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26399509&i=0" target="_top" rel="nofollow">shigio@...</a>>
<br>> PGP fingerprint: D1CB 0B89 B346 4AB6 5663 C4B6 3CA5 BBB3 57BE DDA3
</div></div>----
<br>Hideki IWAMOTO <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26399509&i=1" target="_top" rel="nofollow">h-iwamoto@...</a>
<br><br /> <br />_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26399509&i=2" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><div class="small"><br/><img src="http://old.nabble.com/images/icon_attachment.gif" > <strong>20091118-global-args-file.patch</strong> (16K) <a href="http://old.nabble.com/attachment/26399509/0/20091118-global-args-file.patch" target="_top">Download Attachment</a></div><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>
tag:old.nabble.com,2006:post-26390020
Re: [PATCH] Add new command --file-list (-F) to global, and use it in htags
2009-11-17T05:31:33Z
2009-11-17T05:31:33Z
Shigio YAMAGUCHI-2
Hello again,
<br>> -f, --file files
<br>> Print all tags in the files.
<br>> This option implies the -x option.
<br>> -F, --file-list file
<br>> Read from file a list of file names and print all tags
<br>> in each file in the list.
<br>> If file is -, read from standard input.
<br>> File names must be separated by newline.
<br>> This option implies the -x option.
<br><br>How about adding --args-file option instead of the -F command?
<br><br>Commands:
<br>-f, --file files
<br> Print all tags in the files.
<br> This option implies the -x option.
<br> If the --args-file options is specified, [files] on the command line
<br> is ignored.
<br><br>Options:
<br>--args-file file
<br> Read from file a list of arguments.
<br> If file is -, read from standard input.
<br> Arguments must be separated by newline.
<br> This option can be used only with the -f(--file-list) command now.
<br><br><br>I don't think that a usual user uses the -F command. So, it is not necessary
<br>to offer it as a command. The --args-file=file option can be used with other
<br>commands in the future.
<br><br>What do you think?
<br>--
<br>Shigio YAMAGUCHI <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26390020&i=0" target="_top" rel="nofollow">shigio@...</a>>
<br>PGP fingerprint: D1CB 0B89 B346 4AB6 5663 C4B6 3CA5 BBB3 57BE DDA3
<br><br><br>_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26390020&i=1" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>
tag:old.nabble.com,2006:post-26388361
Re: [PATCH] Add new command --file-list (-F) to global, and use it in htags
2009-11-17T02:46:56Z
2009-11-17T02:46:56Z
Hideki IWAMOTO
> In a word, the 2pass mode speeds up gtags(1), but speeds down htags(1).
<br>> You did an optimization for it, didn't you?
<br><br>Yes. A main purpose of this change is to improve 2-pass modes.
<br>
<br><br>On Tue, 17 Nov 2009 14:47:30 +0900, Shigio YAMAGUCHI wrote...
<div class='shrinkable-quote'><br>> Hi,
<br>> > This patch decreases the frequency that htags executes global command
<br>> > by changing the method of passing a list of path name from command line argument
<br>> > to temporary file.
<br>>
<br>> Bringing this result together:
<br>>
<br>> Htags executing time (second)
<br>>
<br>> default 2pass-mode
<br>> ------------------------------------------
<br>> Current 355 459
<br>> Patch 356 417
<br>>
<br>> In a word, the 2pass mode speeds up gtags(1), but speeds down htags(1).
<br>> You did an optimization for it, didn't you? The effect seems to be clear.
<br>> I feel that we are walking on a correct road.
<br>>
<br>> Thank you for your patch.
<br>> --
<br>> Shigio YAMAGUCHI <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26388361&i=0" target="_top" rel="nofollow">shigio@...</a>>
<br>> PGP fingerprint: D1CB 0B89 B346 4AB6 5663 C4B6 3CA5 BBB3 57BE DDA3
</div><br>----
<br>Hideki IWAMOTO <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26388361&i=1" target="_top" rel="nofollow">h-iwamoto@...</a>
<br><br><br>_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26388361&i=2" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>
tag:old.nabble.com,2006:post-26384695
Re: [PATCH] Add new command --file-list (-F) to global, and use it in htags
2009-11-16T21:36:26Z
2009-11-16T21:36:26Z
Shigio YAMAGUCHI-2
Hi,
<br>> This patch decreases the frequency that htags executes global command
<br>> by changing the method of passing a list of path name from command line argument
<br>> to temporary file.
<br><br>Bringing this result together:
<br><br>Htags executing time (second)
<br><br> default 2pass-mode
<br>------------------------------------------
<br>Current 355 459
<br>Patch 356 417
<br><br>In a word, the 2pass mode speeds up gtags(1), but speeds down htags(1).
<br>You did an optimization for it, didn't you? The effect seems to be clear.
<br>I feel that we are walking on a correct road.
<br><br>Thank you for your patch.
<br>--
<br>Shigio YAMAGUCHI <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26384695&i=0" target="_top" rel="nofollow">shigio@...</a>>
<br>PGP fingerprint: D1CB 0B89 B346 4AB6 5663 C4B6 3CA5 BBB3 57BE DDA3
<br><br><br>_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26384695&i=1" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>
tag:old.nabble.com,2006:post-26370967
[PATCH] Add new command --file-list (-F) to global, and use it in htags
2009-11-16T04:23:47Z
2009-11-16T04:23:47Z
Hideki IWAMOTO
Hi.
<br><br>This patch decreases the frequency that htags executes global command
<br>by changing the method of passing a list of path name from command line argument
<br>to temporary file.
<br><br>Diffstat:
<br> global/global.c | 262 +++++++++++++++++++++++++++++++++++++++----------------
<br> global/manual.in | 7 +
<br> gtags.conf.in | 2
<br> htags/anchor.c | 67 ++++++++++----
<br> htags/anchor.h | 3
<br> htags/htags.c | 16 ---
<br> 6 files changed, 253 insertions(+), 104 deletions(-)
<br><br><br>The frequency of system call under condition of
<br> GTAGSCACHE=134217728
<br> htags_options=-aDfFInosx
<br>is as follows.
<br><br>======== current CVS version ========
<br>$ foreach label ( default 2pass )
<br>foreach? echo ==== label:$label ====
<br>foreach? rm -fr linux-2.6.31; tar xfj ~/download/linux/linux-2.6.31.tar.bz2; sync
<br>foreach? (cd linux-2.6.31; setenv GTAGSLABEL $label; gtags; strace -c -f -F -q -S calls htags)
<br>foreach? end
<br>==== label:default ====
<br>% time seconds usecs/call calls errors syscall
<br>------ ----------- ----------- --------- --------- ----------------
<br> 34.40 122.207052 81 1505061 88 write
<br> 43.17 153.368494 106 1451899 14 read
<br> 8.69 30.868833 82 378630 pread
<br> 2.55 9.048733 50 180246 lseek
<br> 1.78 6.329060 36 175892 mmap
<br> 1.64 5.833524 34 171900 close
<br> 1.74 6.179133 36 170852 fstat
<br> 2.67 9.479246 56 170136 open
<br> 1.82 6.456042 39 167528 munmap
<br> 0.65 2.311625 20 113942 3485 stat
<br> 0.32 1.144488 43 26797 brk
<br> 0.03 0.123552 9 13329 mremap
<br> 0.02 0.079724 17 4602 rt_sigaction
<br> 0.02 0.067396 18 3765 mprotect
<br> 0.01 0.041663 17 2508 rt_sigprocmask
<br> 0.01 0.053235 21 2506 837 access
<br> 0.01 0.020611 16 1255 uname
<br> 0.27 0.958788 1146 837 execve
<br> 0.00 0.011786 14 837 arch_prctl
<br> 0.00 0.005685 13 435 lstat
<br> 0.00 0.007078 17 419 getpid
<br> 0.00 0.012843 31 418 pipe
<br> 0.00 0.009780 23 418 dup2
<br> 0.17 0.611744 1464 418 clone
<br> 0.01 0.019450 47 418 wait4
<br> 0.00 0.006608 16 418 getuid
<br> 0.00 0.006113 15 418 getgid
<br> 0.00 0.005857 14 418 geteuid
<br> 0.00 0.006093 15 418 getegid
<br> 0.00 0.006815 16 418 getppid
<br> 0.00 0.006409 15 418 getpgrp
<br> 0.00 0.003875 11 342 fcntl
<br> 0.00 0.008018 29 277 chdir
<br> 0.00 0.002327 16 146 getcwd
<br> 0.00 0.000107 4 24 getrusage
<br> 0.00 0.000259 32 8 mkdir
<br> 0.00 0.000126 25 5 unlink
<br> 0.00 0.000044 15 3 chmod
<br>------ ----------- ----------- --------- --------- ----------------
<br>100.00 355.302216 4548361 4424 total
<br>==== label:2pass ====
<br>% time seconds usecs/call calls errors syscall
<br>------ ----------- ----------- --------- --------- ----------------
<br> 45.03 206.807069 114 1806818 80 write
<br> 36.08 165.726434 95 1753566 7 read
<br> 6.75 30.985808 84 369432 pread
<br> 2.29 10.540039 58 180246 lseek
<br> 1.65 7.599827 43 175800 mmap
<br> 1.45 6.662274 39 171900 close
<br> 1.62 7.447363 44 170677 fstat
<br> 2.30 10.583143 62 170048 open
<br> 1.58 7.252197 43 167435 munmap
<br> 0.48 2.212786 19 113942 3485 stat
<br> 0.27 1.220022 47 26072 brk
<br> 0.02 0.111370 8 13300 mremap
<br> 0.01 0.063578 14 4602 rt_sigaction
<br> 0.02 0.070244 19 3765 mprotect
<br> 0.01 0.035519 14 2508 rt_sigprocmask
<br> 0.01 0.042286 17 2506 837 access
<br> 0.00 0.019244 15 1255 uname
<br> 0.23 1.071832 1281 837 execve
<br> 0.00 0.012099 14 837 arch_prctl
<br> 0.00 0.005013 12 435 lstat
<br> 0.00 0.005213 12 419 getpid
<br> 0.00 0.018551 44 418 pipe
<br> 0.00 0.009257 22 418 dup2
<br> 0.16 0.724096 1732 418 clone
<br> 0.00 0.018863 45 418 wait4
<br> 0.00 0.007321 18 418 getuid
<br> 0.00 0.005882 14 418 getgid
<br> 0.00 0.005414 13 418 geteuid
<br> 0.00 0.005453 13 418 getegid
<br> 0.00 0.006940 17 418 getppid
<br> 0.00 0.004826 12 418 getpgrp
<br> 0.00 0.008938 32 277 chdir
<br> 0.00 0.004600 18 254 fcntl
<br> 0.00 0.002458 17 146 getcwd
<br> 0.00 0.000108 5 24 getrusage
<br> 0.00 0.000219 27 8 mkdir
<br> 0.00 0.000114 23 5 unlink
<br> 0.00 0.000050 17 3 chmod
<br>------ ----------- ----------- --------- --------- ----------------
<br>100.00 459.296450 5141297 4409 total
<br><br>======== patched version ========
<br>$ foreach label ( default 2pass )
<br>foreach? echo ==== label:$label ====
<br>foreach? rm -fr linux-2.6.31; tar xfj ~/download/linux/linux-2.6.31.tar.bz2; sync
<br>foreach? (cd linux-2.6.31; setenv GTAGSLABEL $label; gtags; strace -c -f -F -q -S calls htags)
<br>foreach? end
<br>==== label:default ====
<br>% time seconds usecs/call calls errors syscall
<br>------ ----------- ----------- --------- --------- ----------------
<br> 33.82 120.521756 80 1505462 73 write
<br> 44.81 159.678560 110 1449895 read
<br> 8.26 29.443912 80 367080 pread
<br> 2.55 9.092706 51 179709 lseek
<br> 1.78 6.348294 38 169017 mmap
<br> 1.62 5.769033 34 167320 close
<br> 1.75 6.237081 37 167047 fstat
<br> 2.76 9.833835 59 166817 open
<br> 1.83 6.534108 39 166050 munmap
<br> 0.34 1.227782 11 109228 1325 stat
<br> 0.31 1.110834 46 24012 brk
<br> 0.04 0.126589 9 13329 mremap
<br> 0.01 0.042422 26 1632 rt_sigaction
<br> 0.01 0.037357 28 1335 mprotect
<br> 0.01 0.029811 34 889 297 access
<br> 0.01 0.029736 33 888 rt_sigprocmask
<br> 0.00 0.012792 29 445 uname
<br> 0.05 0.165096 556 297 execve
<br> 0.00 0.006907 23 297 arch_prctl
<br> 0.00 0.002422 16 152 getpid
<br> 0.00 0.007608 51 148 pipe
<br> 0.00 0.003994 27 148 dup2
<br> 0.01 0.051462 348 148 clone
<br> 0.00 0.010850 73 148 wait4
<br> 0.00 0.002834 19 148 getuid
<br> 0.00 0.002538 17 148 getgid
<br> 0.00 0.002972 20 148 geteuid
<br> 0.00 0.003358 23 148 getegid
<br> 0.00 0.003027 20 148 getppid
<br> 0.00 0.004144 28 148 getpgrp
<br> 0.00 0.001556 12 125 fcntl
<br> 0.00 0.000268 9 30 lstat
<br> 0.00 0.000082 3 24 getrusage
<br> 0.00 0.000126 11 11 getcwd
<br> 0.00 0.000259 32 8 mkdir
<br> 0.00 0.000172 22 8 unlink
<br> 0.00 0.000232 33 7 chdir
<br> 0.00 0.000042 14 3 chmod
<br>------ ----------- ----------- --------- --------- ----------------
<br>100.00 356.346557 4492597 1695 total
<br>==== label:2pass ====
<br>% time seconds usecs/call calls errors syscall
<br>------ ----------- ----------- --------- --------- ----------------
<br> 44.56 186.155840 103 1807202 57 write
<br> 40.49 169.143151 97 1751560 read
<br> 2.68 11.178268 62 179709 lseek
<br> 1.90 7.925562 47 168924 mmap
<br> 1.62 6.780514 41 167320 close
<br> 1.89 7.876922 47 166867 fstat
<br> 2.78 11.602942 70 166727 open
<br> 1.79 7.471974 45 165957 munmap
<br> 0.36 1.483747 14 109228 1325 stat
<br> 1.54 6.417259 86 74518 pread
<br> 0.09 0.381301 29 13300 mremap
<br> 0.07 0.285137 44 6483 brk
<br> 0.01 0.045652 28 1632 rt_sigaction
<br> 0.01 0.037993 28 1335 mprotect
<br> 0.01 0.033648 38 889 297 access
<br> 0.01 0.020896 24 888 rt_sigprocmask
<br> 0.00 0.014595 33 445 uname
<br> 0.12 0.498429 1678 297 execve
<br> 0.00 0.007364 25 297 arch_prctl
<br> 0.00 0.003032 20 152 getpid
<br> 0.00 0.009460 64 148 pipe
<br> 0.00 0.006375 43 148 dup2
<br> 0.08 0.329480 2226 148 clone
<br> 0.00 0.008658 59 148 wait4
<br> 0.00 0.003803 26 148 getuid
<br> 0.00 0.003516 24 148 getgid
<br> 0.00 0.002855 19 148 geteuid
<br> 0.00 0.002961 20 148 getegid
<br> 0.00 0.002904 20 148 getppid
<br> 0.00 0.003153 21 148 getpgrp
<br> 0.00 0.000141 4 35 fcntl
<br> 0.00 0.000222 7 30 lstat
<br> 0.00 0.000080 3 24 getrusage
<br> 0.00 0.000122 11 11 getcwd
<br> 0.00 0.004229 529 8 mkdir
<br> 0.00 0.000311 39 8 unlink
<br> 0.00 0.000387 55 7 chdir
<br> 0.00 0.000051 17 3 chmod
<br>------ ----------- ----------- --------- --------- ----------------
<br>100.00 417.742934 4785336 1679 total
<br><br><br><br>----
<br>Hideki IWAMOTO <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26370967&i=0" target="_top" rel="nofollow">h-iwamoto@...</a>
<br><br /> <br />_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26370967&i=1" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><div class="small"><br/><img src="http://old.nabble.com/images/icon_attachment.gif" > <strong>20091113-global-F.patch</strong> (22K) <a href="http://old.nabble.com/attachment/26370967/0/20091113-global-F.patch" target="_top">Download Attachment</a></div><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>
tag:old.nabble.com,2006:post-26319029
Broken link in download page
2009-11-12T05:03:52Z
2009-11-12T05:03:52Z
Durkin, Michael (NSN - GB/Huntingdon)
Hi,
<br><br>The link <a href="http://www.geocities.com/jadoxa/global/index.html" target="_top" rel="nofollow">http://www.geocities.com/jadoxa/global/index.html</a> is broken as
<br>geocities is no longer active. Is it possible to have the link for the
<br>DOS/Windows 32 version updated please?
<br><br>Thanks
<br><br>Mick Durkin
<br><br><br>_______________________________________________
<br>Help-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26319029&i=0" target="_top" rel="nofollow">Help-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/help-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/help-global</a><br><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Help-f1650.html" embed="fixTarget[1650]" target="_top" >Gnu - Global - Help</a></p>
tag:old.nabble.com,2006:post-26196218
Re: [PATCH] change output of htags --statistics
2009-11-04T05:23:29Z
2009-11-04T05:23:29Z
Hideki IWAMOTO
> There might not be necessity of it.
<br><br>The patch was simplified by dropping support of times.
<br>Attached is updated version.
<br><br><br> configure.ac | 2
<br> htags/htags.c | 50 +----
<br> libutil/Makefile.am | 4
<br> libutil/global.h | 1
<br> libutil/statistics.c | 436 +++++++++++++++++++++++++++++++++++++++++++++++++++
<br> libutil/statistics.h | 98 +++++++++++
<br> libutil/strbuf.c | 15 +
<br> libutil/strbuf.h | 4
<br> 8 files changed, 573 insertions(+), 37 deletions(-)
<br><br>On Wed, 04 Nov 2009 08:19:17 +0900, Shigio YAMAGUCHI wrote...
<div class='shrinkable-quote'><div class='shrinkable-quote'><br>> Hi
<br>> > > #elif HAVE_TIMES
<br>> >
<br>> > I don't know whether support of times(2) is needed.
<br>> > Is there anybody who is using the system that supports times
<br>> > and doesn't support gettimeofday and getrusage?
<br>>
<br>> There might not be necessity of it.
<br>> In FreeBSD, we can see the following description at the head of the DESCRIPTION of times(2).
<br>>
<br>> "This interface is obsoleted by getrusage(2) and gettimeofday(2)."
<br>>
<br>> As for either, those who should use times(2) is small number of people even in case of being.
<br>> --
<br>> Shigio YAMAGUCHI <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26196218&i=0" target="_top" rel="nofollow">shigio@...</a>>
<br>> PGP fingerprint: D1CB 0B89 B346 4AB6 5663 C4B6 3CA5 BBB3 57BE DDA3
</div></div>----
<br>Hideki IWAMOTO <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26196218&i=1" target="_top" rel="nofollow">h-iwamoto@...</a>
<br><br /> <br />_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26196218&i=2" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><div class="small"><br/><img src="http://old.nabble.com/images/icon_attachment.gif" > <strong>20091104-statistics-2.patch</strong> (33K) <a href="http://old.nabble.com/attachment/26196218/0/20091104-statistics-2.patch" target="_top">Download Attachment</a></div><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>
tag:old.nabble.com,2006:post-26188292
Re: [PATCH] change output of htags --statistics
2009-11-03T15:19:17Z
2009-11-03T15:19:17Z
Shigio YAMAGUCHI-2
Hi
<br>> > #elif HAVE_TIMES
<br>>
<br>> I don't know whether support of times(2) is needed.
<br>> Is there anybody who is using the system that supports times
<br>> and doesn't support gettimeofday and getrusage?
<br><br>There might not be necessity of it.
<br>In FreeBSD, we can see the following description at the head of the DESCRIPTION of times(2).
<br><br> "This interface is obsoleted by getrusage(2) and gettimeofday(2)."
<br><br>As for either, those who should use times(2) is small number of people even in case of being.
<br>--
<br>Shigio YAMAGUCHI <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26188292&i=0" target="_top" rel="nofollow">shigio@...</a>>
<br>PGP fingerprint: D1CB 0B89 B346 4AB6 5663 C4B6 3CA5 BBB3 57BE DDA3
<br><br><br>_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26188292&i=1" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>
tag:old.nabble.com,2006:post-26182431
Re: [PATCH] change output of htags --statistics
2009-11-03T08:35:13Z
2009-11-03T08:35:13Z
Hideki IWAMOTO
Hi.
<br><br>> #elif HAVE_TIMES
<br><br>I don't know whether support of times(2) is needed.
<br>Is there anybody who is using the system that supports times
<br>and doesn't support gettimeofday and getrusage?
<br><br><br>On Mon, 02 Nov 2009 10:53:35 +0900, Hideki IWAMOTO wrote...
<div class='shrinkable-quote'><br>> Hi.
<br>>
<br>> Currently, `htags --statistics' outputs elapsed times with the accuracy of second.
<br>> It is preferable that the CPU time is output and the elapsed time is output by higher accuracy.
<br>>
<br>> This patch changes `htags --statistics' as follows.
<br>>
<br>> #if HAVE_GETRUSAGE && HAVE_GETTIMEOFDAY
<br>> Use gettimeofday() to get elapsed time.
<br>> Use getrusage() to get CPU time.
<br>> #elif HAVE_TIMES
<br>> #if HAVE_GETTIMEOFDAY
<br>> Use gettimeofday() to get elapsed time.
<br>> Use times() to get CPU time.
<br>> #else
<br>> Use times() to get elapsed time and CPU time.
<br>> #endif
<br>> #else
<br>> #if HAVE_GETTIMEOFDAY
<br>> Use gettimeofday() to get elapsed time.
<br>> #else
<br>> Use time() to get elapsed time.
<br>> #endif
<br>> #endif
<br>>
<br>>
<br>> The output in each case is as follows.
<br>>
<br>> ======== Case 1 =============================================================
<br>> #define HAVE_GETRUSAGE 1
<br>> #define HAVE_GETTIMEOFDAY 1
<br>>
<br>> $ htags --statistics
<br>> period user[sec] system[sec] elapsed[sec] %CPU
<br>> --------------------------------- --------- ----------- ------------ -----
<br>> Time of making duplicate entries 1.988 0.184 1.420 153.0
<br>> Time of making function index 0.272 0.036 0.281 109.7
<br>> Time of making file index 0.028 0.012 0.042 96.3
<br>> Time of making include file index 1.120 0.092 1.216 99.7
<br>> Time of making hypertext 18.701 1.888 15.474 133.1
<br>> The entire time 22.113 2.220 18.440 132.0
<br>>
<br>> ======== Case 2 =============================================================
<br>> #undef HAVE_GETRUSAGE
<br>> #define HAVE_GETTIMEOFDAY 1
<br>> #define HAVE_TIMES 1
<br>>
<br>> $ htags --statistics
<br>> period user[sec] system[sec] elapsed[sec] %CPU
<br>> --------------------------------- --------- ----------- ------------ -----
<br>> Time of making duplicate entries 1.970 0.170 1.527 140.1
<br>> Time of making function index 0.290 0.020 0.278 111.6
<br>> Time of making file index 0.030 0.010 0.041 97.8
<br>> Time of making include file index 1.110 0.080 1.201 99.1
<br>> Time of making hypertext 18.060 1.980 15.221 131.7
<br>> The entire time 21.460 2.260 18.276 129.8
<br>>
<br>> ======== Case 3 =============================================================
<br>> #undef HAVE_GETRUSAGE
<br>> #undef HAVE_GETTIMEOFDAY
<br>> #define HAVE_TIMES 1
<br>>
<br>> $ htags --statistics
<br>> period user[sec] system[sec] elapsed[sec] %CPU
<br>> --------------------------------- --------- ----------- ------------ -----
<br>> Time of making duplicate entries 2.020 0.190 1.550 142.6
<br>> Time of making function index 0.260 0.040 0.280 107.1
<br>> Time of making file index 0.020 0.030 0.040 125.0
<br>> Time of making include file index 1.070 0.110 1.170 100.9
<br>> Time of making hypertext 18.580 2.190 15.310 135.7
<br>> The entire time 21.950 2.560 18.360 133.5
<br>>
<br>> ======== Case 4 =============================================================
<br>> #undef HAVE_GETRUSAGE
<br>> #define HAVE_GETTIMEOFDAY 1
<br>> #undef HAVE_TIMES
<br>>
<br>> $ htags --statistics
<br>> period elapsed[sec]
<br>> --------------------------------- ------------
<br>> Time of making duplicate entries 1.405
<br>> Time of making function index 0.294
<br>> Time of making file index 0.042
<br>> Time of making include file index 1.157
<br>> Time of making hypertext 15.555
<br>> The entire time 18.461
<br>>
<br>> ======== Case 5 =============================================================
<br>> #undef HAVE_GETRUSAGE
<br>> #undef HAVE_GETTIMEOFDAY
<br>> #undef HAVE_TIMES
<br>>
<br>> $ htags --statistics
<br>> period elapsed[sec]
<br>> --------------------------------- ------------
<br>> Time of making duplicate entries 2.000
<br>> Time of making function index 0.000
<br>> Time of making file index 0.000
<br>> Time of making include file index 1.000
<br>> Time of making hypertext 15.000
<br>> The entire time 18.000
<br>>
<br>>
<br>> Diffstat:
<br>> configure.ac | 2
<br>> htags/htags.c | 50 +----
<br>> libutil/Makefile.am | 4
<br>> libutil/global.h | 1
<br>> libutil/statistics.c | 493 +++++++++++++++++++++++++++++++++++++++++++++++++++
<br>> libutil/statistics.h | 84 ++++++++
<br>> libutil/strbuf.c | 15 +
<br>> libutil/strbuf.h | 4
<br>> 8 files changed, 616 insertions(+), 37 deletions(-)
<br>>
<br>> ----
<br>> Hideki IWAMOTO <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26182431&i=0" target="_top" rel="nofollow">h-iwamoto@...</a>
<br>> ______________________________________________________________________
<br>>
<br>> _______________________________________________
<br>> Bug-global mailing list
<br>> <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26182431&i=1" target="_top" rel="nofollow">Bug-global@...</a>
<br>> <a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a></div><br>----
<br>Hideki IWAMOTO <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26182431&i=2" target="_top" rel="nofollow">h-iwamoto@...</a>
<br><br><br>_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26182431&i=3" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>
tag:old.nabble.com,2006:post-26157218
[PATCH] change output of htags --statistics
2009-11-01T17:53:35Z
2009-11-01T17:53:35Z
Hideki IWAMOTO
Hi.
<br><br>Currently, `htags --statistics' outputs elapsed times with the accuracy of second.
<br>It is preferable that the CPU time is output and the elapsed time is output by higher accuracy.
<br><br>This patch changes `htags --statistics' as follows.
<br><br>#if HAVE_GETRUSAGE && HAVE_GETTIMEOFDAY
<br> Use gettimeofday() to get elapsed time.
<br> Use getrusage() to get CPU time.
<br>#elif HAVE_TIMES
<br> #if HAVE_GETTIMEOFDAY
<br> Use gettimeofday() to get elapsed time.
<br> Use times() to get CPU time.
<br> #else
<br> Use times() to get elapsed time and CPU time.
<br> #endif
<br>#else
<br> #if HAVE_GETTIMEOFDAY
<br> Use gettimeofday() to get elapsed time.
<br> #else
<br> Use time() to get elapsed time.
<br> #endif
<br>#endif
<br><br><br>The output in each case is as follows.
<br><br>======== Case 1 =============================================================
<br>#define HAVE_GETRUSAGE 1
<br>#define HAVE_GETTIMEOFDAY 1
<br><br>$ htags --statistics
<br>period user[sec] system[sec] elapsed[sec] %CPU
<br>--------------------------------- --------- ----------- ------------ -----
<br>Time of making duplicate entries 1.988 0.184 1.420 153.0
<br>Time of making function index 0.272 0.036 0.281 109.7
<br>Time of making file index 0.028 0.012 0.042 96.3
<br>Time of making include file index 1.120 0.092 1.216 99.7
<br>Time of making hypertext 18.701 1.888 15.474 133.1
<br>The entire time 22.113 2.220 18.440 132.0
<br><br>======== Case 2 =============================================================
<br>#undef HAVE_GETRUSAGE
<br>#define HAVE_GETTIMEOFDAY 1
<br>#define HAVE_TIMES 1
<br><br>$ htags --statistics
<br>period user[sec] system[sec] elapsed[sec] %CPU
<br>--------------------------------- --------- ----------- ------------ -----
<br>Time of making duplicate entries 1.970 0.170 1.527 140.1
<br>Time of making function index 0.290 0.020 0.278 111.6
<br>Time of making file index 0.030 0.010 0.041 97.8
<br>Time of making include file index 1.110 0.080 1.201 99.1
<br>Time of making hypertext 18.060 1.980 15.221 131.7
<br>The entire time 21.460 2.260 18.276 129.8
<br><br>======== Case 3 =============================================================
<br>#undef HAVE_GETRUSAGE
<br>#undef HAVE_GETTIMEOFDAY
<br>#define HAVE_TIMES 1
<br><br>$ htags --statistics
<br>period user[sec] system[sec] elapsed[sec] %CPU
<br>--------------------------------- --------- ----------- ------------ -----
<br>Time of making duplicate entries 2.020 0.190 1.550 142.6
<br>Time of making function index 0.260 0.040 0.280 107.1
<br>Time of making file index 0.020 0.030 0.040 125.0
<br>Time of making include file index 1.070 0.110 1.170 100.9
<br>Time of making hypertext 18.580 2.190 15.310 135.7
<br>The entire time 21.950 2.560 18.360 133.5
<br><br>======== Case 4 =============================================================
<br>#undef HAVE_GETRUSAGE
<br>#define HAVE_GETTIMEOFDAY 1
<br>#undef HAVE_TIMES
<br><br>$ htags --statistics
<br>period elapsed[sec]
<br>--------------------------------- ------------
<br>Time of making duplicate entries 1.405
<br>Time of making function index 0.294
<br>Time of making file index 0.042
<br>Time of making include file index 1.157
<br>Time of making hypertext 15.555
<br>The entire time 18.461
<br><br>======== Case 5 =============================================================
<br>#undef HAVE_GETRUSAGE
<br>#undef HAVE_GETTIMEOFDAY
<br>#undef HAVE_TIMES
<br><br>$ htags --statistics
<br>period elapsed[sec]
<br>--------------------------------- ------------
<br>Time of making duplicate entries 2.000
<br>Time of making function index 0.000
<br>Time of making file index 0.000
<br>Time of making include file index 1.000
<br>Time of making hypertext 15.000
<br>The entire time 18.000
<br><br><br>Diffstat:
<br> configure.ac | 2
<br> htags/htags.c | 50 +----
<br> libutil/Makefile.am | 4
<br> libutil/global.h | 1
<br> libutil/statistics.c | 493 +++++++++++++++++++++++++++++++++++++++++++++++++++
<br> libutil/statistics.h | 84 ++++++++
<br> libutil/strbuf.c | 15 +
<br> libutil/strbuf.h | 4
<br> 8 files changed, 616 insertions(+), 37 deletions(-)
<br><br>----
<br>Hideki IWAMOTO <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26157218&i=0" target="_top" rel="nofollow">h-iwamoto@...</a>
<br><br /> <br />_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26157218&i=1" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><div class="small"><br/><img src="http://old.nabble.com/images/icon_attachment.gif" > <strong>20091102-statistics-5.patch</strong> (34K) <a href="http://old.nabble.com/attachment/26157218/0/20091102-statistics-5.patch" target="_top">Download Attachment</a></div><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>
tag:old.nabble.com,2006:post-26035473
Re: [PATCH] Add 2-pass mode to gtags
2009-10-23T18:45:17Z
2009-10-23T18:45:17Z
Hideki IWAMOTO
Hi.
<br><br>> Judging from a little seeing, it seems that new parser is more consistent.
<br>> Please teach if there is something problem.
<br><br>If you prefer consistency than keeping specifications of parser, there is no problem.
<br><br>I will commit the patch to CVS repository, if problem is not found by the next weekend,
<br><br>On Fri, 23 Oct 2009 09:51:37 +0900, Shigio YAMAGUCHI wrote...
<br>> Hi
<br>> > I know that incompatibility exists, and I am not expecting for 2-pass mode be enabled by default.
<br>>
<br>> Judging from a little seeing, it seems that new parser is more consistent.
<br>> Please teach if there is something problem.
<br>> --
<br>> Shigio YAMAGUCHI <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26035473&i=0" target="_top" rel="nofollow">shigio@...</a>>
<br>> PGP fingerprint: D1CB 0B89 B346 4AB6 5663 C4B6 3CA5 BBB3 57BE DDA3
<br><br>----
<br>Hideki IWAMOTO <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26035473&i=1" target="_top" rel="nofollow">h-iwamoto@...</a>
<br><br><br>_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26035473&i=2" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>
tag:old.nabble.com,2006:post-26023209
Re: patch: global -u on win32 with src root at drive root
2009-10-23T02:20:13Z
2009-10-23T02:20:13Z
Shigio YAMAGUCHI-2
Hi
<br>> Since, as a side-effect of trying to fix this problem, i now have win32
<br>> binaries for the current global release - would you be interested to get
<br>> a tar/zip package which can be provided for public download?
<br><br>No, thank you.
<br>Since GLOBAL doesn't support Windows environment, it is not scheduled
<br>to be distributed. Please see Q1 in the FAQ file.
<br>However, if you do it, I would like to introduce your site in the GLOBAL's site.
<br>Anyway, thank you for your offer.
<br>--
<br>Shigio YAMAGUCHI <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26023209&i=0" target="_top" rel="nofollow">shigio@...</a>>
<br>PGP fingerprint: D1CB 0B89 B346 4AB6 5663 C4B6 3CA5 BBB3 57BE DDA3
<br><br><br>_______________________________________________
<br>Help-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26023209&i=1" target="_top" rel="nofollow">Help-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/help-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/help-global</a><br><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Help-f1650.html" embed="fixTarget[1650]" target="_top" >Gnu - Global - Help</a></p>
tag:old.nabble.com,2006:post-26023040
Re: patch: global -u on win32 with src root at drive root
2009-10-23T00:37:14Z
2009-10-23T00:37:14Z
Kai Tetzlaff-2
Shigio YAMAGUCHI wrote:
<br>> Hi Kai,
<br>>> thanks for looking into it. This slightly modified version is working
<br>>> for win32:
<br>>
<br>> Committed. Thank you for your cooperation.
<br><br>Thanks Shigio for taking the time to look into this.
<br><br>Since, as a side-effect of trying to fix this problem, i now have win32
<br>binaries for the current global release - would you be interested to get
<br>a tar/zip package which can be provided for public download?
<br><br>BR,
<br>Kai
<br><br><br><br>_______________________________________________
<br>Help-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26023040&i=0" target="_top" rel="nofollow">Help-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/help-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/help-global</a><br><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Help-f1650.html" embed="fixTarget[1650]" target="_top" >Gnu - Global - Help</a></p>
tag:old.nabble.com,2006:post-26019320
Re: [PATCH] Add 2-pass mode to gtags
2009-10-22T17:41:20Z
2009-10-22T17:41:20Z
Shigio YAMAGUCHI-2
Hi
<br>> I know that incompatibility exists, and I am not expecting for 2-pass mode be enabled by default.
<br><br>Judging from a little seeing, it seems that new parser is more consistent.
<br>Please teach if there is something problem.
<br>--
<br>Shigio YAMAGUCHI <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26019320&i=0" target="_top" rel="nofollow">shigio@...</a>>
<br>PGP fingerprint: D1CB 0B89 B346 4AB6 5663 C4B6 3CA5 BBB3 57BE DDA3
<br><br><br>_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26019320&i=1" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>
tag:old.nabble.com,2006:post-26008398
Re: [PATCH] Add 2-pass mode to gtags
2009-10-22T04:35:57Z
2009-10-22T04:35:57Z
Hideki IWAMOTO
Hi.
<br>> Since there are a lot of changes in the parser, the output should be seen by
<br>> a lot of people before the formal release, I think.
<br>> How about releasing twice like follows?
<br><br>OK.
<br>I know that incompatibility exists, and I am not expecting for 2-pass mode be enabled by default.
<br><br>On Thu, 22 Oct 2009 15:44:28 +0900, Shigio YAMAGUCHI wrote...
<div class='shrinkable-quote'><br>> Hi Hideki,
<br>>
<br>> > To improve the performance of gtags, this patch adds 2-pass mode.
<br>> >
<br>> > The following three points are improved by moving use of function defined()
<br>> > from gtags-parser to gtags.
<br>> > 1. Decreases the frequency of parsing source file from three times to two times.
<br>> > 2. Decreases reading of GTAGS by keeping cache of GTAGS
<br>> > beyond the duration of parser process.
<br>> > 3. Increases parallelism of gtags-parser and gtags on multiprocessor system.
<br>>
<br>> Great!!! I named your code 'Giant steps'.
<br>>
<br>> Since there are a lot of changes in the parser, the output should be seen by
<br>> a lot of people before the formal release, I think.
<br>> How about releasing twice like follows?
<br>>
<br>> include Giant steps? enable Giant steps by default?
<br>> ------------------------------------------------------------------------------
<br>> GLOBAL-5.7.7(beta) Yes No
<br>> GLOBAL-5.8(formal) Yes Yes
<br>>
<br>> Thank you for your wonderful 'Giant steps'!
<br>> --
<br>> Shigio YAMAGUCHI <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26008398&i=0" target="_top" rel="nofollow">shigio@...</a>>
<br>> PGP fingerprint: D1CB 0B89 B346 4AB6 5663 C4B6 3CA5 BBB3 57BE DDA3
</div><br>----
<br>Hideki IWAMOTO <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26008398&i=1" target="_top" rel="nofollow">h-iwamoto@...</a>
<br><br><br>_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26008398&i=2" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>
tag:old.nabble.com,2006:post-26004771
Re: [PATCH] Add 2-pass mode to gtags
2009-10-21T23:36:13Z
2009-10-21T23:36:13Z
Shigio YAMAGUCHI-2
Hi Hideki,
<br><br>> To improve the performance of gtags, this patch adds 2-pass mode.
<br>>
<br>> The following three points are improved by moving use of function defined()
<br>> from gtags-parser to gtags.
<br>> 1. Decreases the frequency of parsing source file from three times to two times.
<br>> 2. Decreases reading of GTAGS by keeping cache of GTAGS
<br>> beyond the duration of parser process.
<br>> 3. Increases parallelism of gtags-parser and gtags on multiprocessor system.
<br><br>Great!!! I named your code 'Giant steps'.
<br><br>Since there are a lot of changes in the parser, the output should be seen by
<br>a lot of people before the formal release, I think.
<br>How about releasing twice like follows?
<br><br> include Giant steps? enable Giant steps by default?
<br>------------------------------------------------------------------------------
<br>GLOBAL-5.7.7(beta) Yes No
<br>GLOBAL-5.8(formal) Yes Yes
<br><br>Thank you for your wonderful 'Giant steps'!
<br>--
<br>Shigio YAMAGUCHI <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26004771&i=0" target="_top" rel="nofollow">shigio@...</a>>
<br>PGP fingerprint: D1CB 0B89 B346 4AB6 5663 C4B6 3CA5 BBB3 57BE DDA3
<br><br><br>_______________________________________________
<br>Bug-global mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26004771&i=1" target="_top" rel="nofollow">Bug-global@...</a>
<br><a href="http://lists.gnu.org/mailman/listinfo/bug-global" target="_top" rel="nofollow">http://lists.gnu.org/mailman/listinfo/bug-global</a><br><p>From forum: <a href="http://old.nabble.com/Gnu---Global---Bugs-f1649.html" embed="fixTarget[1649]" target="_top" >Gnu - Global - Bugs</a></p>