Re: cost of popen()

[Hideki IWAMOTO]

On Sat, 02 Jul 2005 13:18:53 +0900, Shigio Yamaguchi wrote...
> It seems to be unrelated with the performance tuning.
> Would you please change the program without user visible change?
> Let's talk separately about the specification change with other subject.

OK. This is a revised version.

 - The verbose message was returned to the original form.
 - Assembly source is skipped by the original method.
 - Also when using PATHINDEX format, a parser processes multiple files.

---
diffstat

 global/global.c   |   21 ++++------
 gtags/gtags.c     |   97 +++++++++++++++++++++++++++++++++++++++++-------
 libutil/gtagsop.c |  107 +++++++++++++++++++++++++++++++++++++-----------------
 libutil/gtagsop.h |    4 +-
 4 files changed, 167 insertions(+), 62 deletions(-)

--- 
Comparison of execution time

$ tar xfz ~/download/bison/bison-2.0.tar.gz
$ cd bison-2.0

======= current cvs version =======
$ time gtags
0.516u 0.797s 0:01.97 65.9%     0+0k 0+0io 0pf+0w
$ global -P | xargs touch
$ time gtags -i
0.549u 0.813s 0:02.02 66.8%     0+0k 0+0io 0pf+0w
$ time gtags -cc
0.625u 0.992s 0:02.10 76.6%     0+0k 0+0io 0pf+0w
$ global -P | xargs touch
$ time gtags -i
0.641u 1.000s 0:02.15 76.2%     0+0k 0+0io 0pf+0w

======= patched version =======
$ time gtags
0.426u 0.401s 0:01.55 52.9%     0+0k 0+0io 0pf+0w
$ global -P | xargs touch
$ time gtags -i
0.451u 0.409s 0:01.56 54.4%     0+0k 0+0io 0pf+0w
$ time gtags -cc
0.493u 0.458s 0:01.53 61.4%     0+0k 0+0io 0pf+0w
$ global -P | xargs touch
$ time gtags -i
0.509u 0.462s 0:01.51 63.5%     0+0k 0+0io 0pf+0w

----
Hideki IWAMOTO  address@hidden

gtags-add3.patch
Description: Binary data