You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Ben Collins-Sussman <su...@collab.net> on 2004/05/14 21:23:00 UTC
Re: svn commit: r9714 - trunk/subversion/libsvn_wc
On Fri, 2004-05-14 at 16:09, bliss@tigris.org wrote:
> * subversion/libsvn_wc/adm_crawler.c
> (report_revisions): Call svn_io_check_path for each entry instead of
> calling svn_io_get_dirents and comparing strings.
Um... isn't this going to cause a *huge* slowdown when the 'svn commit'
scans for modified files?
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: svn commit: r9714 - trunk/subversion/libsvn_wc
Posted by Ben Collins-Sussman <su...@collab.net>.
On Fri, 2004-05-14 at 16:49, Benjamin Pflugmann wrote:
> I didn't too deep into it, but it looks like one could keep the
> previous logic and only add the svn_io_check_path call if it "looks"
> like the item is missing in order to be sure the file system "thinks"
> the same.
Tobias: whatever the solution ends up being, be sure to nominate it for
1.0.3 backport! This is a serious bug.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: svn commit: r9714 - trunk/subversion/libsvn_wc
Posted by Benjamin Pflugmann <be...@pflugmann.de>.
On Fri 2004-05-14 at 23:40:49 +0200, Tobias Ringström wrote:
> Ben Collins-Sussman wrote:
> >On Fri, 2004-05-14 at 16:09, bliss@tigris.org wrote:
> >
> >>* subversion/libsvn_wc/adm_crawler.c
> >> (report_revisions): Call svn_io_check_path for each entry instead of
> >> calling svn_io_get_dirents and comparing strings.
> >
> >Um... isn't this going to cause a *huge* slowdown when the 'svn commit'
> >scans for modified files?
I wondered the same and was just about to write an email...
> I thought that svn_io_get_dirents already called stat to get the type
> and that it would be the same amount of work. On a closer look (and to
> my big surprise) there seems to be support for a d_type entry in the
> dirent on some systems to avoid having to call stat. I'll continue
> investigating and find another solution if neccessary.
I didn't too deep into it, but it looks like one could keep the
previous logic and only add the svn_io_check_path call if it "looks"
like the item is missing in order to be sure the file system "thinks"
the same.
Bye,
Benjamin.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: svn commit: r9714 - trunk/subversion/libsvn_wc
Posted by Tobias Ringström <to...@ringstrom.mine.nu>.
Ben Collins-Sussman wrote:
> On Fri, 2004-05-14 at 16:09, bliss@tigris.org wrote:
>
>>* subversion/libsvn_wc/adm_crawler.c
>> (report_revisions): Call svn_io_check_path for each entry instead of
>> calling svn_io_get_dirents and comparing strings.
>
> Um... isn't this going to cause a *huge* slowdown when the 'svn commit'
> scans for modified files?
I thought that svn_io_get_dirents already called stat to get the type
and that it would be the same amount of work. On a closer look (and to
my big surprise) there seems to be support for a d_type entry in the
dirent on some systems to avoid having to call stat. I'll continue
investigating and find another solution if neccessary.
/Tobias
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org