You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by kf...@collab.net on 2003/08/15 20:56:20 UTC
issue #1473
Shlomi, in
http://subversion.tigris.org/issues/show_bug.cgi?id=1473
I added this comment:
> Have you been able to determine what specific revision caused
> this bug? I'm pretty sure it appeared only recently. Ben said
> you suspected revision 6687; if you locally back out that change
> and and try to reproduce, does the bug still happen? Knowing
> what change introduced this would probably help a lot.
Don't know if you saw that or not... I heard from Ben that you were
debugging #1473, which is great, but it would also help if the bug's
appearance could be narrowed down to one revision. It would give us
another point to triangulate from when debugging.
I'm probably going to put #1473 in milestone 0.28, so eventually I'll
get around to reproducing it with 6687 locally reverted :-)... If
you're debugging it anyway, could you try that, and report the results
in the issue?
Thanks,
-Karl
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: issue #1473
Posted by Shlomi Fish <sh...@vipe.stud.technion.ac.il>.
On Sun, 16 Aug 2003 kfogel@collab.net wrote:
> Shlomi Fish <sh...@vipe.stud.technion.ac.il> writes:
> > The solution Ben and I came up to was to use a plain copy before the call
> > to svn_wc_add and then afterwards do the translation. This may mean
> > splitting the routine into two.
> >
> > Does anybody have a better idea?
> >
> > For more information, you can consult a log of the IRC session at channel
> > #svn, regarding this subject:
> >
> > http://vipe.technion.ac.il/~shlomif/svn-issue-1473-irc-log.txt
>
> Is this thread linked to from the issue?
>
It is now. ;-)
Regards,
Shlomi Fish
----------------------------------------------------------------------
Shlomi Fish shlomif@vipe.technion.ac.il
Home Page: http://t2.technion.ac.il/~shlomif/
An apple a day will keep a doctor away. Two apples a day will keep two
doctors away.
Falk Fish
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: issue #1473
Posted by kf...@collab.net.
Shlomi Fish <sh...@vipe.stud.technion.ac.il> writes:
> The solution Ben and I came up to was to use a plain copy before the call
> to svn_wc_add and then afterwards do the translation. This may mean
> splitting the routine into two.
>
> Does anybody have a better idea?
>
> For more information, you can consult a log of the IRC session at channel
> #svn, regarding this subject:
>
> http://vipe.technion.ac.il/~shlomif/svn-issue-1473-irc-log.txt
Is this thread linked to from the issue?
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: [PATCH] Fix for Issue #1473 [was Re: issue #1473]
Posted by kf...@collab.net.
Shlomi Fish <sh...@vipe.stud.technion.ac.il> writes:
> Here's a patch that fixes this problem. Here's the log:
> Resolves Issue #1437 (Repos->WC copy of a file with svn:keywords causes
> svn to crash)
Thanks, Shlomi! This patch is on the top of my stack for tomorrow
morning.
-Karl
> * include/svn_wc.h
> added the declaration of svn_wc_add_repos_file_post().
>
> * libsvn_wc/update_editor.c
> (svn_add_repos_file): Change the copy-and-translate command to a plain
> copy one. This is so, the entry won't be needed in order to process
> the keyword.
> Added svn_wc_add_repos_post: this function does the actual translation.
>
>
> * libsvn_client/copy.c
> (repos_to_wc_copy): Added a call to svn_wc_add_repos_file_post(),
> after svn_wc_add().
>
> * copy_test.py
> Added the repos_to_wc_copy_eol_keywords test.
>
> >>>
>
> Regards,
>
> Shlomi Fish
>
> ----------------------------------------------------------------------
> Shlomi Fish shlomif@vipe.technion.ac.il
> Home Page: http://t2.technion.ac.il/~shlomif/
>
> An apple a day will keep a doctor away. Two apples a day will keep two
> doctors away.
>
> Falk Fish
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
> For additional commands, e-mail: dev-help@subversion.tigris.org
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
[PATCH] Fix for Issue #1473 [was Re: issue #1473]
Posted by Shlomi Fish <sh...@vipe.stud.technion.ac.il>.
Here's a patch that fixes this problem. Here's the log:
<<<
Resolves Issue #1437 (Repos->WC copy of a file with svn:keywords causes
svn to crash)
* include/svn_wc.h
added the declaration of svn_wc_add_repos_file_post().
* libsvn_wc/update_editor.c
(svn_add_repos_file): Change the copy-and-translate command to a plain
copy one. This is so, the entry won't be needed in order to process
the keyword.
Added svn_wc_add_repos_post: this function does the actual translation.
* libsvn_client/copy.c
(repos_to_wc_copy): Added a call to svn_wc_add_repos_file_post(),
after svn_wc_add().
* copy_test.py
Added the repos_to_wc_copy_eol_keywords test.
>>>
Regards,
Shlomi Fish
----------------------------------------------------------------------
Shlomi Fish shlomif@vipe.technion.ac.il
Home Page: http://t2.technion.ac.il/~shlomif/
An apple a day will keep a doctor away. Two apples a day will keep two
doctors away.
Falk Fish
Re: issue #1473
Posted by Shlomi Fish <sh...@vipe.stud.technion.ac.il>.
Hi good people.
I talked with sussman on the IRC regarding this issue and here's what we
discovered so far:
1. svn does indeed segfault when doing a repos->Working Copy copy of a
file where the svn:keywords is set.
2. The reason is that svn_wc__get_keywords (a function that is called by
the translate routine) tries to access the entry in the .svn/entries file,
and since the entry was not created yet it returns NULL.
3. svn_wc_add() is responsible for creating the entry, but it has not
called yet when the translation takes place.
4. The translation takes place in svn_wc_add_repos_file(), a recently
added function that is responsible to instantize all the file data based
on the input from the repository. It is called right before svn_wc_add().
The solution Ben and I came up to was to use a plain copy before the call
to svn_wc_add and then afterwards do the translation. This may mean
splitting the routine into two.
Does anybody have a better idea?
For more information, you can consult a log of the IRC session at channel
#svn, regarding this subject:
http://vipe.technion.ac.il/~shlomif/svn-issue-1473-irc-log.txt
(I am rindolf)
Regards,
Shlomi Fish
----------------------------------------------------------------------
Shlomi Fish shlomif@vipe.technion.ac.il
Home Page: http://t2.technion.ac.il/~shlomif/
An apple a day will keep a doctor away. Two apples a day will keep two
doctors away.
Falk Fish
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: issue #1473
Posted by Shlomi Fish <sh...@vipe.stud.technion.ac.il>.
On Fri, 15 Aug 2003 kfogel@collab.net wrote:
> Shlomi, in
>
> http://subversion.tigris.org/issues/show_bug.cgi?id=1473
>
> I added this comment:
>
> > Have you been able to determine what specific revision caused
> > this bug? I'm pretty sure it appeared only recently. Ben said
> > you suspected revision 6687; if you locally back out that change
> > and and try to reproduce, does the bug still happen? Knowing
> > what change introduced this would probably help a lot.
>
> Don't know if you saw that or not...
I saw it - I received an E-mail with it.
> I heard from Ben that you were
> debugging #1473, which is great, but it would also help if the bug's
> appearance could be narrowed down to one revision. It would give us
> another point to triangulate from when debugging.
>
OK, in revision 6686, I don't get a segfault, but instead get:
<<<
subversion/libsvn_client/copy.c:732: (apr_err=160013)
svn: Filesystem has no item
svn: path
`///home/shlomi/Download/unpack/prog/SCM/svn-test/new-issue/copy-test/t1/repo/trunk/bar'
not found in revision `1'
>>>
> I'm probably going to put #1473 in milestone 0.28, so eventually I'll
> get around to reproducing it with 6687 locally reverted :-)... If
> you're debugging it anyway, could you try that, and report the results
> in the issue?
>
I'll report what I got there.
Regards,
Shlomi Fish
> Thanks,
> -Karl
>
----------------------------------------------------------------------
Shlomi Fish shlomif@vipe.technion.ac.il
Home Page: http://t2.technion.ac.il/~shlomif/
An apple a day will keep a doctor away. Two apples a day will keep two
doctors away.
Falk Fish
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org