You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@subversion.apache.org by st...@apache.org on 2011/06/22 17:50:54 UTC
svn commit: r1138516 - /subversion/trunk/subversion/libsvn_client/ra.c
Author: stsp
Date: Wed Jun 22 15:50:54 2011
New Revision: 1138516
URL: http://svn.apache.org/viewvc?rev=1138516&view=rev
Log:
* subversion/libsvn_client/ra.c
(svn_client__repos_locations): PEG_REVNUM used to be initialised from
copyfrom info (if any) before r1092150. Since this revision we've
accidentally left PEG_REVNUM initialised to SVN_INVALID_REVNUM all
the time (Bert confirmed that this change was accidental).
This didn't cause any test failures, and neither does this change.
But we're now back to the original semantics.
Modified:
subversion/trunk/subversion/libsvn_client/ra.c
Modified: subversion/trunk/subversion/libsvn_client/ra.c
URL: http://svn.apache.org/viewvc/subversion/trunk/subversion/libsvn_client/ra.c?rev=1138516&r1=1138515&r2=1138516&view=diff
==============================================================================
--- subversion/trunk/subversion/libsvn_client/ra.c (original)
+++ subversion/trunk/subversion/libsvn_client/ra.c Wed Jun 22 15:50:54 2011
@@ -621,7 +621,7 @@ svn_client__repos_locations(const char *
const char *repos_relpath;
svn_boolean_t is_copy;
- SVN_ERR(svn_wc__node_get_origin(&is_copy, NULL, &repos_relpath,
+ SVN_ERR(svn_wc__node_get_origin(&is_copy, &peg_revnum, &repos_relpath,
&repos_root_url, NULL,
ctx->wc_ctx, local_abspath_or_url,
FALSE, subpool, subpool));
Re: svn commit: r1138516 -
/subversion/trunk/subversion/libsvn_client/ra.c
Posted by Daniel Shahaf <d....@daniel.shahaf.name>.
Stefan Sperling wrote on Wed, Jun 22, 2011 at 22:53:13 +0200:
> On Wed, Jun 22, 2011 at 04:43:38PM -0400, Greg Stein wrote:
> > Is there a way that this can be tested? ie. avoid similar situations
> > in the future?
>
> I don't know. I came across this because I saw code that did:
>
> peg_rev = INVALID_REVNUM;
> /* do nothing with peg_rev */
> if (peg_rev == INVALIV_REVNUM)
>
> Which is obviously bogus so I praised the file at various revisions
> until I found the cause. I don't know more than that :)
>
> But I agree our tests should have caught this if it is important.
I'm fine with letting tools (compilers and static analysis) catch this
sort of stuff. Though if there was a library-consumer-layer-visible bug
here, agree that our tests might want to be extended too.
Re: svn commit: r1138516 -
/subversion/trunk/subversion/libsvn_client/ra.c
Posted by Stefan Sperling <st...@elego.de>.
On Wed, Jun 22, 2011 at 04:43:38PM -0400, Greg Stein wrote:
> Is there a way that this can be tested? ie. avoid similar situations
> in the future?
I don't know. I came across this because I saw code that did:
peg_rev = INVALID_REVNUM;
/* do nothing with peg_rev */
if (peg_rev == INVALIV_REVNUM)
Which is obviously bogus so I praised the file at various revisions
until I found the cause. I don't know more than that :)
But I agree our tests should have caught this if it is important.
Re: svn commit: r1138516 - /subversion/trunk/subversion/libsvn_client/ra.c
Posted by Greg Stein <gs...@gmail.com>.
Is there a way that this can be tested? ie. avoid similar situations
in the future?
On Wed, Jun 22, 2011 at 11:50, <st...@apache.org> wrote:
> Author: stsp
> Date: Wed Jun 22 15:50:54 2011
> New Revision: 1138516
>
> URL: http://svn.apache.org/viewvc?rev=1138516&view=rev
> Log:
> * subversion/libsvn_client/ra.c
> (svn_client__repos_locations): PEG_REVNUM used to be initialised from
> copyfrom info (if any) before r1092150. Since this revision we've
> accidentally left PEG_REVNUM initialised to SVN_INVALID_REVNUM all
> the time (Bert confirmed that this change was accidental).
> This didn't cause any test failures, and neither does this change.
> But we're now back to the original semantics.
>...