You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Bert Huijben <be...@qqmail.nl> on 2012/05/13 23:26:05 UTC
RE: svn commit: r1337781 - /subversion/trunk/subversion/libsvn_ra_serf/getlocationsegments.c
> -----Original Message-----
> From: gstein@apache.org [mailto:gstein@apache.org]
> Sent: zondag 13 mei 2012 7:33
> To: commits@subversion.apache.org
> Subject: svn commit: r1337781 -
> /subversion/trunk/subversion/libsvn_ra_serf/getlocationsegments.c
>
> Author: gstein
> Date: Sun May 13 05:33:01 2012
> New Revision: 1337781
>
> URL: http://svn.apache.org/viewvc?rev=1337781&view=rev
> Log:
> Switch get-location-segments over to v2 xml parsing.
>
> * subversion/libsvn_ra_serf/getlocationsegments.c:
> (gls_context_t): remove SUBPOOL, INSIDE_REPORT, DONE
> (INITIAL, REPORT, SEGMENT): add enum for parse states
> (gls_ttable): new transition table
> (start_gls, end_gls): removed; obsolete.
> (gls_closed): new close handler. call the receiver with the segment
> information.
> (svn_ra_serf__get_location_segments): drop the old parsing stuff.
> add the new v2 parsing. use run_one(). remove the detection of
> ending within the report element.
The serf/Win32 buildbot currently reports two test failures:
FAIL: basic_tests.py 32: info on file not existing in HEAD
FAIL: log_tests.py 33: log with unrelated peg and operative rev targets
Both appear to be caused by a leaked error generated from a server report via util.c line 902 (called from svn_ra_serf__expect_empty_body()) for svn_ra_serf__get_locations() calls.
The serf call currently returns success, while it probably should have reported this error.
(You might have fixed an old error as I remember that there used to be some behavior difference between ra implementations here)
[I haven't tracked this issue to this specific revision. But this appears to be the most relevant in the batch]
Bert
RE: svn commit: r1337781 - /subversion/trunk/subversion/libsvn_ra_serf/getlocationsegments.c
Posted by Greg Stein <gs...@gmail.com>.
Thanks, Bert. I'll take a look in a little while.
On May 13, 2012 11:26 AM, "Bert Huijben" <be...@qqmail.nl> wrote:
>
>
> > -----Original Message-----
> > From: gstein@apache.org [mailto:gstein@apache.org]
> > Sent: zondag 13 mei 2012 7:33
> > To: commits@subversion.apache.org
> > Subject: svn commit: r1337781 -
> > /subversion/trunk/subversion/libsvn_ra_serf/getlocationsegments.c
> >
> > Author: gstein
> > Date: Sun May 13 05:33:01 2012
> > New Revision: 1337781
> >
> > URL: http://svn.apache.org/viewvc?rev=1337781&view=rev
> > Log:
> > Switch get-location-segments over to v2 xml parsing.
> >
> > * subversion/libsvn_ra_serf/getlocationsegments.c:
> > (gls_context_t): remove SUBPOOL, INSIDE_REPORT, DONE
> > (INITIAL, REPORT, SEGMENT): add enum for parse states
> > (gls_ttable): new transition table
> > (start_gls, end_gls): removed; obsolete.
> > (gls_closed): new close handler. call the receiver with the segment
> > information.
> > (svn_ra_serf__get_location_segments): drop the old parsing stuff.
> > add the new v2 parsing. use run_one(). remove the detection of
> > ending within the report element.
>
> The serf/Win32 buildbot currently reports two test failures:
> FAIL: basic_tests.py 32: info on file not existing in HEAD
> FAIL: log_tests.py 33: log with unrelated peg and operative rev targets
>
> Both appear to be caused by a leaked error generated from a server report
> via util.c line 902 (called from svn_ra_serf__expect_empty_body()) for
> svn_ra_serf__get_locations() calls.
>
> The serf call currently returns success, while it probably should have
> reported this error.
>
> (You might have fixed an old error as I remember that there used to be
> some behavior difference between ra implementations here)
>
>
> [I haven't tracked this issue to this specific revision. But this appears
> to be the most relevant in the batch]
>
> Bert
>
>
Re: svn commit: r1337781 - /subversion/trunk/subversion/libsvn_ra_serf/getlocationsegments.c
Posted by Greg Stein <gs...@gmail.com>.
On Sun, May 13, 2012 at 5:26 PM, Bert Huijben <be...@qqmail.nl> wrote:
>
>
>> -----Original Message-----
>> From: gstein@apache.org [mailto:gstein@apache.org]
>> Sent: zondag 13 mei 2012 7:33
>> To: commits@subversion.apache.org
>> Subject: svn commit: r1337781 -
>> /subversion/trunk/subversion/libsvn_ra_serf/getlocationsegments.c
>>
>> Author: gstein
>> Date: Sun May 13 05:33:01 2012
>> New Revision: 1337781
>>
>> URL: http://svn.apache.org/viewvc?rev=1337781&view=rev
>> Log:
>> Switch get-location-segments over to v2 xml parsing.
>>
>> * subversion/libsvn_ra_serf/getlocationsegments.c:
>> (gls_context_t): remove SUBPOOL, INSIDE_REPORT, DONE
>> (INITIAL, REPORT, SEGMENT): add enum for parse states
>> (gls_ttable): new transition table
>> (start_gls, end_gls): removed; obsolete.
>> (gls_closed): new close handler. call the receiver with the segment
>> information.
>> (svn_ra_serf__get_location_segments): drop the old parsing stuff.
>> add the new v2 parsing. use run_one(). remove the detection of
>> ending within the report element.
>
> The serf/Win32 buildbot currently reports two test failures:
> FAIL: basic_tests.py 32: info on file not existing in HEAD
> FAIL: log_tests.py 33: log with unrelated peg and operative rev targets
>
> Both appear to be caused by a leaked error generated from a server report via util.c line 902 (called from svn_ra_serf__expect_empty_body()) for svn_ra_serf__get_locations() calls.
>
> The serf call currently returns success, while it probably should have reported this error.
Yup. It was an error that came from the server. No client-side error
occurred, so success was returned. We now look for potential server
errors on most responses. (in process on getting all of them to
look...)
>
> (You might have fixed an old error as I remember that there used to be some behavior difference between ra implementations here)
Fixed in 1338128. Thanks!
Cheers,
-g