On Tue, 12 Jun 2007, Blair Zajac wrote:
> In looking through the 1.4.4 changes listed in the STATUS, it appears
> that these revisions are not accurate:
>
> SWIG/python bindings:
> - fixed: memory leak whenever C APIs returned errors (r21453, r23468)
>
> as they refer to
>
> $ svn log -r 21453
> ------------------------------------------------------------------------
> r21453 | pburba | 2006-09-12 07:43:41 -0700 (Tue, 12 Sep 2006) | 28 lines
>
> Consolidate the many conflict file singleton handlers into one within
> tree.py.
>
>
> $ svn log -r 23468
> ------------------------------------------------------------------------
> r23468 | lundblad | 2007-02-22 13:42:19 -0800 (Thu, 22 Feb 2007) | 12 lines
>
> Fix regression introduced in r21720: update of a replaced file *without*
> history uses the revert-base. Only replaced files *with* history have a
> revert-base.
>
>
>
> It appears that the listed revision should be
>
> r23544 | dlr | 2007-03-01 14:29:11 -0800 (Thu, 01 Mar 2007) | 6 lines
> Changed paths:
> M /branches/1.4.x
> M /branches/1.4.x/STATUS
> M
> /branches/1.4.x/subversion/bindings/swig/python/libsvn_swig_py/swigutil_py.c
> M
> /branches/1.4.x/subversion/bindings/swig/python/libsvn_swig_py/swigutil_py.h
>
> Merge r23521 from trunk into the 1.4.x branch, fixing an error leak
> the Python bindings that occurred when an error was returned from the
> C API, translated into a Python exception, but then not cleared.
>
>
> And that (r21453, r23468) refer to issue #2618.
>
> If nobody disagrees, I'll make the changes in the trunk and 1.4.x file.
+1, nice catch Blair.
I'm not sure how r23468 fixes an error leak -- were you planning on
listing this change separately, or was it not backported to the 1.4.x
line?
- Dan
- application/pgp-signature attachment: stored
Received on Wed Jun 13 17:34:59 2007