You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Branko Čibej <br...@xbc.nu> on 2002/07/06 18:36:36 UTC
Re: svn commit: rev 2418 - branches/issue-531-dev/subversion/libsvn_fs
branches/issue-531-dev/subversion/libsvn_delta
Well, it looks like the combiner is now properly integrated into the
filesystem on branches/issue-531-dev, and most tests pass (except
fs-test 27, have to figure that one out). It needs a lot more pounding
on before we merge it onto the trunk, though; and I have some ideas
about how to reduce memcpys in the undeltification pass.
I'll be away next week; much obliged if people could bang on this code a
bit and lt me know the results.
Oh, and the 2417 and 2419 commits can/should be merged to the mainline
when convenient; they're just test harness improvements.
brane@tigris.org wrote:
>Author: brane
>Date: Sat, 06 Jul 2002 11:46:52 -0500
>New Revision: 2418
>
>Modified:
> branches/issue-531-dev/subversion/libsvn_delta/compose_delta.c
> branches/issue-531-dev/subversion/libsvn_delta/delta.h
> branches/issue-531-dev/subversion/libsvn_fs/reps-strings.c
>Log:
>Working on issue #531.
>
>* subversion/libsvn_fs/reps-strings.c (window_handler, window_handler_baton):
>Copy the window into a known pool, so that it survives the svndiff parser.
>(get_one_window): Stricter check for cur_chunk overflow. Fix assertions.
>(rep_undeltify_range): Move A and composite pool and window decls inside loop.
>And don't destroy the pool that contains the window we just got.
>
>* subversion/libsvn_delta/compose_delta.c (svn_txdelta__compose_windows)
>Extreact the window-copying code into new funceion svn_txdelta__copy_window.
>* subversion/libsvn_delta/delta.h (svn_txdelta__copy_window): New prototype.
>
>
--
Brane Čibej <br...@xbc.nu> http://www.xbc.nu/brane/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: svn commit: rev 2418 - branches/issue-531-dev/subversion/libsvn_fs
branches/issue-531-dev/subversion/libsvn_delta
Posted by Branko Čibej <br...@xbc.nu>.
Branko Čibej wrote:
> Branko Čibej wrote:
>
>> Well, it looks like the combiner is now properly integrated into the
>> filesystem on branches/issue-531-dev, and most tests pass (except
>> fs-test 27, have to figure that one out).
>
>
>
> By the way, in case I don't manage to fix this before leaving (should
> be on the road right now ...) and somebody cares to take this on: This
> test passes if SVN_STREAM_CHUNK_SIZE is 512, but fails if it's the
> default 102400.
Interesting. The digest comparison fails sometimes, but succeeds later
on; e.g., the first failure is at revision 24, but then revs 23..17
succeed. Most interesting, because that suggests the data is undeltified
correctly, but the digests are somehow wrong.
--
Brane Čibej <br...@xbc.nu> http://www.xbc.nu/brane/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: svn commit: rev 2418 - branches/issue-531-dev/subversion/libsvn_fs
branches/issue-531-dev/subversion/libsvn_delta
Posted by Branko Čibej <br...@xbc.nu>.
Branko Čibej wrote:
> Well, it looks like the combiner is now properly integrated into the
> filesystem on branches/issue-531-dev, and most tests pass (except
> fs-test 27, have to figure that one out).
By the way, in case I don't manage to fix this before leaving (should be
on the road right now ...) and somebody cares to take this on: This test
passes if SVN_STREAM_CHUNK_SIZE is 512, but fails if it's the default
102400.
--
Brane Čibej <br...@xbc.nu> http://www.xbc.nu/brane/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org