[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: svn commit: r33938 - in trunk: . notes/tree-conflicts subversion/include subversion/include/private subversion/libsvn_auth_kwallet subversion/libsvn_client subversion/libsvn_subr subversion/libsvn_wc subversion/svn subversion/tests/cmdline subversion/tests/cmdline/svntest subversion/tests/libsvn_subr tools/buildbot/slaves/win32-xp-VS2005 tools/server-side www

From: Neels J. Hofmeyr <neels_at_elego.de>
Date: Wed, 29 Oct 2008 02:18:00 +0100

sbutler_at_tigris.org wrote:
> Author: sbutler
> Date: Tue Oct 28 09:22:25 2008
> New Revision: 33938
>
> Log:
> Merge the tree-conflicts-notify branch to trunk.

This

> Merge uses a different set of callbacks, which we haven't touched yet,
> but calls the same notify() function. To be fixed separately.

is related to this:

> Modified: trunk/subversion/tests/cmdline/merge_tests.py
> URL: http://svn.collab.net/viewvc/svn/trunk/subversion/tests/cmdline/merge_tests.py?pathrev=33938&r1=33937&r2=33938
> ==============================================================================
> --- trunk/subversion/tests/cmdline/merge_tests.py Tue Oct 28 08:56:45 2008 (r33937)
> +++ trunk/subversion/tests/cmdline/merge_tests.py Tue Oct 28 09:22:25 2008 (r33938)
> @@ -12256,7 +12256,7 @@ def del_differing_file(sbox):
> "--- Merging r2 into '%s':\n" % dir_G2,
> "C %s\n" % dir_G2,
> "Summary of conflicts:\n",
> - " Tree conflicts: 1\n",
> + " Text conflicts: 1\n",
> " Skipped paths: 1\n"])
>
> svn_merge(s_rev_pi, source, target, [
> @@ -12264,7 +12264,7 @@ def del_differing_file(sbox):
> "--- Merging r3 into '%s':\n" % dir_G2,
> "C %s\n" % dir_G2,
> "Summary of conflicts:\n",
> - " Tree conflicts: 1\n",
> + " Text conflicts: 1\n",
> " Skipped paths: 1\n"])
>
>
> @@ -12287,7 +12287,7 @@ def del_differing_file(sbox):
> "--- Merging r2 into '%s':\n" % dir_G3,
> "C %s\n" % dir_G3,
> "Summary of conflicts:\n",
> - " Tree conflicts: 1\n",
> + " Text conflicts: 1\n",
> " Skipped paths: 1\n"])
>
> svn_merge(s_rev_pi, source, target, [
> @@ -12295,7 +12295,7 @@ def del_differing_file(sbox):
> "--- Merging r3 into '%s':\n" % dir_G3,
> "C %s\n" % dir_G3,
> "Summary of conflicts:\n",
> - " Tree conflicts: 1\n",
> + " Text conflicts: 1\n",
> " Skipped paths: 1\n"])
>
> os.chdir(saved_cwd)

This change feels very wrong to me. I'd rather XFAIL these tests until merge
notification is fixed. It says "Text conflicts", but these are tree
conflicts!! Furthermore, some tests I wrote about yesterday are inherently
depending on victims' parent dirs being in conflict, so it's a complete
delusion to pass them while taking away this behaviour.

Maybe it's a little obsessive, but it just feels so wrong to PASS this
behaviour. We could go back to counting text conflicts on directories as
tree-conflicts. But we'd use the time better changing merge notification.

Oh, I guess I've volunteered. ;)

So about the strategy. Someone said along the way it would be better to use
an editor instead of the diff callbacks. But we're not going into this now,
are we? I guess it'd take a lot of time. Let's enhance the diff_callbacks
for now. Do you think this is the right decision?

~Neels

Received on 2008-10-29 02:18:49 CET

This is an archived mail posted to the Subversion Dev mailing list.