kfogel@collab.net wrote:
>Okay, that brings us to the present. The two tests still failing are:
>
> diff_tests.py 4: diff_replace_a_file()
> update_tests.py 3: update_ignores_added()
>
>Although they are failing for "the same" reason, in a sense, the
>failures are very different. I'll address each one separately.
>
>The 'diff_replace_a_file' failure is actually because the #1553 C-code
>changes *partially fix* a bug in our diff output. The bug is
>described in detail in new issue #2261. Unfortunately, the test is
>currently written to expect the bogus output shown in #2261. In other
>words, the test currently succeeds when it really shouldn't.
>
>Solution: we should decide how we want the diff output to look, change
>the test to expect that (this may be complex, due to the highly
>abstracted way diff_tests.py is written), and run it XFail until the
>bug is fully fixed. Fixing the diff-a-replaced-file bug is rather
>beyond the scope of #1553; it's an accident that the #1553 patch made
>it change in the right direction at all, really.
>
>
I think XFAILing the test is O.K.
>The 'update_ignores_added' failure is a manifestation of a deeper
>question: what revision number should status report for a
>schedule-replace file? With current Subversion, here's what happens:
>
> $ svn status -vuq README
> 13832 12812 kfogel README
> Status against revision: 13844
> $ svn rm README
> D README
> $ echo "totally new file" > README
> $ svn add README
> A README
> $ svn st -vuq README
> R 13832 12812 kfogel README
> Status against revision: 13844
> $
>
>With the latest #1553 patch applied, it does this instead:
>
> $ svn status -vuq README
> 13832 12812 kfogel README
> Status against revision: 13844
> $ svn rm README
> D README
> $ echo "totally new file" > README
> $ svn add README
> A README
> $ svn st -vuq README
> R - 12812 kfogel README
> Status against revision: 13844
> $
>
>It's not clear to me what the right answer is. One the one hand, the
>user asked for the status of 'README' as it currently stands, which is
>a new schedule-add file, so "-" for a working revision seems right.
>On the other hand, it's not like the user had any *other* way to refer
>to the old, now-shadowed 'README'. And what's with the 12812?
>
>
I think the output is exactly right. The status line is
<schedule> <current rev> <prev rev> <prev author> <name>
When <schedule> is R(eplace), I find it most logical that current-rev
would be "-" (after all, it's not been committed yet), and that
prev--rev and prev-author show information about the item that's being
replaced.
-- Brane
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sat Apr 2 01:09:21 2005