Max, just a ping regarding a cvs2svn issue I mentioned in a private
mail to you a while back. This may be the first time the dev list has
seen this (although anyone watching the verification/ directory would
have known about it already).
Do you know anything more about this problem than I did when I
committed the notes below (in rev 6665)?
-Karl
> There's one unfiled but important issue that I know of (referred to in
> issue #1504, maybe you saw the reference). The directory
>
> tools/cvs2svn/test-data/verification/
>
> contains in-progress data that I was using to map out how the result
> of converting test-data/main-cvsrepos to Subversion should look. The
> reason I was doing this was so I could make run-tests.sh expect the
> correct results -- obviously, we can't depend on cvs2svn.py to do it,
> since that's what we're testing! :-) And as it happens, I found a
> correctness issue as I was verifying. It's described in
>
> tools/cvs2svn/test-data/verification/commits
>
> This paragraph describes the bug:
>
> [Okay, problem: This revision creates branch B_MIXED, by copying
> revision 19 of trunk. But it fails to recopy 'proj/sub3/default'
> and 'proj/sub2/subsubA/default' from some revision earlier than
> 19, which it needs to do because in those files, B_MIXED sprouts
> from 1.2 and 1.1 respectively, not 1.3 and 1.2 (the latter being
> what you get if you just copy from revision 19). This means that
> the contents of those files are wrong in branches/B_MIXED/'s first
> incarnation. Even if they get fixed up in some later revision,
> there is still a period of wrongness.]
>
> I stopped the verification process there; there was no point
> proceeding until I'd fixed that bug, and I haven't had time to do
> that.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Mon Sep 29 17:39:27 2003