On Sun, 2008-10-05 at 04:17 +0200, Neels J Hofmeyr wrote:
> > Karl Fogel wrote:
> >> "Neels J. Hofmeyr" <neels_at_elego.de> writes:
> >>> Let me rephrase that:
> >>> Is it currently possible to do a diff between
> >>> URL_at_REV
> >>> and
> >>> an unrelated working copy with local mods
> >>> , and if yes, how?
> >>>  unrelated, as in not of the same URL.
> >> $ cd svn_trunk_working_copy
> >> $ echo "This is a local mod." >> README
> >> $ svn diff --old=http://svn.collab.net/repos/svn/trunk/INSTALL@33019 --new=README
> >> -Karl
> > Ah, thanks for that. I was quite unlucky (cough) to have missed that
> > combination.
> > Alas, it looks like julianf and me found a case where `svn diff' reports
> > wrongly. (It is a crucial diff case for implementing directory comparison
> > for tree-conflicts.)
> > I've attached test scripts and an output screenshot when using current trunk.
> > "Create two similar trees, but leave some of the changes uncommitted in
> > the second tree. Then diff the url of the first to the working copy of
> > the second tree."
> > When reading the "Results" in the output, note that
> > 2) the other local change (the file addition of `baz') is sort of not seen
> > by diff and reported as if baz was missing in `--new' (wrong).
Sadly, there are several cases of "svn diff" that are just broken. This
appears to be one of them.
> > 3) Also, the diff is sort of ambiguous about reporting the file name:
> > Which is it, quux/baz or foo/baz?
> > "--- quux/baz (.../file:///arch/elego/svn/test/repos/foo/baz) (revision 2)"
Yes, it's broken there too, though of course these header lines are
unimportant for our purposes.
> > Is this really an error or am I still missing something?
It's really an error.
Can you maybe add a test for it, and/or file an issue, and/or fix it, or
just ignore this case, while still writing the directory compare thing?
To unsubscribe, e-mail: dev-unsubscribe_at_subversion.tigris.org
For additional commands, e-mail: dev-help_at_subversion.tigris.org
Received on 2008-10-07 23:43:33 CEST