>
> This would have been easier to understand if you had included an actual
> example, but I think I know what you mean. I believe Subversion behaves
> like this:
Sorry about the lack of an example. There's an eternal conflict
between brevity and verbosity raging in me. Some days one side
wins, some days the other :) But your example was correct.
> However, although what I did made the output repeatable, it didn't
> distinguish files from directories so the output would still "jump
> around" in terms rather like it currently does, in the sense that the
> diffs of sibling files in one directory would be separated by diffs of
> subdirectories wherever those appear in lexical order of their names.
That would be *slightly* better, but only slightly.
> I would be very interested in seeing this done, for reasons like yours
> (e.g. being able to compare one diff with another). I don't know
> whether it would be feasible to group all of the subdirectories
> separately from the files. Is that important to you?
I think "important" is too strong a word, but it is certainly
the behaviour I expect. There is nothing functionally wrong with
what svn currently does. It produces diffs just fine. However, it
is (at least for me) fairly common to have to break up a large
output from svn diff (previously cvs diff) into smaller patches.
That task is made more difficult, but not impossible, with the
current mechanism.
Thanks to pointers from Dan Berlin, I have a direction to head in.
I am certainly going to try for breadth-first, but will settle for
simply lexically sorted if the breadth-first thing seems to be too
invasive.
Kean
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Nov 15 21:04:39 2005