Re: CVS update: MODIFIED: libsvn_wc ...

Greg Hudson <ghudson@MIT.EDU> writes:

> > If a directory is a 'target' of a commit or update, examine its parent:
>
> It sounds like there are two potentially better solutions here:
>
> 1. Anchor the commit or update at the parent of the target directory,
> but don't actually examine the parent working directory.

I don't think we can just blindly anchor an editor on a
non-revision-controlled directory. IIRC, this leads to badness.

Honestly, I'm not sure at all why people object to examining a parent;
we're only doing it once, to determine where to anchor an editor. Is
there a great philosophical wrongness? I mean yes, it's a little ugly
to think about. But then again, so is the whole problem of versioning
directories. I've just never been clear on Greg/Karl's objections.

> 2. Fix the editor interface to allow changes (properties and removal)
> to the root directory.
>
> The second option has the advantage that it lets us set properties on
> the repository root directory, something I guess we just can't do now.

Yeah, this might be the key. In addition to the editor having a
"replace_root()" (which gives you back a dir_baton for editing
children), we could add "delete_root()" and "add_root()". :-)

This is a big change though...

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sat Oct 21 14:36:37 2006