Julian Foad <julianfoad_at_btopenworld.com> writes:
> The "tree-conflicts" branch stores tree-conflict information in the
> admin dir's "entries" file. Can anybody comment on whether it need to
> be bumping the WC format in order to do so?
What would be the consequences of an old client ignoring the
tree-conflict information, or of writing out the .svn/entries file
without that information?
If the consequences of either would be bad, then I think the format
needs to be bumped. It might be preferable to bump it only if/when the
tree-conflicts code does something, so that merely touching an old
working copy with a new client wouldn't magically render the working
copy unusable by the old client.
-Karl
> <markphip> I assume it bumps the WC format?
>
> <julianf> Huh... I haven't looked at the WC-recording part! That's one
> of the parts that works well enough for me to leave it alone for now,
> so I don't actually know. I'll find out.
>
> <julianf> No... it seems to leave the format at 9, which I think is
> where v1.5.0 has it.
>
> <markphip> I wonder if that is just for convenience and has to be changed?
> <markphip> for example, sparse directories clearly needed to bump the
> format but lived in trunk a long time before it did
>
> <julianf> Indeed.
>
> <markphip> you should not be able to use a 1.5 client and commit with
> conflicts. If we bumped the WC you would not be able to[.]
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe_at_subversion.tigris.org
For additional commands, e-mail: dev-help_at_subversion.tigris.org
Received on 2008-05-23 22:56:36 CEST