[svk-devel] Why push svk:mirror?

[Chia-Liang Kao]

> I am a bit puzzled about the svk:mirror property. I understand why it is
> needed, but why is it pushed to the upstream server? When mirroring a
> server, the information about the merge status is only relevant for the
> local copy, not for the upstream server. At least I do not see how he
> can make use of it, any the local repository knows anyway.

ITYM svk:merge?  you can use --no-ticket on merge operations, but svk
won't be able to do smart merge at all.

> In fact I would find it much more pleasant if svk:mirror would only be
> stored locally: First, it would make svk usage completely transparent
> (to the upstream server), svk would just be a much more powerful svn
> client.  And second, the svk:mirror property grows with the number of
> svk users (i.e. developer), which is unaesthetic. Admittedly these are
> not real problems, but I would still think a solution without upstream
> svk:mirror nicer.

The reason the merge meta data are in the mirror is because you don't
just merge between branches in your local depot, there could be merges
between published branches.  And not storing the meta data or only
having it overlaying on the local mirror will make it impossible for
other svk users to perform smart merges.

Cheers,
CLK