On 6/7/12 20:25 , Nicolás Alvarez wrote:
> I think this could justify putting the client and the server in
> separate git repositories. Each repository would have its own tags and
> branches. The server.git/stable branch would have no client code, and
> client.git/7.0 would have no server code, thus no branch would have
> code in an "indeterminate state".

One can achieve that by using development/idea branches during
development, without separating the code. Those branches would be merged
into master when they're ready (tested). This way master would be much
more stable/reliable. Server code releases should then be branched of
from master and maintained in those branches (as the client releases are
today).

> The problem is the shared code in lib/. Putting it in a third
> repository and configuring it as a git-submodule of the other two
> would work, but might be too cumbersome.

That's the point. There's no reason to do that (see above). I'm in favor
of separation where it's useful but I'm not if it just adds complexity
without actual need or increases the risk of loosing code coherence.


Oliver
_______________________________________________
boinc_dev mailing list
[email protected]
http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev
To unsubscribe, visit the above URL and
(near bottom of page) enter your email address.

Reply via email to