Hi Joerg,

On Thursday 26 April 2007 09:47, Thorsten Behrens wrote:

> > What would be an acceptable time for a cwsresync for you? 36 hours
> > certainly seems much too high, at least when the resync is to be done
> > regularly. But would 1 hour perhaps be acceptable? And if with
> > Subversion a resync were possible within 1 hour, would the scenario
> > which Bernd described be imaginable for your work?
>
> well, given that git is able to perform the task in about no time, why
> would anything that takes an order of magnitude longer be acceptable?

Yes, Thorsten is right.  Few minutes is acceptable, an hour is not:  we have 
1000 patches, let's say one feature consists of 10 patches, it would be 100 
branches, making it 100 hours in the worst case.  Of course I don't know what 
would be the average one, but even 10 is too much.

Currently the update to new milestone in ooo-build is not limited by tooling 
at all, it's just the speed I (or whoever does the update) can edit 
patches...

> Besides that, my trouble with that centralized approach are e.g. that
> this mandates exactly one policy - for commit rights, licensing,
> process, you get the idea. Having a staging tree of DSCM repos allows
> for relaxed rules on some of them, while still maintaining our
> standards on the 'vanilla' repo. I mean, a dev can simply clone that
> repo, and commit right away on his disk, without any administrative
> overhead whatsoever.

Exactly.  And still - up-stream can use these changes after signing the JCA, 
etc. extremely easily.

Regards,
Jan

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to