Slime project considering Mercurial

Paul Franz theandromedan at gmail.com
Fri Mar 7 12:20:41 CST 2008



Adrian Buehlmann wrote:
> On 07.03.2008 12:46, Paul Franz wrote:
>> So the local India developers would pull from the local gate 
>> repository but push to the one in the US, correct?
>
> Yes.
>
> If you have your main repository A in the US, you can keep
> a clean replica B (a clone) of that in India on a computer
> on their local network.
>
> The first cloning of A to B may be slow, if
> it's a large repo with many files and changesets
> (see http://www.selenic.com/mercurial/wiki/index.cgi/RepoSamples
> for some real numbers).
Before versioning it is:

Number of Files: 44630
Total Size of Files: 1.4GB

>
> Subsequent pulls to update B from A will be
> very fast, since only missing changesets are transferred.
>
> You can do a "hg verify" on B to verify that B is fine -
> which is reasonably fast and perfect, since every tiny little bit
> is covered by the SHA1 hashes of all heads.
>
> Individual developers in India should then clone B for their work
> (instead of A) and push to A (better not to B in order to keep it
> a clean copy of A).
>
> The main thing with this setup would be that B would
> never diverge from A, just lag behind, if it is not up
> to date. So no merges would have to be done in B.
>
I think the minor delay would be ok. Since currently, the developers 
only update once a day (at least that is my understanding)
> If you don't want every single developer in India to
> push to A, you might create an outgoing repo C in India, which
> contains all integrated changes from developers in India
> (contains merges) and push/pull from C to A (integrate
> into A).
>
 From other people in the thread it seems I would need a gatekeeper to 
resolve integration issues.
> A single developer might do the same for himself. Having
> his own clean "B" on his computer and always clone that for
> specific work.
>
> Just a possible working scenario.
Possibly. I am hoping that any "playing around" would be done on a local 
branch and therefore avoid a need for 2 clones on the box.


Paul Franz


More information about the Mercurial mailing list