[Biojava-l] Re: New snapshot

[Ewan Birney]

On Fri, 28 Jul 2000, Thomas Down wrote:

> On Fri, Jul 28, 2000 at 10:37:27AM +0200, Gerald Loeffler wrote:
> > 
> > Excellent contribution - and a lot of volunteering (-:
> > 
> > Why not skip the branching-off of the release candidate branch (steps 1
> > to 4 below)? I've never actually seen anyone doing that. Why not  simply
> > fork the release branch off the main branch when everyone is happy
> > calling it 1.0?


I have seen the branch, bug-fix, merge branch before, but I would suggest
(from bioperl experience) the following format

Someone is the release point man (? Matt)

declare a "closed period" on the cvs head. Everyone who does a check in
from now on has to coordinate with Matt their check-ins.

Matt makes 2 or 3 release candidates on the main trunk encouraging people
to pick them up and bash them hard (? do you have a test suite yet?)


Matt then declares that people are ready to branch. Puts in the branch
tag, checks out the branch. Does any final silly fixes (eg, 1.0 in
documentation files etc etc).

First release should come quite cleanly off the branch.





The problem I found in bioperl was that people did not respect the closed
period and would send me patches at the end of the release candidate
period for large changes in code, trying to make it in "before the
release". 


I wasn't strong enough to tell them that they couldn't get in ;)




Have fun out there ...


-----------------------------------------------------------------
Ewan Birney. Mobile: +44 (0)7970 151230, Work: +44 1223 494420
<birney@ebi.ac.uk>. 
-----------------------------------------------------------------