[Bioperl-l] Final BioPerl 1.6 release

[Chris Fields]

On Jul 16, 2010, at 10:23 AM, Robert Buels wrote:

> My thoughts on this:
> 
> It sounds like a good plan.
> 
> Also, it increases the urgency for the stuff on the development roadmap for the 1.7 release (well, releases!), the biggest of which is breaking things up into smaller distributions with independent versions.

Precisely.

> Jonathan Leto and I did some pilot work on the nuts and bolts of doing this, and we discovered something a bit counterintuitive: it's much better to do any large-scale splitting of BioPerl by starting at the root of the dependency graph, splitting off Bio::Root first.  The reason for this is that each distribution's build system must keep a list of the modules that it depends on.
> 
> If splitting is not done bottom-up, the dependency lists of dists that have already been split off have to be continually changed (manually, usually) to keep them up to date, which is not scalable for the large numbers of distributions that will be involved here.
> 
> So, something to think about going forward.
> 
> Rob

I mentioned this at BOSC (it's briefly touched upon in the talk); it was agreed that the bottom-up approach is the best.  Basically, if we can somehow emulate the larger distribution by allowing installation of the various packages then it makes life much easier for users, just have to decide what is the best approach.  

chris