[Bioperl-l] Withdraw Bio::Graphics and Bio::DB::SeqFeature from bioperl distribution?

[Mark Johnson]

On Mon, Nov 10, 2008 at 3:46 PM, Chris Fields <cjfields at illinois.edu> wrote:
> Personally, I think if Bio::Graphics remains in bioperl we have to do two
> things.  We should release the full bioperl-live as-is to CPAN as an
> official release (TODO any bugs) ASAP.  No RCs; we'll post point releases
> along the way for bug fixes (I like the 'release early/release often'
> mantra).  I can work on this over the next couple of weeks, aiming for
> Thanksgiving for a 1.6, but I probably won't get rolling until this weekend
> (too much going on this week).  We can aim for more regular point releases
> then.

    Agreed.  We (the Genome Center at Washington University in St.
Louis) switched to a nightly build about a month ago.  Prior to that,
we had been on 1.5.2.  What's in the trunk at the moment may not be
perfect, but it's the best to be had anytime soon.  Slap the 1.6
sticker on it, tag it, bag it and ship it.

> Following that, I think a more stable long-term solution is to split off
> some of the non-core-like modules so that we can speed up releases (this has
> been discussed in the past,
> http://www.bioperl.org/wiki/Proposed_1.6_core_modules).  Basically, make a
> 'bare-bones' well-tested core containing the base classes and interfaces
> that remain stable long-term, such as Bio::Root, Bio::Seq/PrimarySeq,
> Bio::SeqFeature::*, with as few dependencies as possible.
>
> Everything else requiring constant maintenance, not actively supported, or
> under development would go into a separate monolithic distribution listing
> the new core as a dependency; this could feasibly have it's own release
> schedule.  If we go this route, Bio::Graphics and related could also be in a
> second distribution (and thus also on a distinct release schedule).  This
> could be worked out in a separate subversion directory, so bioperl-live
> wouldn't be affected until we switch over.  Does that seem feasible?

    I don't disagree with anything you've said.  However, I wonder if
maybe there isn't something to be learned from the way the Linux
Kernel Development process changed with 1.6 (such that there is no 1.7
development branch)?  You're the closest thing we've got to Linus, so
you've got my vote for 'benevolent dictator'.