[Bioperl-l] Pushing towards 1.0

[Hilmar Lapp]

Ewan Birney wrote:
> 
> It should be perfectly obvious to most people that I would love to be able
> to say we have reached 1.0 branching for the Oreilly conference at the end
> of Janurary.

I think that's just what _at least_ the core feels. However, see
below.

> [...]
> I actually think that we are pretty close to a feature-complete - in
> particular with the Structure objects coming in we have covered alot of
> bioinformatics. Our only missing hole is the Expression objects - noone
> really looked at bringing in the objects from BMS.

I don't even think we'd really need this integrated to declare
1.0, because for the time being it could very well live
stand-alone (and maybe even should be a stand-alone module
initially, just to throw a thought).

> 
> I'd like to suggest we go for a 0.9 series developer release this week
> (Jason - Hilmar - Heikki - sounds good?) I am happy to push it through its
> paces.
> 

Definitely. The developer releases are for 'release early and
frequently'. We still don't do this very well -- the problem being
that making the package build cleanly (the only basic requirement
for a developer release) is less easy to accomplish than it seemed
to be. The reason for this as I see it is that actually the
development activities have risen substantially since summer,
i.e., it is 'good', but as I said it makes cutting cleanly
building development releases more difficult.

> 
> We have 24 bugs in incoming, with I estimate about half of them being not
> really active, so we are looking good there.

I wouldn't underestimate this. A year ago, resolving bugs took a
significant time and a concerted effort, and don't forget that
fixing bugs can unforeseenly trigger API changes if you want to do
it cleanly because they can be caused or made dreadful by
mis-designs. You don't want to postpone bugs after 1.0 unless
you're relatively sure their impact is neglible and/or fixing
won't trigger an API or behaviour change.

> 
> Before christmas we should discuss what we need to check off before the
> 1.0 release

I'd love to, but I have to admit that I for instance won't have
any time before next weekend due to several deadlines. Making this
a year-end goal exploiting the sit-back-and-have-a-second-look
time during the holidays seems more realistic to me.

> 
> Is there anything else people really, really want to see in before we
> branch for 1.0?
> 

Consistently applied factory pattern. Ideally, in 1.0 nowhere in
the code the class of an instantiated object is hard-coded except
in Bio::Factory::*. Otherwise you can't really claim Bioperl is
extensible (well, you can, but have to admit at the same time that
there are serious limitations).

Implementing this is not what anyone is really keen on in the
first place, the reason being that it doesn't add any productive
functionality in the first place (the reward comes only later).

> Jason/Hilmar/Heikki - do you think I am mad for pushing for this?

I more than agree that it would be nice. However, given the fact
that presently Jason and maybe Heikki are the only core developers
who can dedicate the majority of their time to polishing up the
package, and even they can't for a prolonged time I guess, I don't
think it is realistic if 1.0 shall live up to the principles we
had when not declaring any of the previous releases 1.0. Alone the
time for letting people test-drive the new functionality is not
sufficient, even if you count from now. Declaring 1.0 hastily
after not having declared it for 5 years is not the best idea I
can think of.

What I can image as very realistic could be e.g. calling the 0.9.x
current at that time the 1.0-alpha pre-release or something
similar.

To conclude with the obvious, all this is entirely my own opinion,
and in addition the opinion of someone who presently doesn't even
produce bioperl code ... So, if you really want to aim at a real
1.0 by mid/end of January, go ahead, and I'll do what I can to
support it, which may not be that much.

	-hilmar
-- 
-----------------------------------------------------------------
Hilmar Lapp                              email: hilmarl@yahoo.com
San Diego, Ca. 92130                     phone: +1 858 812 1757
-----------------------------------------------------------------