[Bioperl-l] new cvs module bioperl-runnable

[Jason Stajich]

On Thu, 27 Jun 2002, Elia Stupka wrote:

> > It is unclear how the tests should run - should they be separated into the
> > specific cvs modules or still located in the t/ directory of bioperl-live.
> > I am leaning towards bioperl-live t directory since we will plan to
> > distribute the modules as a single tarball - devs will just need to know
> > how to set their own PERL5LIB.  I think this is how ensembl does the
> > distributed modules with a single test suite in the ensembl cvs module.
> > Opinions?
>
> In Ensembl we do have tests in each of the cvs module, and I must say it
> kinda makes sense to me intuitively to find the tests there. Also if I
> have not checked out bioperl-run because I am not interested in it, why
> would I want to run the tests?
>
> Finally, specifically in the case of bioperl-run, most tests will really
> only run if the binaries are installed, so again, if I have no binaries
> and am not interested, I will just see tons of tests being skipped.
>
> Strongly in favour of tests in each cvs module... ;)
>
okay - I'll buy that - we'll need to migrate a bunch piddly files from the
t and t/data directories from bioperl-live then - can I count on you/your
team to move the relavant files for your modules.

> (About LazySeqs, I was proposing EmptySeq because I thought the whole
> issue was not how to represent lazy fetched sequences, but simply when one
> does not want to have a sequence at all, but knows its length and has
> features...)

I think we're mixing two things together:

An EmptySeq - we would never expect to get a sequence string back when we
call $seq->seq(), but can call length(), molecule/alphabet, is_circular,
etc.  Additionally SeqI and RichSeqI interfaces can be inherited as well
to build a composite of these interfaces.

Should LazySeq should really be an implementation specific soln and
not necessarily an interface?

>
> Elia
>
>

-- 
Jason Stajich
Duke University
jason at cgt.mc.duke.edu