[Bioperl-l] updates, new code, invitation to contribute

[Hilmar Lapp]

On Tuesday, April 26, 2005, at 01:11  PM, Jason Stajich wrote:

> Bio::Factory::FTLocationFactory
>  Hierarchical parsing works even for joins of joins now.  This is 
> using a regexp solution which seems to be suitably fast but has the 
> problem of not working on perls < 5.6.1.
>

It won't be difficult to make the test conditional on the perl version, 
but I wonder what happens if a perl 5.005 or lower loads the module and 
needs to compile the expression. Haven't had a chance to investigate 
yet.

> [...]
> Bug fixing help!
> There are a lot of bugs sitting in the bugzilla queue.  I realize it 
> may not seem easy for someone to jump in and work on, but that is how 
> you learn the toolkit (it's how I learned to do things on this 
> project).

What would also help I think if people can write up small short test 
cases that expose the bug.

> [...]
> I think it may be worth considering moving stuff that has lots of 
> dependancies out of the main core code and into separate installable 
> packages.  IO::String is not a large dependancy, but LWP starts to be 
> add any of the XML modules, Graph, etc and it seems to be too large of 
> a hurdle for many new folks.  Maybe anything which depends on code 
> linked to an external C library would be a candidate.

In a way I agree but what I'm not sure about is how separating out 
modules with compiled dependencies will help if you want those modules 
anyway. If you don't want them then failure to install those 
dependencies should not be a cause to worry, and in fact one should 
then not even bother to try to install them.

So what is happening when someone "fails" to install bioperl? Did that 
person try to install all dependencies and then gave up over the 
ensuing hassle? Maybe we need to better document which dependencies are 
worth installing and which aren't unless you are certain you'll use a 
particular set of modules or functionality.

Maybe it's the Bundle that's trying to do too good of a job by 
including virtually all dependencies? Maybe there should be different 
bundles for different intended uses?

How would separating out modules with compiled dependencies help an RPM 
maintainer if she wants to build an RPM for the entire bioperl anyway 
(in either one or a set of RPMs)?

Installing bioperl is indeed a PITA, I've done it too many times. The 
main pain, however, in my experience comes from LWP, GD, and sometimes 
expat. None of these are needed for much of the core functionality, but 
based on list traffic many users out there do want GBrowse and/or 
Bio::DB::* modules, for which you'll have to have them.

	-hilmar
-- 
-------------------------------------------------------------
Hilmar Lapp                            email: lapp at gnf.org
GNF, San Diego, Ca. 92121              phone: +1-858-812-1757
-------------------------------------------------------------