[Bioperl-l] new cvs module bioperl-runnable

Catherine Letondal letondal@pasteur.fr
Thu, 20 Jun 2002 11:48:06 +0200


Heikki Lehvaslaiho wrote:
> 
> Separating the modules to an other cvs module might help manging the large 
> number of modules

yes, definitely

> but it does not help the main problem: how to oganize the 
> wrappers and make sure that users are not confused by modules when binaries 
> are not available.

Since they are run on a remote server, they don't have to be available locally?
I agree that whether the API is for local or remote execution
should be more explicit in the bioperl hierarchy.
Maybe Bio::Tools::Run::Local and Bio::Tools::Run::Remote distinction could be
useful? Or, on the contrary should it be hidden to the user of bioperl?

> How about using the Registry to centrally manage the 
> names of the binaries? We have not really started using Registry even for 
> databases!
> 

A registry would tell where to find the program, I guess?
But again, you don't have to take care of this in using a client. It's the
server's responsibility. In the perl code you just have to do:
	 my $program = $factory->program('genscan');
What is currently done (for now!) in Pise factory, is to give a default
location for the server of a program.

--
Catherine Letondal -- Pasteur Institute Computing Center