[Biopython-dev] KEGG support

[Renato Alves]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

- From Peter on 02/10/2010 10:37 PM:
> On Wed, Feb 10, 2010 at 8:12 PM, Kyle <kellrott at gmail.com> wrote:
>> I think external library dependancies should be avoided unless necessary.
>>  Would a tool like wsdl2py produce code that isn't dependent on an installed
>> library? Alternatively, suds is LGPL based, could we just cannibalize the
>> source code for the important classes?
> 
> Working with SOAP is so complicated that using an external library
> would be the sensible option. It would be an optional dependency
> (and would not be an install time dependency like NumPy), much
> like how we have a optional dependency on ReportLab just for
> Bio.Graphics, and now also the option to use NetworkX with the
> new Bio.phylo code.

Yes that would be my idea on the SOAP interface. If doable we could even
evaluate the possibility of having some abstraction layer that could
enable the use of SOAPpy or suds if either is already available on the
system.

> Package management (e.g. under Linux distros) can mark these
> external modules as suggestions or soft requirements, making
> this quite straight forward.

The 'or' case for soap libraries would also fit in this scheme since
most package managers already support this kind of feature.

> Regarding some of Giovanni's points, modularising the distribution
> of Biopython (which can already be considered to be a core plus
> assorted domain-specific modules like Bio.PDB, Bio.Cluster,
> Bio.Graphics and so on) seems premature to me give the current
> state of python distribution.

Could you elaborate a little on what you mean by 'current state of
python...'. Are you referring to the python3 transition?

Renato
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.14 (GNU/Linux)

iEYEARECAAYFAktzVO0ACgkQYh11EUYTX9S1ngCfYFiW7VeNu6atl0J1eViqquSo
PCIAn3KO2p//fRYpZVC0QSp2gITP/n2I
=uTTc
-----END PGP SIGNATURE-----