[BioPython] Plans for release 0.95d01

Jeffrey Chang jchang@SMI.Stanford.EDU
Thu, 1 Feb 2001 17:46:41 -0800 (PST)


On Thu, 1 Feb 2001, Brad Chapman wrote:

> Jeff:
> > - Figure out how to deal with dependencies.  We've been accumulating
> > dependencies on external packages, i.e. Martel, mxTextTools (through
> > Martel), Numeric, and Spark.  Brad suggested writing code that will help
> > install those packages.  Are you planning on doing that for this
> > release?
> 
> I don't know -- I've kind of had misgivings about trying to write a
> hackish install package -- it might just make life harder for people,
> especially people like Johann who are trying to package Biopython for
> their platforms. If Martel is included in the Biopython tree (what was 
> the final word on this? Andrew?) then I think we should install it,
> but for other external tools, I'm kind of nervous about this. What do
> you guys think?

That sounds reasonable.  This may be a problem that catalog-sig will solve
for us if we leave it alone.


> > If not, I propose 1) adding Spark since it's only 1 file, 
> 
> I've already done this (using my time machine :-) -- it's in
> Bio.Tools.Parsers

Great!


> > 2) making > setup.py check for those packages, and if they don't
> exist, complain and > ask the user if they want to continue anyway.
> 
> How about just writing out a warning at the end of the install?
> Attached is a modified setup.py script which does this -- what do
> people think about this plan for now, and see how it goes?

Looks good, except that check_install should be generalized.  
import_string should actually be a function that checks whether the system
is installed on the system.  This way, the client can write code that does
more checks than seeing if something's on the system, e.g. can check for
the proper version.

Jeff