[BioPython] Martel problems and MS-Windows

[Brad Chapman]

Hey guys;

Michiel:
> >You are right, Martel did get skipped because it was already present 
> >on my machine. Why is it necessary to check version number of Martel? 
> >Is it possible that a version of Martel is already present that is 
> >newer than the one included with Biopython?

Jeff:
> This was my doing.  At the time (and still now, I believe), Andrew 
> wanted also to distribute Martel as a separate package, that people can 
> use outside of Biopython.  So thus, the CVS version, and possible 
> distributions, were ahead of Biopython.  But now that it appears that 
> Martel development is stabilizing, perhaps it is time to revisit that 
> decision.

My vote is that it's not really necessary, since I thought that
distutils would already do the work of determining if the files to
be installed are newer than the stuff in site-packages. Of course, I
guess this based on time-stamps so it's not a perfect system, but
should work for most cases and make things a little less confusing.

On the other hand, it's not too tough to bump the version on making
new releases or if new changes go into CVS -- I think I was just
sleeping (well, really studying for exams) during the whole Martel
install debate a while back, so I didn't know to do this on the last
release. I added it to the build release directions, so it shouldn't
be a problem in the future if we leave things the way they are.

By the way, Michiel -- the quick fix is to just remove Martel from
your site-packages before building the installer. We'll have a real
fix of some sort for the next release.

Brad