[BioPython] distutils / configure and Makefiles

[Andrew Dalke]

Jeff:
> The down side is that they'll need to install either the distutils
> package or Python 1.6, which comes with a copy.


I believe in the interim some of the packages distribute distutils
along with the package itself.

>I still have some concerns about distutils, though.  I'm not sure whether
>there's a good way to deal with testing code or how it handles
>configuration.  However, I must admit that I don't keep up with the SIG,
>and these might already be solved!  Anyone more up on distutils?


You would think that my being on the distutils mailing list means I
keep up with the SIG, but you would be wrong.  From what I recall, you
should be able to do any sort of check you want to do.  It's something
like making your own method for handling certain cases.

>I'm not against using a more conventional configure/Makefile type
>build/installation process for unix-like environments, provided that the
>solution is reasonably cross-platform (at least Solaris, Irix, Linux) and
>handles extension packages.  However, someone will need to put that
>together!


That's what my mods to GNU's automake do.  Because the starship site
crashed, I've put a copy at ftp://ftp.ks.uiuc.edu/pub/group/dalke/ .
I've tested it under Solaris, IRIX and Linux.  It does take a bit getting
used to, since automake/autoconf are somewhat strange creatures.
The Seq proposal I had uses it, if anyone wants to see how it works.

I haven't had a chance to work with distutils yet.  I know Greg Ward
(Mr. Distutils) has some good plans for it - it was contributed as
one of the Software Carpentry tools.  For what I was doing it wouldn't
have been appropriate, since we needed the build system to handle non-
Python and non-Python-extensions.


To summarize this post, I don't know the answer, but distutils is
probably better.

Mmmm, need to see how it handles regression tests.

                Andrew
                dalke@acm.org