[Mailman-Developers] Re: Python-2.2 disaster

[David Champion]

On 2002.03.05, in <514667500.1015300858@[192.168.0.1]>,
	"Carson Gaspar" <carson@taltos.org> wrote:
> OK, I "fixed" the _socket problem by hacking setup.py to change how it 
> found ssl_incs and ssl_libs, and by adding 'runtime_library_dirs = 
> ssl_libs' to the exts.append clause.
> 
> Can one of the Python gurus here tell me if I just missed an obvious 
> "correct" way to fix this, or if this is really a broken build process?
> 
> Now to try and fix audioop and fpectl...

I managed to get it to compile by ditching all my preferences and
letting it do whatever it wants to. (I should have learned by now that
my computer always knows better than I do.) It made me ornery, too,
but it got me a python build. In particular, I couldn't figure out
how to make it compile without fpectl, so I now have a dependency on
libsunmath.so (which is a licensed product, and ergo not installed on
all my systems, so my python won't work on all my systems). And I had to
settle for Solaris's curses, rather than using my ncurses installation.
Dommage!

I'm hopping back to 2.0.8 next time, and hoping 2.2+ is a little
friendlier when finally I'm compelled to submit to the force of
"progress", rather than just assuming the latest is the greatest.

-- 
 -D.	dgc@uchicago.edu	NSIT	University of Chicago