/usr/bin/env: python: No such file or directory

[Paul Boddie]

quinn at groat.ugcs.caltech.edu (Quinn Dunkan) wrote in message news:<slrna22ic9.7k5.quinn at groat.ugcs.caltech.edu>...
>
> I don't know how much time I've wasted fighting with too-clever install
> scripts, broken Makefile-mazes that want to rebuild the world every time, GNU
> configure picking broken settings, setup.py deciding to start using
> 'gcc -shared' to make shared objects on SunOS when only 'gcc -G' works and
> what do you edit to change that, and programs which generate perl scripts
> which generate imakefiles which generate makefiles which generate shell
> scripts which sed out a bunch of magic tokens in configuration files, which
> then breaks, etc.  etc. when all I really needed was 'cc -c *.c -o program;
> cp program /usr/local/bin'.

This kind of reminds me of all those projects which use automake
instead of autoconf, or at least I assume they use automake, because
instead of pertinent questions being asked in the configure script,
you get hundreds of checks being run for no obviously good reason:

  Checking for maiden name of aunt... Jenkins

>From what I've seen of distutils, it does seem to do a reasonably good
job, and it certainly seems a lot more workable than the old
Makefile.pre.in/Setup thing (which, in turn, at least managed to do
the right thing most of the time, unlike certain parts of the KOffice
build process), but I haven't actually attempted to employ it in
anything I've written. Indeed, that's probably because I mostly stay
clear of writing C-level extensions. Perhaps the issue here is how to
get distutils to scale down so that it's trivial to use - even for
pure Python projects.

Having written the above, I now anticipate an example of distutils'
triviality... ;-)

Paul