Sebastian Hilbert wrote: > >Actually no. :-) >There will be packages if someone packages GNUmed. This person makes the >decision. Another person might disagree and package GNUmed with another >technology. It's about doing it not making decisions. One thing I have >learned about GNUmed is that you can only make decisions for yourself. You >can make it as easy as it gets for other people to do your work. It simply >won't happen :-) > > I agree. Personally I think all python programs should be packaged with distutils, so that in the end there is a graphical installer for windows, and a command line installer for linux, which looks like:
python setup.py build (not always necessary and probably wouldn't be for us, since we have to C code AFAIK) python setup.py install The setup.py install accepts other arguments. This is how most python programs work and it seems to work very well. Mind you I have seen some excellent python programs which did not use distutils but made their own setup.py script, without importing distutils methods. Linux distro packagers just unzip the tarball and then run "setup.py intstall" with proper arguments for that distro. I tried to get this working before, but then stopped. It might be worth looking into again. BTW, I think the windows gui installer stuff can be produced using distutils by doing something like python setup.py build_dist_win... -- David J. Grant http://www.davidandnasha.ca _______________________________________________ Gnumed-devel mailing list [email protected] http://lists.gnu.org/mailman/listinfo/gnumed-devel
