lost interest?

Skip Montanaro skip at pobox.com
Tue Dec 11 13:23:29 EST 2001


    >> CPAN is marvelous.  The way the Perl and Python communities operate,
    >> it fills a niche perfectly in the Perl world, but I think would be a
    >> bit of a solution looking for a problem in the Python world.

    Alex> If it encourages more reuse of code in module A by the author of
    Alex> module B, I think it would be a VERY good influence "in the Python
    Alex> world".  Why *shouldn't* the useful 3rd-party modules we use be
    Alex> free to rely on other useful 3rd-party modules yet?!

Python authors are free to rely on any modules they like, third-party or
otherwise.  That they don't suggests to me that there's generally fairly
little need to do that, at least for widely used packages.

To make the example concrete, here are the "whatrequires" dependencies for
the various python-related packages installed on my Mandrake system:

    *** python-devel ***
    no package requires python-devel
    *** rpm-python ***
    no package requires rpm-python
    *** python-imaging ***
    no package requires python-imaging
    *** python-docs ***
    no package requires python-docs
    *** python ***
    python-numeric-20.1.0-1mdk
    pygtk-0.6.8-7mdk
    rpm-python-4.0.3-0.27mdk
    tkinter-2.1.1-3mdk
    python-imaging-1.1.2-3mdk
    gnumeric-0.70-3mdk
    gif2png-2.4.2-1mdk
    python-devel-2.1.1-3mdk
    python-docs-2.1.1-3mdk
    *** python-numeric ***
    pygtk-0.6.8-7mdk
    *** tkinter ***
    python-imaging-1.1.2-3mdk
    pysol-4.72-2mdk

The only dependency listed between two non-core packages is between gtk and
numeric.  (Tkinter is really a core package, and is probably only separated
for convenience for people running servers that require no X capability at
all.)  I installed and use gtk happily without ever having installed
numeric, so I conclude that either there is some optional stuff installed in
Mandrake's pygtk RPM or the dependency is an error.  All other listed
dependencies are against the core package.

Skip




More information about the Python-list mailing list