pan pan pan

Milan Zamazal pdm at debian.org
Sat Aug 22 14:47:47 UTC 2009


>>>>> "RS" == Rupert Swarbrick <rswarbrick-Re5JQEeQqe8AvxtiuMwx3w at public.gmane.org> writes:

    RS> I don't think that debian packages are a waste of time. 

It is not, the problem is who will do the work.  I definitely prefer a
small set of well working core CL packages over hundreds of obsolete CL
packages in a poorly maintained state.

    RS> On the contrary, I spent flipping ages trying to work out how to
    RS> get a newer version of clsql working a couple of weeks ago. At
    RS> first, there was a problem with not having a new enough cffi and
    RS> then there was a different problem with an incompatible version
    RS> of libmysqlclient... "Known working" combinations, which can be
    RS> ensured using debian's package systems, would have saved me
    RS> several hours...

Yes, I've got a similar experience with something else.  Debian can help
here only if 1. upstream (including the used libraries) acts reasonably
with respect to releases, 2. the Debian maintainers have enough time to
follow upstream development.  And it may be still hardly possible when
recent versions are required, as is often the case with packages used
for development.  Would it help something to get a consistent set of
relatively up-to-date CL packages (just) at the time of each Debian
release?

BTW this problem is not specific to CL, CL just makes it less urgent due
to its small user base and the fact it's usually easier to install a CL
package than a shared C++ library.




More information about the pkg-common-lisp-devel mailing list