Re: Complied into debs please (Python/Emacs packagers take note)
On Sun, Feb 24, 2002 at 09:50:44AM +1100, Brian May wrote:
> On Sat, Feb 23, 2002 at 04:58:12PM +0100, Torsten Hilbrich wrote:
>
> > > Subverting the Debian package management system seems a bit odd for
> > > a Debian package to do, yet this is the impression I get from each
> > > Python package I install here. (Emacs packages tend toward the same
> > > stupidity, and this is why I try very hard to avoid installing python
> > > or emacs packages on most of the machines I administer.)
> >
> > The Emacs bytecode is incompatible with the XEmacs one. The same can
> > be true for new versions of these programs. So there are good reasons
> > do not distribute the byte code for all different Emacsen versions in a
> > debian package but to compile it on installation.
>
> Why can't the bytecode be compatible?
>
> Is there some good reason why every LISP interpreter must use
> incompatible byte code?
If every Lisp interpreter or compiler was semantically equal then perhaps
such a project could be feasible. However there are far more pressing
issues for the Lisp community to deal with; supporting backwards, poorly
implemented, out-of-date Lisps such as Emacs Lisp is not one of them.
The Common Lisp packages in Debian suffer from a similar problem; every
package needs to be compiled for each compiler you have installed.
The FASL file-formats happen to be something which can differ from
compiler to compiler (and in the case of SBCL, for good reasons), and
even from major-version to next. Currently the common-lisp-controller
package manages this, but people complain about the re-compilation on
upgrades all the time. Hopefully, if we find a better solution, it may
be applicable to Emacs and Python packages as well.
--
; Matthew Danish <mdanish@andrew.cmu.edu>
; OpenPGP public key: C24B6010 on keyring.debian.org
; Signed or encrypted mail welcome.
; "There is no dark side of the moon really; matter of fact, it's all dark."
Reply to: