Re: 21.0 showstopper bytecompile-time -vs- run-time dependencies
Oscar Figueiredo <Oscar.Figueiredo(a)di.epfl.ch> writes in xemacs-beta(a)xemacs.org:
>>>>> "sb" == SL Baur
<steve(a)xemacs.org> writes:
sb> Before I forget ...
sb> The current REQUIRES field is only a bytecompile-time dependency not a
sb> run-time dependency and should *never* be enforced at run-time,
sb> including package fetching.
sb> Since not all run-time dependencies should be enforced either
sb> (eg. Gnus run-time dependency on rmail), the solution isn't a
sb> straightforward download everything.
sb> Suggestions?
If I understand things well, this confusion probably stems from the
following
line in package-info.in that can be found in most packages, I believe:
requires (REQUIRES)
This isn't the right fix, alas.
Editing this line by hand for packages that need it should be enough
to
introduce the distinction between bytecompile-time and
runtime/package-installation-time dependencies.
No, that's not the case at all. To pick a specific example, there are
many packages which have a dependency on reportmail, but *don't* have
a bytecompile-time dependency on mail-libs. You don't need functional
outgoing email to run cc-mode ...
The run-time dependencies need to be in something completely separate.