Re: 21.0 showstopper bytecompile-time -vs- run-time dependencies
>>>> "sb" == SL Baur <steve(a)xemacs.org>
writes:
sb> Olivier Galibert <galibert(a)pobox.com> writes in xemacs-beta(a)xemacs.org:
[...]
>> Suggestions?
> Three fields:
> - minimal requires for runtime
> - requires for compile time
> - full requires for maximal operation
sb> Exactly correct. The minimal requires would be force-downloaded by
sb> the package-get stuffs. The maximal requires would need some kind of
sb> extra option.
sb> For a first cut, I'd be satisfied to have minimal requires handled
sb> correctly as the SUMO tarballs will handle everything else.
> But maybe that would be too much a hassle to manage...
sb> At the present time, yes. We *really* need some tool to compute
sb> run-time dependencies.
The "full requires for maximal operation" could certainly be determined
automatically. I doubt that the minimal runtime requires can. For instance,
EUDC can use three different backends none of which is strictly required. Each
of them depends on some other package. I can't see how a tool could determine
automatically which requirements are really necessary and which just enhance
functionality without my writing it explicitely.
Oscar