Re: CVS is the `released version'

[David Kastrup]

Richard Stallman <address@hidden> writes:

>     rms> 1. It could reduce the incentive for people to assign copyright
>     rms> on their code.
>
>     Yeah, this is definitely to be preferred.  I'm not comfortable
>     requiring it, though, because there are very useful Emacs Lisp
>     packages which, AIUI, will never be assigned to the FSF.
>
> I'm not comfortable with having Emacs automatically install and load
> anything that isn't copyright FSF.  So if we do install a feature like
> this, it will use only one repository, on gnu.org.

I am not comfortable with having Emacs automatically install and load
anything, period.  I already expressed my opinion that requiring a
particular _repository_ would be quite a mistake.  While I agree that
something like that should likely be preconfigured to point to
something under our supervision, I consider the "fixed repository"
approach something that does not work out.  Each package should carry
with itself the information where to ask for updates.

There is a lot of code around for Emacs for which we do not even
_want_ copyright assignments and control since we have neither the
manpower nor the willingness to cater for it.  Providing for a simple
mechanism and file layout that will make it easy for the package
writers to provide the Elisp files, documentation, supplementary files
in a way where two or three Emacs commands will pull the package into
the Emacs' private site-lisp tree and byte-compile it there, would be
quite beneficial.

It would also decrease the motivation of Debian developers to come up
with their own system of placing third-party packages into Emacs
variants, a system that neither Emacs developers nor Debian developers
actually really understand in its complications and side-effects.

If we provided a standard way of getting some package layout pulled
into Emacs, possibly we could persuade Debian to stop providing Emacs
and package versions that force people to deinstall the Debian Emacs
version and replace it with a hand-installed one if they ever plan on
contributing to external Emacs projects (like AUCTeX and other).

So while I agree that the connection "central
repository"-"GNU"-"copyright FSF" seems somewhat natural, I don't want
a central repository as a necessary component of a package system or
packaging policy.

That, indeed, would not buy us so very much beyond just putting
everything into Emacs CVS itself.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum