Re: What is GNU ELPA?

[Dmitry Gutov]

Hi Eli,

On 16.05.2020 17:43, Eli Zaretskii wrote:

> > - More generally give more freedom to the authors (many authors are
> >    *not* interested in having their code in emacs.git because of the
> >    extra constraints that it implies).
> >
> > The last two problems *also* affect GNU ELPA, ironically, tho to
> > a lesser extent.
>
> I'm sorry to say, but I feel we are trying to retroactively invent
> reasons to justify the actual situation and the goals that got shifted
> from their original.  As the description of ELPA shows, the original
> goal was to have an extension of Emacs that is separate for minor
> technical reasons.

You seem to be trying to produce logical rules for governing ELPA based 
on two lines from the manual written by Glenn likely on a whim 4 years 
after GNU ELPA was created.

And does it really matter why it was created originally? Perhaps it was 
made to fight forest fires or raise awareness about the surveillance 
state. Does that really matter now, 10 years later?

GNU ELPA is where it is because it turned out useful enough for both 
sides (users and package authors) that the latter contributed a 
considerable number of packages, and a significant fraction of them 
continues to be updated. Making the situation worse for either side can 
break the balance, and there will be fewer updates, as a result it will 
be less useful for the users, and in the end we could simply be forced 
to write it off as a failure.

The current situation is, it is okay-ish for the authors (for reasons 
Stefan described), though probably not ideal for minor technical 
reasons. As well a copyright assignment reasons, but that acts as a filter.

This filter, however, makes it not great for the users, because the 
result is, we provide less functionality out-of-the-box. Hopefully I 
don't have to explain why this is a problem.

On the authors' side, however, if you decide to dispense with some of 
the reasons people contribute to GNU ELPA, that can also lead to fewer 
packages in there, and the result for the users is predictable.