Re: ELPA policy

[Artur Malabarba]

2015-11-17 23:17 GMT+00:00 John Wiegley <address@hidden>:
>>>>>> Richard Stallman <address@hidden> writes:
>
>> We want people to arrange to contribute their Emacs add-ons to Emacs, and
>> that suggests that whenever someone says a reason why he didn't want to put
>> his code in GNU ELPA, we should try to correct it.
>
> Agreed. ELPA should become a more positive experience, both for developers,
> contributors and users.

I'm deeply in favor of making it easier for people to contribute code
to Elpa, and (like I said), I'm willing to help write the code for
that.
But I think the hardest part of the problem is not being discussed here.

Whatever this method is, it needs to meet two criteria simultaneously:
1. Not commit any code to Elpa until someone with push access has OK'd
it. This is just to ensure nothing malicious is being done.
2. Be fairly automated (not completely), so that the Elpa maintainers
don't have to manually commit+push all code that gets sent to them.
This is to preserve the sanity of the maintainers. The current model
where everything is commited by us just isn't scalable.

While those two points are not mutually exclusive, they are quite
difficult to reconcile. It needs to show the maintainers a diff of the
changes and then only proceed with the commit after receiving some
very minimal interaction (a reply to an email, or the click of a
button).