Re: What to do about unmaintained ELPA packages

[Tim Cross]

Dmitry Gutov <dgutov@yandex.ru> writes:

> On 30.05.2022 00:34, Philip Kaludercic wrote:
>> There are some popular packages on GNU ELPA (and I expect NonGNU ELPA)
>> that are practically unmaintained.  One example would be Yasnippet that
>> has been gathering issues and pull requests on GitHub, mostly without
>> any comments whatsoever.  For example, see
>> https://github.com/joaotavora/yasnippet/issues.  Does anyone know of any
>> other packages of this kind?
>
> Talking about yasnippet in particular, it more in the "stable" rather than
> "bitrotten" category, so I wouldn't worry about it too much.
>
> Or definitely not resort to measures like removing the reference to the
> upstream.
>
>> I'd like to ask, if there some point at which should one should go from
>> regarding packages like these from "de facto unmaintained" to "actually
>> abandoned"?  Perhaps if there was no real activity for over a year,
>> despite constant contributions?  Would it make sense to call for anyone
>> new to take over maintaining the package?  Or depending on how long the
>> package has been unmaintained, how popular the package is, how much
>> effort it would take to apply the changes one could modify the package
>> in elpa.git/nongnu.git and inform the maintainers that if they decide to
>> start working on the package again, that there are downstream changes
>> that they should look at.
>
> Personally, carrying over the development on ELPA would seem 
> counter-productive.
> Both due to the reduced potential community of contributors and reporters, and
> because of the wealth of reports, discussions and docs that reside at the
> currently dormant upstream. Kinda passive-aggressive, too.
>
> I think the best step right now would be to try to contact Noah and ask to 
> share
> commit access. And if not Noah, then Joao -- he's definitely still around.


I would agree. First step is to try contacting the repository owner. I
notice in the case of yasnippets, they are active in other projects, so
there may be a good reason none of the issues or PRs are getting action
in that repo. 

I'm not sure there is a 'one size fits all' answer here. We probably
need to look at each case individually. 

I do think both ELPA and non-GNU ELPA would likely benefit from some
statistic showing number of weekly/monthly downloads and/or possibly
some heuristic quality metric i.e. number of open issues, whether the
package has a test suite, number of compiler warnings, days since last
update etc. While none of these are likely sufficient metrics on their
own, perhaps a combination could be useful as an indicator. 

One unfortunate tendency in the current climate is to consider anything
which has not had an update in some time to be abandoned. However, it
could simply be it has reached a stable point of maturity where fewer
updates are necessary or critical enough.