Re: Switch fork, version going backward

[Greg Troxel <gdt%lexort.com@localhost>]

Emmanuel Dreyfus <manu%netbsd.org@localhost> writes:

> mail/openarc is bsed on https://github.com/trusteddomainproject/OpenARC
> PKGNAME is openarc-20180921
>
> I recently submitted pull request for a bug fix and was told it was 
> alrezady fixed on a develop branch, for which releases seems to not
> be the priority. I am pointed to a fork that focuses on backporting
> fices to a stable release:
> https://github.com/flowerysong/OpenARC
>
> I tested switching from the former to the later with success, but there
> is a caveat: PKGNAME becomes openarc-1.3.0, which a is a lower version
> number than openarc-20180921.
>
> How should that be dealt with?

This is quite messy.

Ideally, upstream would make releases, but you report they aren't.
  - Do they advise that people just run the tip of master?
  - The latest tag is 1.0.0Beta3; I don't see 20180921.
  - What does upstream say about flowerysong?

  - What do you think about upstream.  Is it "they don't like releases"
    or "nobody is there and it's abandoned"?

flowerysong's fork is confusing.  The README looks to be the same, and
doesn't explain what it is: the old release plus fixes, a stabilized
version of develop, or something else.  It's not a fork in the github
sense; I don't know if it is in terms of git.  I would file a bug that
the README for a fork should make all this clear.

Generally, pkgsrc moves to a fork when the upstream is entirely
unmaintained or the larger community has consensus that the new repo is
the rightful place.  I don't get that impression so far.  What are other
packaging systems doing?



So:

  The 20180921 seems off to me.  The package is a tag from git.

  Given a tag now, why not just advance to some later tag?

  If flowerysong is a good thing for some users to run, perhaps
  openarc-flowerysong with flowerysong's numbers, as a different
  package, and people can run it if they want.