Re: New version of www/gumbo-parser

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

Kevin Bloom <kevin.bloom%posteo.net@localhost> writes:

> So I've began working on porting wip/newsraft and I've ran into an
> unexpected roadblock, namely, www/gumbo-parser. Turns out Google
> has stopped supporting gumbo-parser but a new version of the project
> has been forked here:  https://codeberg.org/grisha/gumbo-parser
> What is the protocol for supporting this project? Do we update the
> existing www/gumbo-parser to now point to this repo or do we create
> a new recipe with a new name for historical reasons (if other
> programs depend on the original gumbo)? If we do the latter, what
> do we name the new one since they're still below version 1 (0.10.1
> vs 0.11.0, we can't call the new one gumbo-parser11 since it's not
> actually version 11 and will likely increment subversion quickly)?

It's very situation dependent.

Basically we have to decide between:

  A) The original project is abandoned and the new repo is a "continuation
  fork" that is widely recognized by the community as the rightful heir
  to the name.   As part of this, a useful question is "if package A
  that depends on gumbo-parser doesn't work with a release from the
  fork, do we say that this is evidence that A is not maintained?"

  B) The new repo is *a* continuation fork, but it isn't, or it isn't
  clearly, the sole logical successor.  It's ok for depending packages
  to choose not to support it.

Arguing for A typically is some blend of

  original upstream has blessed the fork

  other packaging systemes are treating it as the project

  there is no evidence of a competing fork

Arguing for B is any kind of "it's complicated" when figuring it out.

A quick web search leads to not finding anything else competing as a
continuation fork.  I was not able to quickly understand what Debian is
doing (which is about me not understanding how to figure out Debian
things, not about them).

So I am leaning to A this minute, weakly, but very ready to accept
falsifyin evidence.


Regardless, there is the question of "do we need the old version".  In
this case, it looks like the old version is really old.  pkgsrc users
are

  print/mupdf
  print/zathura-pdf-mupdf
  www/litehtml
  www/p5-Alien-LibGumbo
  www/p5-HTML-Gumbo

so it's a fair question to ask those upstreams if they see the fork as
legitimate and for someone to check if those package build with it.

You could locally update and test build those, and test-run mupdf.