Re: Gumbo-parser package in NetBSD

[Grigory Kirillov <txgk%bk.ru@localhost>]

Hi, Greg

> It would probably help if you are able to add an updated package to wip,
> as it is easier for us to review and update with the exact desired state
> being readable and tested, with a COMMIT_MSG file that has the summary
> of NEWS.

I'd really like to help to the extent that it's up to me as the maintainer of
the project, but creating recipes for packages in the repositories is in my
opinion work overload pretty much. The situation is also complicated by the fact
that I have never used the pkgsrc system, let alone created packages in it. I
can assume that for this I need to install and configure some SDK, as well as
figure out how it all works. Unfortunately I have to leave this work on the
shoulders of the pkgsrc maintainers of the gumbo-parser package.

> Have you been able to contact the original author?  My standard question
> would be if the fork is blessed,

Yes, I emailed last committers to the original repository of gumbo-parser
(Andrew Whalley and Jonathan Tang) with a request to bless this project as the
heir but didn't have the luck to get an answer from any of them yet.

> or if there is anyone else doing this,
> in terms of if there was/is a community.   Or perhaps you have forked
> and have not found anyone else who is dealing with this software.
> There are no fixed rules for pkgsrc here, but I like to understand "why
> is this fork the true heir".

I got a similar question from the package maintainers in the Debian
repositories, here is what I replied:

Many projects continue to use slightly modified versions of the original
gumbo-parser (through vendoring it in) to suit their needs; but some projects,
like sigil-gumbo and nokogiri, implement big additions and changes that make
their fork incompatible with the original gumbo-parser. These, in my opinion,
are separate projects now and making them new gumbo-parser most certainly will
cause reverse-dependencies to break.

Maintained project for gumbo-parser which is started by me, on the other hand,
is just a continuation of original gumbo-parser with an aim to fix bugs and do
some performance improvements while continuing to work as before until HTML6
replaces HTML5. Major changes with a huge degree of probability will not occur,
but if it becomes unavoidable, before releasing a compatability-breaking
version, we will release a preliminary version with deprecation warnings for a
year or two, so that all dependent projects have time to change their code to
work correctly. And only after all this big version will be released, which will
be accompanied by a change in the major version and soname numbers, as it's
supposed to. Again, it's very unlikely that some big changes will come.

These are my thoughts why gumbo-parser package should switch to Codeberg
repository grisha/gumbo-parser (Gentoo, Alpine and LiGurOS already made the
transition and I think some other repositories are on the way).

Best regards, Grigory Kirillov