Re: git: 9942b0aef604 - main - archivers/liblz4: revert switch to meson

From: Matthias Andree <mandree_at_freebsd.org>
Date: Thu, 06 Nov 2025 19:44:28 UTC
Am 06.11.25 um 18:37 schrieb Daniel Engberg:
> On 2025-11-06 18:02, Matthias Andree wrote:
>> Am 06.11.25 um 17:43 schrieb Gleb Popov:
>>> On Sun, Nov 2, 2025 at 12:39 PM Matthias Andree <mandree@freebsd.org> 
>>> wrote:
>>>>
>>>> The branch main has been updated by mandree:
>>>>
>>>> URL: https://cgit.FreeBSD.org/ports/commit/? 
>>>> id=9942b0aef6049d260a451b255faf4cadeb44b7b0
>>>>
>>>> commit 9942b0aef6049d260a451b255faf4cadeb44b7b0
>>>> Author:     Matthias Andree <mandree@FreeBSD.org>
>>>> AuthorDate: 2025-11-02 09:24:36 +0000
>>>> Commit:     Matthias Andree <mandree@FreeBSD.org>
>>>> CommitDate: 2025-11-02 09:37:50 +0000
>>>>
>>>>      archivers/liblz4: revert switch to meson
>>>
>>> I wonder if using recently introduced USES=meson:muon would be a 
>>> better fix?
>>
>>
>> Why or how would switching to a new and less well understood tool 
>> that's not even mentioned upstream be an improvement and is documented 
>> to not be fully compatible to meson?
>>
>> What remaining problem does it solve that would outweigh our deviating 
>> from upstream choices that raise the support effort on both ends 
>> should something surface that needs support?
>>
> It works and builds fine including unit tests and doesn't run into any 
> conflicts with Python. Given the general adoption I think its good way 
> to lightly start having something in place because we'll likely need it 
> further down the road.
> 
> https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=290774

I cannot shake off the feeling we're barking up the wrong tree here - 
sidestepping the build system isn't something we "need", it is *one* way 
to solve a certain subset of problems, assuming the framework were 
carved in stone, which it is not.
   And it does not scale well at all and piles up technical debt because 
for every issue we see where we want to discuss with upstream we have 
then added the additional burden onto ourselves to sift and filter is 
the issue there because we did change the build system.  Only then can 
we sensibly talk to upstream maintainers or file bug reports.

-- 
Matthias Andree
FreeBSD ports committer