[Fedora-packaging] Proposal to reduce anti-bundling requirements

Haïkel hguemar at fedoraproject.org
Fri Sep 11 22:19:54 UTC 2015


2015-09-11 23:37 GMT+02:00 Dominik 'Rathann' Mierzejewski
<dominik at greysector.net>:
>
> It's not an improvement as such. We already require the Provides:
> bundled(foo) thing, though we rely on the good will of maintainers
> (both in Fedora and upstream) because we have practical means of
> enforcement.
>

It's currently not enforced and as some maintainers bypass the current policy,
it's not gonna change.

>
> Which SIGs? FPC is the the right general SIG for this. I guess you
> meant something along the lines of: Perl SIG would grant/deny exceptions
> for perl-* packages, Python SIG would do the same for python* packages,
> and so on. That... isn't entirely a bad idea IMHO, but I think it'd
> have to be tracked anyway.
>

+1 for FPC
But analyzing pros and cons of unbundling takes a lot of time, so if
we could allow some
SIGs. ie: python, perl, big data to do overrides FPC on a limited and defined
set of packages, it'd scale better.

Off course, SIGs would need blessing from fesco or FPC, or it
would be to easy.


> What if one provenpackager grants an exception and another disagrees?
>

 Good point, if we ended going that way, it would be formal rules.


>> If you disagree, please come up with a counter-proposal even if it
>> means dropping half of Fedora packages.
>
> Please don't speak in unsubtantiated hyperboles. If you have specific
> numbers detailing the extent of bundling in existing packages then
> please present them.
>

I have no specific numbers, but I regularly find bundled code in existing
packages. Latest was CFEngine bundling code from RPM ...
Some modules of interpreted languages fork base libs and ship them too.

Some critical packages like Puppet regularly requires shipping bundled
libs. Some packages like Hadoop are stuck to very old releases, as unbundling
libs is a neverending task. Upstream even advise users not to use
Fedora packages.

We could ignore them but we'll be less and less relevant and our influence
on upstream projects to bring good practice will fade.
A vibrant Fedora that can influence upstream to improve is better than
ignored Fedora


>> Keeping status quo is at best diverting our eyes from bad practices,
>> at worst, hypocrite.
>
> Not at all, IMHO. The FPC is dealing with cases of bundling as they're
> brought to our attention. I believe there are enough cases that are not
> straightforward to decide to warrant analyzing them closely and voting.
> As you've probably noticed, not everyone on the FPC has exactly the same
> stance on bundling, though we all agree that it's bad in the general
> case.
>
> Regards,
> Dominik

I'm not criticizing FPC here, you're doing a great job while having a
lot pressure
on your shoulders.
It's rather about people who refuse to discuss that topic.

In a decade, the world has changed, in a good or bad way, not up to me
to decide.
I agree that bundling is a terrible practice, but we need to weight
pros and cons
and accept that we may change our policies.


> --
> Fedora http://fedoraproject.org/wiki/User:Rathann
> RPMFusion http://rpmfusion.org
> "Faith manages."
>         -- Delenn to Lennier in Babylon 5:"Confessions and Lamentations"
> --
> devel mailing list
> devel at lists.fedoraproject.org
> https://admin.fedoraproject.org/mailman/listinfo/devel
> Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct


More information about the devel mailing list