[fedora-arm] ExcludeArch tracker doesn't appear to be effective

[Matthew Garrett]

On Tue, Jun 10, 2014 at 09:49:58PM +0100, Peter Robinson wrote:
> > What's depressing is the trend, not the absolute count. I'd expected it
> > to head rapidly towards zero after the first release, but instead it's
> > still growing.
> 
> Is it? Where's your proof? From the patches and dealings with it
> personally that's not my understanding and if it is the case it's not
> due to the ARM team but because packagers aren't following the
> packaging process.... and in this case we actively fix them when we're
> made aware of the incident.

In the past 6 months, 6 bugs added, 2 bugs closed - 
https://bugzilla.redhat.com/show_activity.cgi?id=485251 .

> > Anyone who has a usecase that relies on one of those packages will have
> > an inconsistent experience if they attempt to reproduce it on ARM.
> > That's harmful. It makes us look bad. It gives the appearance that we're
> > willing to release a worse product simply in order to claim ARM support.
> 
> And if they engage with us about that experience we do our best to
> deal with them where possible. There's a few cases where I'm aware of
> that we don't package because the HW is actively not supported on ARM
> or similar style cases. In those cases I would argue that it's better
> not to build the packages as arguably no experience is better
> experience than a bad experience especially if there's potential of
> issues that offer a worse experience, hardware breakage or data
> corruption. The fact is that the x86 and ARM use cases don't match up
> 100% and I don't see the point in trying to glue those together 100%
> for the sake of a check box.

Where there's reliance on specific hardware functionality that's absent 
then yes, absolutely, there's no benefit in building the packages. 
Figuring out how to communicate that to users isn't an entirely solved 
problem, but with luck nobody's actually buying ARM hardware with 
unrealistic expectations of its functionality.

But I can't find any examples of those in the tracker. They all seem to 
be cases where packages are either missing dependencies, take too long 
to build or are just missing support. They're code. We can fix them.

> > I don't think the current state of the ARM port is good enough. That's
> > not a reflection on the people involved. That's not a criticism of the
> > amount of effort that's been made. I just want to know how we can get
> > from where we currently are to where we want to be.
> 
> Well why didn't you say that from the start rather than coming in and
> bullying the people actively involved and make them feel like the
> massive effort myself and many others have been putting in is useless
> and a waste of time. Don't be a Magpie Board Member and fly in and
> shit over everything and than fly awau with no concept of what's
> happening below. Every time you've had any attempt at opinion that's
> the way you've done it and all you do is get all our backs up and make
> the problem worse rather than better.

I'm genuinely sorry if I gave the impression of bullying here. I want to 
feel comfortable pointing at the ARM port as an equal to the x86_64 one. 
I don't feel entirely comfortable doing so at the moment, and the 
current process doesn't seem to be getting us to the point where I would 
be.

> >  Individual package
> > maintainers seem happy to just add an ExcludeArch, maybe file a bug
> > against upstream and then forget about the issue. Given a lack of direct
> > incentive for them to care about ARM, that's not terribly surprising.
> > What can we do about that? Is the only realistic answer to find the
> > resources to have a team to hunt down and fix portability issues that
> > are sufficiently far from the core that the existing ARM community can't
> > justify the time? And if so, is there any way we can make that happen?
> 
> I'm not sure I agree with your outline here, you've given no real
> concrete examples here. I agree there's no real incentive but there's
> over 15,000 source packages in Fedora and there's less than 100 (last
> time I looked, unfortunately there's no easy way off checking this
> without downloading the entire src.rpms or checking out all 15K git
> repos) or so excluded from ARM and the vast majority of those are due
> to HW support. There's some like D where upstream has yet to port the
> stack. I'm sure there's others I'm unaware of but it's not because of
> the ARM team but rather the packager following procedures or engaging
> us for assistance.

The quantity of the archive that's built and working on ARM so far is a 
testament to the amount of effort that the ARM community have put into 
this port. The question is how to finish that. All I'm saying here is 
that the current approach of filing bugs doesn't appear to be resulting 
in people actually fixing their packages. It's unreasonable to expect 
you to do all of it. So what do we do?

-- 
Matthew Garrett | mjg59 at srcf.ucam.org