RE: [OE-core] [PATCH 1/3] insane: Improve patch warning/error handling

[Peter Kjellerstedt <peter.kjellerstedt@axis.com>]

> -----Original Message-----
> From: openembedded-core@lists.openembedded.org <openembedded-
> core@lists.openembedded.org> On Behalf Of Richard Purdie
> Sent: den 21 januari 2023 00:01
> To: Alexander Kanavin <alex.kanavin@gmail.com>; Bruce Ashfield
> <bruce.ashfield@gmail.com>
> Cc: Ross Burton <Ross.Burton@arm.com>; OE-core <openembedded-
> core@lists.openembedded.org>
> Subject: Re: [OE-core] [PATCH 1/3] insane: Improve patch warning/error
> handling
> 
> On Fri, 2023-01-20 at 20:38 +0100, Alexander Kanavin wrote:
> > On Fri, 20 Jan 2023 at 20:29, Bruce Ashfield <bruce.ashfield@gmail.com> wrote:
> > > Because I'm simply not going to insist on it in all the patches. I
> > > need all the contributions I can get, and I'm not going to
> > > pedantically insist on that.
> > >
> > > meta-virt is not oe-core, I do the lifting. Therefore, if bitbake
> > > errors, I have to fix it.
> >
> > But you do not need to insist on the needed metadata or fix it after
> > the fact. Bitbake will do the insisting for you, when contributors
> > test the change locally *before* they send it to you. If bitbake
> > errors on your side, this means they never built their contribution,
> > and you should raise a concern for that reason, and not for the
> > missing metadata.
> 
> It isn't that simple since this is a configurable QA warning, all it
> takes is one layer/distro to disable it and it is disabled for all
> layers that user works on.
> 
> This is why "core" is a separate config to "noncore" but we can't have
> a config for every layer and even if we did, people would still turn it
> off.

Rather than having separate QA tests for "patch-status-core" and 
"patch-status-noncore", couldn't we have a single "patch-status" and then 
configure it using a separate variable that specifies the layers that 
require the Upstream-Status trailer? Then each layer with this requirement 
can add itself in its layer.conf file and thus it is up to the maintainer 
to decide whether they want it or not.

> If it is turned off, it means people send patches and Bruce has to fix
> them, or ask them to resubmit which is extra overhead to the
> maintainer.
> 
> I've been thinking about this and if I do make it the default, it will
> mean warnings show up on other CI systems and layer maintainers will
> get patches or complaints about the warnings. I'm not sure I really
> want to get into this.
> 
> I do think it is something the project should be doing but I don't want
> to burn out our existing maintainers. Since there isn't wide community
> buy in, I suspect I should just drop the idea.
> 
> Cheers,
> 
> Richard

//Peter