> -----Original Message-----
> From: Richard Purdie <richard.pur...@linuxfoundation.org>
> Sent: Tuesday, February 28, 2023 4:04 AM
> To: michael.opdenac...@bootlin.com; Tom Hochstein
> <tom.hochst...@nxp.com>; OE-core <openembedded-
> c...@lists.openembedded.org>
> Cc: YP docs mailing list <d...@lists.yoctoproject.org>
> Subject: Re: [OE-core] [DISTRO|MACHINE]_FEATURES_BACKFILL_CONSIDERED:
> confusing variable names?
>
> On Tue, 2023-02-28 at 10:52 +0100, Michael Opdenacker via
> lists.openembedded.org wrote:
> > Greetings,
> >
> > Thank you all for your replies and comments!
> >
> > On 24.02.23 at 00:24, Tom Hochstein wrote:
> > > I agree. The problem I have with these variable names is that it's
> > > ambiguous
> whether the features will be included or excluded from the backfill. All you
> can
> tell from the names is that the values were 'considered'. What does that mean?
> You have to go to the manual to find the full meaning.
> > >
> > > What's important about the variables is that they will be 'excluded' from
> > > the
> backfill, so this does look much clearer to me:
> > >
> > > DISTRO_FEATURES_BACKFILL_EXCLUDED += "some-feature"
> >
> >
> > Richard, would you accept it if I sent a patch replacing
> > "BACKFILL_CONSIDERED" by "BACKFILL_EXCLUDED" in the two variables?
> >
> > Of course, I can understand if that's too late for the next release, and
> > instead document this oddity, at least before we can make a change again.
>
> I don't want to rush into this. The idea of the variable was that the
> list were things that distro maintainers would need to consider when
> upgrading and then "mark as considered" so in that context, excluding
> them makes less sense. I'm sure there was discussion about the name at
> the time.
Not sure I understand. When you say 'idea of the variable', are you referring
to the BACKFILL variable? I assume so since that seems to be the one that the
distro maintainers will 'consider', right? After they've 'considered' the
BACKFILL, then they will add the ones they _don't want_ to
BACKFILL_CONSIDERED(_AND_NOT_WANTED). The ones they _do want_ are not marked.
>
> I'm nervous about renaming variables in general as it does create pain
> for users upgrading and some community feedback has been that we're
> bordering on frustrating users. I know we have mechanisms now to help
> but it isn't pain free.
This I totally understand 😊
>
> > > Is there a good reason why we have this separate variable and can't just
> > > do
> this?
> > >
> > > DISTRO_FEATURES_BACKFILL:remove = "some-feature"
> >
> >
> > Good question. Maybe because ":remove" is final, and cannot be
> > overridden, if I understood correctly?
>
> Right, the concern would have been it couldn't be undone and the distro
> setups from when this dates from were more complex.
Since every variable is so afflicted by :remove, perhaps this isn't a good
argument, particularly when there is a more general pattern for this: to add an
extra variable at the point of the removal:
MY_DISTRO_FEATURES_BACKFILL_REMOVALS ??= "some-feature"
DISTRO_FEATURES_BACKFILL:remove = "${ MY_DISTRO_FEATURES_BACKFILL_REMOVALS}"
>
> Cheers,
>
> Richard
>
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#177856):
https://lists.openembedded.org/g/openembedded-core/message/177856
Mute This Topic: https://lists.openembedded.org/mt/97187180/21656
Group Owner: openembedded-core+ow...@lists.openembedded.org
Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub
[arch...@mail-archive.com]
-=-=-=-=-=-=-=-=-=-=-=-
