On Fri, 2023-01-06 at 10:19 +0100, Alexander Kanavin wrote:
> I understand where it comes from, and that's not what I'm trying to
> express. Why do we need to insert those REQUIRED_DISTRO_FEATURES lines
> into every recipe that, directly, or indirectly depends on an opengl
> implementation, even when the component itself uses nothing from the
> opengl stack?
> 
> With gtk4 becoming one of those items, this really does become a
> maintenance burden.
> 
> What would happen if we instead drop them all, and keep
> REQUIRED_DISTRO_FEATURES only in the opengl provider itself, which
> would be mesa?

There is a bitbake 'limitation' where these things don't automatically
flow through dependency chains. The net result of that is that if you
just do this in mesa, "bitbake world" would show errors that XXX can't
find dependency mesa or opengl.

The first thought is, 'lets just make bitbake flow them' however the
trouble is bitbake can't know when this is appropriate and when it is
not.

For example, imagine the dependency is "virtual/libc" and somehow you
break the providers of it, i.e. break the glibc recipe. All of a sudden
"bitbake world" would return success even though it built nothing since
anything depending on virtual/libc was magically removed (which is
nearly everything at some point).

The conclusion I came too last time I thought about this was they we
really did want to mark up the cases where things have a specific
requirement, even if at times that is painful.

Cheers,

Richard
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#175571): 
https://lists.openembedded.org/g/openembedded-core/message/175571
Mute This Topic: https://lists.openembedded.org/mt/95842308/21656
Group Owner: [email protected]
Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub 
[[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-

Reply via email to