On Wed, 2026-01-28 at 20:02 +0000, Weber (US), Matthew L via 
lists.openembedded.org wrote:
> I was curious where the Yocto binary distro work ended up?  I was
> looking at Michael's email thread and talk.
> 
> https://lists.openembedded.org/g/openembedded-architecture/topic/102755562#msg1846
> 
> https://rootcommit.com/pub/conferences/2024/elce/yocto-binary-distro/yocto-binary-distro.pdf
>   (https://www.youtube.com/watch?v=UdJZ0xz-QLk)
> 
> I've looked through parts of the wiki
> (e.g.https://wiki.yoctoproject.org/wiki/Binary_Distro_Process ).  
> 
> Does my understanding as follows sound correct?  Yocto doesn't
> provide a binary distro, so we first build with Yocto from source,
> e.g., a small generic x86 kernel and then setup a feed of kernel
> module pkgs and pkg world.  Can I assume a user's only option to
> build a system, would be to boot a system to use the feed "online" to
> make a rootfs have everything they need? Or are there more options
> for composing offline so the end users grabs a kernel and a script
> that pulls a rootfs together offline from emulation/hw/yocto
> env?(debootstrap like)
> 
> Thank you for any insight on this so we can find the right starting
> point and list of open issues (if that exists)!

You are correct that we don't provide binary packages. We've opted to
provide binaries in the form of sstate instead. We've never been
totally convinced that binary packages make sense for the project given
the ways people want to tweak the targets they build for.

We're also not entirely convinced that a traditional package manager
approach is always a good way to deliver or update what we produce.
Yes, internally we use package managers to build our images but that
isn't the only way end users update or maintain them.

We do already have tools like the esdk which can assemble images,
however these haven't seen widespread adoption.

I therefore think the right question might be, "what do you want to use
such a thing for?". I'd like to see how we can support user usecases
but I'm not sure we understand which usecases we should be aiming to
support.

Once we understand the usecases we need to support, that naturally
leads to the list of issues we need to address. There are lots of these
in bugzilla for example but I think the plan/usecase needs resolving
next to move forward.

Cheers,

Richard
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#2227): 
https://lists.openembedded.org/g/openembedded-architecture/message/2227
Mute This Topic: https://lists.openembedded.org/mt/117514584/21656
Group Owner: [email protected]
Unsubscribe: https://lists.openembedded.org/g/openembedded-architecture/unsub 
[[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-

Reply via email to