core: Add composefs digest for local commits too #4594
Conversation
|
Needed for ostreedev/ostree#3021 |
cgwalters
force-pushed
the
client-composefs-meta
branch
from
eb1736f
to
48db662
Compare
cgwalters
added
needs-work/failing-ci
cgwalters
force-pushed
the
client-composefs-meta
branch
from
48db662
to
7be0056
Compare
|
Always nice when the CI legitimately fails and you know your CI is working. |
cgwalters
force-pushed
the
client-composefs-meta
branch
from
7be0056
to
1dee2d4
Compare
|
OK at some point I'll actually debug this test locally... |
cgwalters
force-pushed
the
client-composefs-meta
branch
from
1dee2d4
to
85047d6
Compare
|
Oh god of course the problem is that somewhere in the giant manual process of building ostree for 3 active Fedora releases I got side tracked and only did rawhide... |
cgwalters
force-pushed
the
client-composefs-meta
branch
from
85047d6
to
246b2cc
Compare
|
This will now just need to sit here, blocked while we wait ~a week for no one to add karma to https://bodhi.fedoraproject.org/updates/FEDORA-2023-13b41d2e37 until it times out and ships... |
|
why is |
I stopped adding it because sometimes it got 3 karma from people who weren't really testing things, although that seems to have stopped thankfully. (This is a complex topic; more ideally we gate on some FCOS tests and really upstream tests, but at least there is the silverblue test there now...) |
For the composefs metadata API.
This pairs with coreos@dbe7821 where we generate composefs metadata for base images. It's going to be useful to do it for client side builds too, so that we can treat them symmetrically. It doesn't have a very high cost, just a metadata traversal.
cgwalters
force-pushed
the
client-composefs-meta
branch
from
246b2cc
to
c115600
Compare
|
OK, this should now work since the new ostree went stable. |
|
/retest |
|
Blocking on coreos/fedora-coreos-tracker#1579 |
|
/retest |
This pairs with dbe7821 where we generate composefs metadata for base images.
It's going to be useful to do it for client side builds too, so that we can treat them symmetrically. It doesn't have a very high cost, just a metadata traversal.