ci: changes for core24 snaps to build

[soumyaDghosh]

This PR makes some necessary changes to make core24 snaps build with remote build.

1. disables force-fallback
2. restricts other architectures

[jnsgruk]

Will take a look tomorrow - this logic is getting more complicated than I'd like by now!

[soumyaDghosh]

Please take your time.

Will take a look tomorrow - this logic is getting more complicated than I'd like by now!

Disabling fall-back isn't necessary, I just did it as I found the scope. (The new backend is very fast and feels very nice to see 🥲 ) But restricting the architectures are very necessary as needed by the core22 snaps before.

[jnsgruk]

I'm a little confused because the architecture restriction should happen through the use of the --platforms arg which is added on the next line. I think I'm going to hold on this, as we're still ironing out a couple of bugs in the new builder.

As far as I know, the current code works for core24 - there is an example run here, with the corresponding snapcraft.yaml.

Could you point me to where this fails for you?

[soumyaDghosh]

I'm a little confused because the architecture restriction should happen through the use of the --platforms arg which is added on the next line. I think I'm going to hold on this, as we're still ironing out a couple of bugs in the new builder.

As far as I know, the current code works for core24 - there is an example run here, with the corresponding snapcraft.yaml.

Could you point me to where this fails for you?

https://github.com/soumyaDghosh/webkitgtk-sdk/actions/runs/9530513048/job/26270468434

https://github.com/soumyaDghosh/webkitgtk-sdk/actions/runs/9530526618/job/26270504427

See, it's not working

[jnsgruk]

I don't think that's because of the platforms definition. @mr-cal - any ideas (I'm afk and only have my phone for a few days!)

@soumyaDghosh can you reproduce locally?

[soumyaDghosh]

@soumyaDghosh can you reproduce locally?

Yes, I first tested this locally and then went to try the ci.

[lucyllewy]

This might be an issue with your authentication token not having the full ACL required. The LaunchpadError looks like it is trying to issue an API call which is failing. The first thing is to confirm which ACL entries you need on your authentication token and regenerate it to ensure you have a valid token with the appropriate ACL to see if that fixes it.

[soumyaDghosh]

Nope tried with new secret. Still the same

https://github.com/soumyaDghosh/webkitgtk-sdk/actions/runs/9533180559/job/26276132086

[soumyaDghosh]

@jnsgruk Builds once again failed here

https://github.com/soumyaDghosh/ffmpeg-sdk/actions/runs/9548427970/job/26315614173

[mr-cal]

I'll try remote building your snaps locally and see if I can reproduce.

If I can't, updating the CI job to set CRAFT_DEBUG=1 and --verbosity=trace in CI or cat'ing the full snapcraft logs would be very useful.

[mr-cal]

@soumyaDghosh - can you try one of the debugging strategies I mentioned above?

I've tried a few variations of remote-builds of webkitgtk-6-gnome-2404-sdk and I'm run into this known bug but I can't reproduce the errors you are seeing. They look more like launchpadlib api-related issues.

[jnsgruk]

I think this is ready to be merged - this is a small change that only realistically affects a small portion of our snaps, some of which are blocked by the change. I'll merge this, and we experience problems it can be reverted.