Make lttng-ust an optional dependency.

[tmds]

Closes #107406.

@jkotas ptal.

cc @dotnet/distro-maintainers

[jkotas]

cc @dotnet/dotnet-diag

[tommcdon]

@noahfalk

[noahfalk]

The principal that the build allows disabling lttng looked fine to me. I'll defer to @ViktorHofer and @jkoritzinsky on the specific --with-system-libs mechanism for signaling that.

I'm interested in renaming the feature flag to something like FeatureLttng, but thats probably better kept separate so that the change in the build isn't obscured by renaming.

[tmds]

The principal that the build allows disabling lttng looked fine to me. I'll defer to @ViktorHofer and @jkoritzinsky on the specific --with-system-libs mechanism for signaling that.

We discussed and concluded that using --with-system-libs is ok for this PR and it is desired to replace it with the a dedicated feature flag from the vmr build.sh script.

[tmds]

@noahfalk or @jkotas , can you merge this please?

[tmds]

@noahfalk or @jkotas , can you merge this please?

Friendly reminder

[noahfalk]

I re-ran some tests because the build analysis reported that it had lost contact with one of the agents. I don't want to merge the code until build analysis is green.

[noahfalk]

I just rebased to pull in changes from #114268 which will hopefully address the networking test timeouts. If that still doesn't fix the CI issues then I'll ask you to take a closer look at the failures @tmds. I don't want merge a not passing PR.

[tmds]

Looking at https://github.com/dotnet/runtime/pulls, I doubt we'll see a full pass. I'll take a look at the CI results when this run is complete.

[noahfalk]

Not every leg has to pass, just the 'Build Analysis' one. Build analysis does some work to try triaging known failures so its typical that GH will show a red X on the PR even though Build Analysis determines the PR is OK. As an example: #114459

[tmds]

Build Analysis passed!

[noahfalk]

Done! :)