I am having some issues regarding CI tests running for hours and then being cancelled due to exceeding the time. I am not sure what the issue might be. I tried to trigger some of them individually but they still fail: https://github.com/nf-core/differentialabundance/pull/705 The PR only updates modules/subworkflows from nf-core in the diff abundance pipeline. Can I get some hints on how to find the root cause? Thanks!

can you try running the test in a github codespace?

Yes, let me try that and see what happens

This is the same issue @Matthieu Muffato came across, with processes being both consumers and contributors to the

versions

topic. Basically we need to prevent any process downstream of the topics aggregation from also emitting to the topic. In this case the QUARTONOTEBOOK and MAKE_REPORT_BUNDLE processes. I think you're basically going to have to patch those modules in the pipeline to achieve that.

That makes sense. I'll check that.

Pretty sure that's it, just had an adventure with Claude 🙂

Indeed, that was the cause. I've patched the modules for now. I wonder what the long-term solution is? What are people doing in other pipelines?

I don't see a way around a solution like this if we keep using the topic mechanism. MultiQC doesn't emit a versions topic at all, but we can't do that for modules that aren't always terminal processes in pipelines so patching is probably the thing for now.