Hi everyone, I've been working on adding

topics

to the template and I think I've found a nice way to implement this without breaking much: So before we had something like this:

output:
...
path('versions.yml'), emit: versions

What I would make it is something like this:

output:
...
tuple val("${task.process}"), val('tool1'), eval("tool1 --version"), topic: versions, emit: versions1
tuple val("${task.process}"), val('tool2'), eval("tool2 --version"), topic: versions, emit: versions2

So I think having both the

topic

and the

emit

is the nicest way to go at this. The main advantage is that we can easily test the versions output by using the

emit

names. It also ensures that the

meta.yml

creation tool doesn't break on this and actually documents the whole versioning process. What do you think of this solution? (You can see a POC implementation in this PR: https://github.com/nf-core/tools/pull/3779)

how about we do:

output:
...
tuple val("${task.process}"), val('tool1'), eval("tool1 --version"), topic: versions, emit: versions
tuple val("${task.process}"), val('tool2'), eval("tool2 --version"), topic: versions, emit: versions2

Yes I agree and we don't need versions in subworkflows, it would kind of remove the advantage of using topics if we kept using the regular outputs

do you think it would be bad if we named the channels

versions_tool1

,

versions_tool2

, etc? I don't love having numbers in the channel names

Oooh I like this!

I am also not sure about the

meta.yml

, I think it is better to add a new section

topics

, at the same level of

ìnputs

and

outputs

(while keeping the versions output)

For me we shouldn't put too much effort into naming this as it should really only be temporary until a newer nf-test handles topics properly

Good idea on the topics section!

I think the topics section was your idea originally 😄 but it makes the most sense if we want to be able to replicate the main.nf from the information in meta.yml

Yeah true, I'll see how much work it is to add that in 😛