Ben Sherman

@adamrtalbot check out the rnaseq PR that I linked above, it is less verbose than what you did. The syntax has evolved somewhat since Paolo's original proposal. Even so, the...

In the Nextflow PR there are some [docs](https://github.com/nextflow-io/nextflow/pull/4784/files#diff-275884dd33d2ba55282992b0ab18894b8e8b0636252614af64394b77750575ee) which explain the feature in more detail. Unfortunately the deploy preview isn't working so you'll have to look at the diff >...

The current prototype simply maps the output channels to the publish directory structure, but we still need to get these outputs to the top level whereas currently they are nested...

I managed to move everything to the top-level workflow, so it should be executable now (though there are likely some bugs, will test tomorrow). I ended up using topics for...

We can definitely move some of these topic mappings into the modules and subworkflows, that was going to be my next step. I also suspect that nf-core will be able...

I moved most of the topic mappings into their respective subworkflows. It gets tricky when a workflow is used multiple times under a different name and with different publish behavior....

Notes on the latest update: - Topics are no longer used. Nextflow simply maintains a global map of channels to "rules" under the hood - The output DSL is no...

> Would it be feasible at some point to use some optional dynamism in the modules, to facilitate repeated usage? Maybe in a future iteration. But related to this, we...

The redirect to `null` simplifies the top-level publish def somewhat. The remaining rules could also be moved into the workflow defs since they only rename paths. It just might be...

I don't think the inspect command needs to know the work directory