Mike Lin

BTW, I showed the python example just to clarify the distinction, not because I think we necessarily have to follow its model. I don't have a big problem with making...

I'm sensitive to the inconvenience caused by enforcing the distinction; namely, to have optional inputs that are passed through from workflow to task, to apply a default one would either...

Opened #464 to clarify in the spec. It would be good to confirm that the proposed clarification matches what Cromwell actually does...

cf. https://github.com/broadinstitute/cromwell/issues/5540 https://github.com/openwdl/wdl/issues/371

I strongly support, at the least, a statement that tasks "[SHOULD](https://datatracker.ietf.org/doc/html/rfc2119#section-3)" treat their input files as read-only. But it may be challenging to require engines to enforce this -- that...

@aofarrel I agree with your points *when we assume* the engine already has to copy/localize the input file(s) before starting the task. However, both miniwdl and dxCompiler have the ability...

miniwdl supports v1.1 with some errata documented [here](https://miniwdl.readthedocs.io/en/latest/runner_reference.html#wdl-interoperability). I'd be hesitant to direct users to 1.1 by default without at least a timeline from Cromwell, maybe it's a chicken and...

@illusional thanks! I'm happy to get this clarified too. I'd love to hear from somebody who has poked this specific behavior in Cromwell to see if it's captured correctly here,...

@jdidion the pickle we're in (see the linked prior discussions) is that this > `call foo { input: x = select_first([x, 0]) }` isn't a good general solution because it...

I'd def rather find a better future approach, but, I personally just haven't seen the others discussed so far as better -- each one is a compromise. The abbreviated passthrough...