Guidelines for additional output parameters

[bernt-matthias]

In a recent (short) discussion (#4006 (comment)) we were wondering what's the best way to choose which outputs a tool should produce:

• multiple booleans
• one select

Wondering if we should try to standardize this in the IUC guidelines?

For me a select is favorable for the following reasons:

• its more compact (except for one output)
• one can contrain them, e.g. requiering at least one (min="1") or requiring exactly one (type="radio")

Here is how the two options look:

Maybe add thumbs up/down reactions to the issue if you support/not support standardization... or add comments with arguments.

[bgruening]

You can move this to https://github.com/galaxy-iuc/standards/issues if you like.

[bgruening]

I prefer the multi-select.

[wm75]

One disadvantage of the select is that you cannot provide separate help/command line args for the options (as visible in the screenshots).

[wm75]

Also, restrictions on the select can only be enforced in the command section (after the user hits Execute). So it's not very clear what the allowed combinations are.

[wm75]

It's also easy to abuse the one select box like e.g. here:

Clearly the VCF or BCF output is the main output of Delly, while the other two are optional outputs. The single select box doesn't convey that at all.

[bernt-matthias]

Closing this here galaxy-iuc/standards#66 .. Also copied the comments