Better error intan when missing files in the "one-file-per-signal" type.

[h-mayorquin]

From the discussion on #1693, @zm711

[h-mayorquin]

Maybe there are some missing ones that are acceptable. But I think we can start here and be explicit which is better.

[h-mayorquin]

So we have some tests that don't have digital output and supply voltage.

What else is possible not to have?

[zm711]

I think this is a tough check to make carefully. For one-file-per-channel you can basically shut anything off. For one-file-per-signal that are certain combination that have to show up together but you could shut off dig in /dig out, aux channels. So I think don't think we have a realistic core. I think we should pull the info from the header and whatever the header says is enabled we raise an error no? We do that when we read the header. Most things have a boolean check to see if that signal is enabled?

[zm711]

Look at line 1185 for the check I'm referring to. We see that the signal is enable. Then we could use that information for the formats I think. Maybe you could try an implementation with that instead?

[h-mayorquin]

For one-file-per-signal that are certain combination that have to show up together but you could shut off dig in /dig out, aux channels. So I think don't think we have a realistic core.

Yeah, this is the case that I want to cover. one-file-per-channel seems too complicated at the moment but might do it later.

I think we should pull the info from the header and whatever the header says is enabled we raise an error no? We do that when we read the header. Most things have a boolean check to see if that signal is enabled?

Yes, this is the way. Thanks.