CI is failing, probably because the InterpPass option isn't set when running the tests in main_test.go. One simple way to fix this is to also allow the empty string to indicate "all" in compileopts/config.go, or by reversing the logic:

func (c *Config) InterpProgram() bool {
	switch c.Options.InterpPass {
	case "package", "none":
		return false
	}
	return true
}

Thank you for the feedback @aykevl. I've update the code.

Any feedback on this now @aykevl?

On the subject of configuring interp, being able to customize the timeout value (currently one minute) would also be useful.

So I still don't think this change is correct in general. Some packages absolutely require the interp pass to run for correct functioning. Although, maybe this is only true for the runtime package and not for other packages.

One thing that would be correct is to disable the whole program interp pass (interp.Run), that one can I think be safely disabled. Maybe that already solves the biggest issues?

being able to customize the timeout value (currently one minute) would also be useful.

I guess? But I think that would only be for debugging. If it runs more than a few seconds I'd argue that there must be a bug somewhere (even if it's just a performance bug). So, not sure about this but not against it either.

The idea behind this PR was to disable the interp passes for debugging, and not to generate fully functional code.
In other words, this is purely for debugging and should not be used for anything else.

Would you be more confortable if we prefix the parameter with -internal as you suggested in #2003 (review) or should I go ahead and remove the package option, only leaving the program option?

Thinking a bit more about this, perhaps one piece of information missing here is the fact that this was meant to be used together with -printir to capture the generated IR without the interp (package) pass.