Re: [PATCH][gdb/testsuite] Handle unrecognized command line option in gdb_compile_test

[Joel Brobecker <brobecker@adacore.com>]

> [ To reiterate my earlier point, it's counter-pattern to generate a FAIL
> for something that's not a problem in the tool-under-test.  So much
> counter-pattern that I had to resist the reflex to commit this as obvious. ]
> 
> Indeed, there could be an error in the testcase.  Also, there could be
> no error in the test-case.
> 
> In the former case, having a FAIL instead of UNSUPPORTED could increase
> the change of noticing this, agreed.  In the latter case, having a FAIL
> instead of UNSUPPORTED when testing an older compiler, is an annoyance
> which drowns out FAILs that do need attention.
> 
> So, I'm not saying that noticing test-case errors causing compilation
> problem is not worthwhile.  But I'm saying that turning the compilation
> errors into FAILs is the wrong solution.

I see what you mean. I don't really have a strong opinion on this.
Speaking for myself, I tend to blindly accept UNSUPPORTED results,
so I am unlikely to notice incorrect ones, which suggests that, for me
at least, using UNSUPPORTED is no better. But then again, the way
I have being doing testing in the past is by "diff-ing" the testsuite
report before and after my change -- so it's not like I would notice
a FAIL better ;-). So it's also be no worse (again, just speaking for
myself, as YMMV in this case).

-- 
Joel