Testcases, was Re: [Patch, fortran] PR29786

[Brooks Moses]

Bernhard Fischer wrote:
> On Sat, Jun 09, 2007 at 09:56:28AM +0200, Paul Thomas wrote:
> Thanks alot for taking care of this!

Seconded.  :)

>>>> +  subroutine int4_int4
>>>> +      integer(4)         NUNITS(4)
>>>> +      integer(4)         o
>>>> +      equivalence (o,nunits(3))
>>> "o" seems like a singularly poor choice for a variable name, and 
>>> "NUNITS" looks like it should mean something but it doesn't.  Could 
>>> these perhaps be A and B instead?
>> 'tis the reporter's choice - I take your point and will change it.
> 
> It was the name that was used in the COLUMBUS package (i should have
> noted this in my report, sorry). I just reduced it and didn't bother
> to change the names too much. Unfortunately i do not not have the
> time to see if this app compiles and works as expected now..

Certainly fair enough.  IMHO, there's no point in regularizing names and 
clarifying code when it's reported as PRs, but I think it's useful to 
add some editing between there and the testcase in the testsuite, to 
make things clearer and easier to read out of context.

I'm less certain about whether it's useful to simplify the code itself 
between the PR and the testsuite.  The code in PRs often still has 
unnecessary weirdness aside from the actual problem at hand.  On one 
hand, that weirdness means that we get a testsuite that's a patchy 
collection of random code that happens to cover a lot of things (but is 
difficult to look at and see if it covers "everything" or not); on the 
other hand, the weirdness means that extra stuff is tested, including 
some things that we wouldn't otherwise think to test.

- Brooks