Module expert needed

[Paul Richard Thomas]

Dear FX,

Yes, I should have read the manual to find -fintrinsic-modules-path!
The patch is looking OK.

On the submodule files, I have agonized about that. It would be
possible, at the expense of more complexity, to only write the private
entities to the .smod file. Or, I could introduce a flag
-fno-submodules. I guess that I would go for the former.

Cheers

Paul

On 6 August 2015 at 13:03, FX <fxcoudert@gmail.com> wrote:
> Hi Thomas,
>
>> I used local copies of the IEE modules because f951 was not finding
>> them. In this case, the condition in read_module that permits the
>> setting of current_intmod is false. Once corrected, I do not find any
>> problem with the values of sym->from_intmod.
>
> The driver passes -fintrinsic-modules-path to f951, which is how we know where to load intrinsic modules from (and how we detect that they are, indeed, intrinsic modules). So yes, for debugging f951 with IEEE modules present, you need to manually pass that. And yes, I forget every time, at first.
>
> I’ve stop trying to get the generic symbols marked, as I can’t find a way to do that reliably. Checking the specifics instead, as I did in the attached patch, feels a bit roundabout, unless I’ve missed a more simple way to do it (this is not the area of the compiler I know best).
>
> Thanks for your advice and help. I should have a formal patch to submit soon on the use of IEEE intrinsics in constant expressions.
>
> Cheers,
> FX
>
>
> PS: I have not entirely followed the recent developments on the submodule work, but thanks. One quick question: is it normal that we write two-almost identical files as .mod and .smod in the current state of things? Even when submodules are not used?
>



-- 
Outside of a dog, a book is a man's best friend. Inside of a dog it's
too dark to read.

Groucho Marx