Problems with intrinsic modules

[zbeekman]

Hi Stefano,
I’m running into this error:

Attention: the file ‘src/tests/test1.f90’ depends on 'iso_c_binding' that is unreachable

And the build is failing. However, iso_c_binding is an intrinsic module and does not require any source file to be built, same with iso_fortran_env and ieee_…

Is there a way to tell FoBiS.py not to try to build the intrinsic module, and not to fail if it can’t seem to find it?

[jacobwilliams]

Here's an idea: if FoBiS determines that there is a module it can't see, just go ahead and compile everything anyway. Then, if it's really a problem, it will fail when compiling (with the normal compiler error message about not being able to locate a module with the given name). That way, you could account for this situation without having to have a list of the intrinsic modules somewhere in the code.

(Maybe in verbose mode it could print out the list of things it can't find).

[zbeekman]

@jacobwilliams Sounds like a plan to me!

Also, @jacobwilliams, just curious if you saw my PR from yesterday for json-fortran

[szaghi]

Arghhhhh

I was sure to have trapped the intrinsic modules usage, but evidently I was wrong! The idea of Jacob is very nice... tomorrow I will push the new 1.5.1 version with this feature 👍

[zbeekman]

My vote is for this behavior:

• FoBiS.py automatically detects standard intrinsic modules and handles them correctly
• If a missing dependency of any type is encountered a warning message is printed and then the build is attempted any way

This will likely be helpful for compiler vendor language extensions, etc.

[szaghi]

Hi all,

I just understand why my intrinsic modules usage is correctly handled and your not. In my OCD workflow I declare explicitly their are intrinsic module by the intrinsic keyword, thus FoBiS has a specific regex to catch these cases, but in your example I think your source is like use iso_c_binding and not use, intrinsic:: iso_c_binding consequently, due to the absence of keyword intrinsic FoBiS assumes that it is a standard module.

I like your feature request to collect the warning on missing dependencies and trying always to make the building, this fixes also the intrinsic modules issue, but I am thinking to change their trap: I do not like false positive warning about them... the standard defines their name thus I think it is better to catch them explicitly. Anyhow the intrinsic keyword trap remains.

I will push an update later this morning.

[szaghi]

Dear all,

in v1.5.1 should be fixed. It is worthy to note that your feature request should be also accomplished: now if a missing dependency of any type is encountered a warning message is printed and then the build is attempted any way.

See you soon.

[zbeekman]

Great! Thanks so much. Does this mean that use iso_c_binding will print a warning now or have the (3?) intrinsic modules been added to the regexp?

[szaghi]

Yes (or not :-) ), no warnings are now printed for the 5 standard intrinsic modules usage (with or without the explicit keyword intrinsic).

I have test with only iso_c_binding and iso_fortran_env, the ieee_xxxx modules are not tested.

See you soon.

[zbeekman]

Awesome, thanks! 👍