Skip to content

Conversation

@JDPailleux
Copy link
Contributor

After the merge of llvm/llvm-project#125115 to flang, two gfortran tests compile and run successfully after the support of co_reduce but they're marked "xfail".

@JDPailleux JDPailleux requested a review from klausler March 5, 2025 09:11
@DavidSpickett DavidSpickett requested a review from omjavaid March 5, 2025 09:23
@omjavaid
Copy link
Contributor

omjavaid commented Mar 5, 2025

@JDPailleux if these tests are passing consistently then shouldnt we remove the xfail decorator instead of disabling them

Fortran/gfortran/regression/tests.cmake:compile;coarray_collectives_10.f90;xfail;-fcoarray=single -std=f2008;;
Fortran/gfortran/regression/tests.cmake:compile;coarray_collectives_13.f90;xfail;-fcoarray=single -std=f2008;;

@JDPailleux
Copy link
Contributor Author

@omjavaid I don't know if this is the right way to do it because “normally”, when passing the -std=f2008 flag, this test will fail because co_reduce is not supported by the 2008 standard. (like the coarray_collectives_2.f90 on co_sum and more, I've adopted the same way of ignoring it because I think it's better than removing the xfail).
I think it's best to ignore it for now, but maybe I'm wrong.

@omjavaid
Copy link
Contributor

omjavaid commented Mar 5, 2025

@omjavaid I don't know if this is the right way to do it because “normally”, when passing the -std=f2008 flag, this test will fail because co_reduce is not supported by the 2008 standard. (like the coarray_collectives_2.f90 on co_sum and more, I've adopted the same way of ignoring it because I think it's better than removing the xfail). I think it's best to ignore it for now, but maybe I'm wrong.

Ack. Agreed Thanks

Copy link
Contributor

@omjavaid omjavaid left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

Copy link
Contributor

@tarunprabhu tarunprabhu left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks.

It seems like theese tests should "xfail" in the future when flang implements strict standards checks. In that case, disabling the tests temporarily is the right thing to do.

There is a mechanism to override the DejaGNU annotations in case we need it. See the README in the Fortran/gfortran for details. The source files in the gfortran test suite directory must not be changed.

@klausler
Copy link
Contributor

klausler commented Mar 5, 2025

The semantics checks in f18 don't structure features by standard revisions, so don't depend on GNU-like -fstd=... options that won't be implemented.

@tarunprabhu
Copy link
Contributor

@klausler,

There are a number of tests in the gfortran test suite that check for standards compliance. Will flang ever have something similar?

@klausler
Copy link
Contributor

klausler commented Mar 5, 2025

@klausler,

There are a number of tests in the gfortran test suite that check for standards compliance. Will flang ever have something similar?

No, that's my point, we don't check conformance with old standards, and it's very unlikely that we ever will.

@JDPailleux JDPailleux merged commit c9d3b5a into llvm:main Mar 6, 2025
1 check passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants