Add semantics unit test for SYNC ALL statement

[rouson]

Summary

[flang] add semantics test for sync all

Test a range of acceptable forms of SYNC ALL statements,
including combinations with and without the stat-variable
and errmsg-variable present. Also test that several invalid
forms of SYNC ALL call generate the correct error messages.

[rouson]

Expand coverage of SYNC ALL statement semantics test:

1. Add checks for both constraints in the standard.
2. Check more valid and invalid statement forms.

[ktras]

flang/test/Semantics/synchronization01.f90

24

An addition of a sync all statement with the correct type for the errmsg specifier, but the incorrect name of the specifier, as you have done here for sync_status, would be a good addition. Perhaps the incorrect specifier name could be close to the correct one, such as saying errormsg=error_message instead of the correct errmsg=error_message.

[rouson]

1. Reorganized and regrouped checks.
2. Increased the coverage of checks for non-standard-conforming code.

[ktras]

Please see the inline comment.

flang/test/Semantics/synchronization01.f90

34

The variable non_scalar has not been declared, so the error would be a different one than I believe you are intending to test.

[rouson]

Added required variable declaration.

[ktras]

LGTM. The only note is that this test should pass if you take away the XFAIL directive, but doesn't because flang doesn't currently enforce the following constraints stated in the standard: C1171 - that no specifier shall appear more than once in a given sync-stat-list, and C1172 - that a stat-variable or errmsg-variable in a sync-stat shall not be a coindexed object. It seems appropriate to open up an issue on the github repository pointing this out after this commit is pushed, and referencing this test in the issue.

[awarzynski]

Thanks for working on this, @rouson !

I've just noticed #53915 - thank you for flagging this! LLVM Flang has only recently gained F95 support and I imagine it might be a while (perhaps years) before F18 support is available. It is possible that we won't be re-visiting this in the meantime and once we do come back to this we might find the rationale behind XFAILing this test rather unclear (or hard to track). I think that it would be worth adding a comment that there's no F18 support available yet and hence this test can neither be run nor tested.

Thank you,
-Andrzej

flang/test/Semantics/synchronization01.f90

+! RUN: %python %S/test_errors.py %s %flang_fc1
+! Check for semantic errors in sync all statements
+
+program test_sync_all
+  implicit none
+
+  integer sync_status
+  character(len=128) error_message
+  logical invalid_type
+  
+  sync all
+  sync all(stat=sync_status)
+  sync all(stat=sync_status, errmsg=error_message)
+ 
+  !ERROR: expected execution part construct
+  sync all(invalid_argument_keyword=sync_status)
+  !ERROR: expected execution part construct
+  sync all(invalid_type)
+  !ERROR: expected execution part construct
+  sync all(sync_status)
+  !ERROR: expected execution part construct
+  sync all(     sync_status, errmsg=error_message)
+  !ERROR: expected execution part construct
+  sync all(                  errmsg)
+  !ERROR: expected execution part construct
+  sync all(     sync_status, errmsg)
+ 
+end program test_sync_all