On 9/17/26 12:17 AM, Paul Richard Thomas wrote:
Sorry, the subject line is wrong in my posting of 15 minutes ago.
Paul
On Thu, 17 Sept 2026 at 08:01, Paul Richard Thomas
<[email protected]> wrote:
Hello All,
The fix for this PR turned into a catch-all for parameterized string components:
Apart from the third chunk, the patch recognises fully that
pdt_strings are represented by descriptors. The third chunk corrects
an incorrect fail, when gfc_dep_compare_expr returns -3 for unlike
variables. This is overly restrictive and should be subject to a
runtime check.
The test, pdt_95.f03, grew like Topsy but clearly identifies the
original report from wrinkles that have been fixed.
It passes regression testing on FC44/x86_64 - OK for mainline and
after a decent interval, 16-branch?
Regards
Paul
This looks OK Paul. I wonder of it changes the results on the new test cases
Mikael posted with ne PR in Bugzilla. I did not try it.
Jerry
