Hello All,

Now that we are approximately one month away from the appearance of a
release candidate for GCC 16, I have decided to leave PDT bug fixes
until 17 branch opens and get on with the regressions that are
assigned to me. The PDT bug 115315 now exposes a regression, which I
hope to fix, even if it doesn't fix the original bug.

It has been a bit of an uphill battle because right from the beginning
the implementation was designed with PDTs with LEN parameters in mind.
It turned out that this was slowing progress on PDTs with only KIND
parameters. It took me rather longer than it should have done to
realise that this was the case. Additionally, many of the PDT bugs
turned out to be due to other problems; mostly in nesting of structure
and array constructors, as found in fiats.

In all 77 bugs were fixed between 12th August 25 and today. 5 bugs
remain. PRs 82649 and 121972 involve parts of gfortran that I do not
want to touch until 17-branch opens.

Caveats aside, gfortran's implementation of PDTs is now very much more
compliant with the standard and very much more usable. Thank you to
all those that have helped, most especially Jerry Delisle and Damian
Rouson.

Regards

Paul

Reply via email to