PR fortran/19269 (partial fix): Lengths of reshaped arrays

[Richard Sandiford]

Paul Thomas <paulthomas2@wanadoo.fr> writes:
> Yes, as I said to you off-list, this was discussed some while back:

Argh!  Of course you did, sorry.  At the time, I was just concentrating
on the "should the callee or caller calculate the length?" question.
I'd completely forgotten that you'd talked about the library problem too.

> Indeed, you can drop the character length arguments for intrinsic
> functions and they all work fine.  I tested this and it works for each
> and every case.
>
> However, Paul B did not like it, on consistency grounds, and so the
> second solution is the preferred one.  I discussed it with Thomas
> Koenig, who is our expert on this part of the library; I am not sure
> whether he has put the job in his own queue or was assuming that I would
> do it!

OK ;)  I'm willing do this if you and Thomas agree.  Just let me know.
I'll unassign myself from the PR in the meantime.

Richard