the only knowledge I have of Fortran not being dead is that its use in scientific computing is brought up in EVERY SINGLE ONE OF THESE PL CONVERSATIONS I've bee

the only knowledge I have of Fortran not being dead is that its use in scientific computing is brought up in EVERY SINGLE ONE OF THESE PL CONVERSATIONS I've been in dating back to the 1970s. (It was the first language I learned if you you don't count HP calculators) COBOL too. Do the people maintaining these libraries get harangued on a weekly basis about how Fortran has undefined behavior and the PDP-11 and IBM 360 ISAs are not appropriate for modern multi level cached pipelined speculatively executed CPUs? How can their matrix multiplication libraries be as fast as they can be if they are assuming/relying on memory allocation layouts that don't exist any more in the hardware?