Nuclear simulation codes have been rewritten to run in more modern languages.[0] Not only is the DOE not required by law to detonate a nuke to make non-technical changes to their codes, they have invested many billions of dollars into programs for validating entirely new codes without nuclear testing, such as the NIF.[1]
[0] https://www.sciencedirect.com/science/article/abs/pii/S03064...
[1] https://lasers.llnl.gov/science/nif-stockpile-stewardship#:~...
Let's apply some critical thinking.
If the F90 code works it gives the same answers as the F77 code, no need to test any bombs.
Move along folks, nothing more to see.
That's not how validation works in systems like that. It may be the same. But you still need to fill out and sign a lot of forms to validate it. It's a bit like you can't compile some code and say "this works the same as FIPS module, I'm FIPS compliant".
You have to fill out a lot of forms but you don’t need to fire another bomb unless you believe the data you have isn’t sufficient to build some new kind of weapons that involves physics that isn’t well modeled.
FORTRAN '77 conforming code is (almost) always Fortran '90 conforming code too.
my first job: tend to 70k lines F77 on RSX-11M (pdp/11-44 pair)
interfaced to it with SunOS 4 & C
enjoyed my time with SunOS, not so much with RSX
If there actually were a Fortran 77 compiler industry, it would deserve to die.
Why?