Closed mandli closed 4 years ago
Note that at this point I cannot get this to pass the tests but it looks like it has something to do with the adjoint tests and I am having difficulty understanding why.
shoudln't have anything to do with it. There must a different thing going on - versionitis?
— Marsha
On Dec 17, 2019, at 6:07 PM, Kyle Mandli notifications@github.com wrote:
@mandli https://github.com/mandli requested your review on: #254 https://github.com/clawpack/amrclaw/pull/254 Switch mstart from being a local variable to the one stored in amr_module.
— You are receiving this because your review was requested. Reply to this email directly, view it on GitHub https://github.com/clawpack/amrclaw/pull/254?email_source=notifications&email_token=AAGUGC7QMUG62T6XZLBFGQ3QZFLSFA5CNFSM4J4CZPB2YY3PNVWWK3TUL52HS4DFWZEXG43VMVCXMZLOORHG65DJMZUWGYLUNFXW5KTDN5WW2ZLOORPWSZGOVRNUTJA#event-2891663780, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAGUGC3KZQOIRAF6NVU5UILQZFLSFANCNFSM4J4CZPBQ.
I agree and it looks like Travis is passing so I think this is probably safe to merge.
What test isn't passing? I tried this and it seems to work fine for me, both all the tests and examples/acoustics_1d_adjoint
.
As far as I can see, the only place mstart is used in these subroutines is in call outtre(mstart,.false.,nvar,naux)
immediately before stop
in the case of an error, so I don't see this change can affect cases that are running properly.
I agree but it's better to have it fixed. I think my own testing is messed up right now as well so I would not worry about it if this looks good.
This appears to be an issue when the f90 refactor occurred. This fixes that.