tmpstr not defined in ice_init.F90

[dabail10]

Looks like this code is scheduled for deprecation, but it does not compile because tmpstr is not defined.

! each task gets unique ice log filename when if test is true, for debugging
if (1 == 0) then
call get_fileUnit(nu_diag)
write(tmpstr,'(a,i4.4)') "ice.log.task_",my_task
open(nu_diag,file=tmpstr)
endif

[eclare108213]

Just need to change tmpstr to tmpstr2. Amazing this hasn't been caught before now. Which compiler?

[apcraig]

This is never compiled in standalone mode, it's within a CESMCOUPLED ifdef. This section of code is only on if you manually change (1 == 0) to (1 == 1). What it does is give each task a unique ice.log file to write to, so you can separate the output from each task by file. This would only be used for debugging. I wonder if compilers are skipping that code because it's guaranteed to not be used unless you set (1 == 1). This is not scheduled for deprecation. I think it's a hidden feature. What machine did it not compile on? Were you trying to use it (i.e. setting 1==1)?

[dabail10]

This was on cheyenne intel. I can't recall if I had optimization turned off.

[dabail10]

This was within CESM, so CESMCOUPLED is turned on.

[dabail10]

I have a PR coming with the saltflux stuff and I will fix it there.

[apcraig]

I knew it was in CESM, but was just pointing out that the standalone model would not have caught this. I just checked RASM. The code is there and CESMCOUPLED is on. We have a slightly older version. tmpstr is defined there. My guess is that when I was cleaning up compiler warning messages about unused variables and such, I removed tmpstr from the module. This would not have been needed for standalone. I suspect it's my fault and probably happened fairly recently. Sorry about that. Thanks @dabail10 for fixing this.

[dabail10]

No worries. Testing my changes now.

[eclare108213]

Can we close this issue now?

[dabail10]

Fixed in #799.