cam history always chooses first averaging flag when using duplicate fields

[peverwhee]

What happened?

If, in user_nl_cam you're trying to do something like:

fincl2='Q:I', 'Q:A'

in the hopes that you'll get an h1i and h1a file, both with 'Q', it won't work!

You'll get whichever flag you specify first. So, in this case, you'd just get an h1i with 'Q'

What are the steps to reproduce the bug?

Run CAM with the user_nl_cam described above. Be sure your nhtfrq /= 1 for the test, since that overrides everything to "I" anyway.

What CAM tag were you using?

cam6_4_032

What machine were you running CAM on?

CGD machine (e.g. izumi)

What compiler were you using?

GNU

Path to a case directory, if applicable

No response

Will you be addressing this bug yourself?

Any CAM SE can do this

Extra info

Not sure exactly, but this seems like the problem area (in fldlst)

      listentry => masterlinkedlist
      do while(associated(listentry))
        mastername = listentry%field%name

        call list_index (fwrtpr(1,t), mastername, ffld)
        if (ffld > 0) then
          prec_wrt = getflag(fwrtpr(ffld,t))
        else
          prec_wrt = ' '
        end if

        call list_index (fincl(1,t), mastername, ffld)

        if (ffld > 0) then
          avgflag = getflag (fincl(ffld,t))
          call inifld (t, listentry, avgflag,  prec_wrt)
        else if ((.not. empty_htapes) .or. (is_initfile(file_index=t))) then
          call list_index (fexcl(1,t), mastername, ffld)
          if (ffld == 0 .and. listentry%actflag(t)) then
            call inifld (t, listentry, ' ', prec_wrt)
          else
            listentry%actflag(t) = .false.
          end if
        else
          listentry%actflag(t) = .false.
        end if
        listentry=>listentry%next_entry

      end do

[adamrher]

Peanut gallery here.

Be sure your nhtfrq /= 1 for the test, since that overrides everything to "I" anyway.

You can't have nhtfrq = 1 and use the "A" flag? This is useful for outfld calls in the macmic loops. For those variables I often set them to "A" and nhtfrq=1, so it is represents an average over all subcycles in a single time-step.

[peverwhee]

@adamrher hmmmm... I'm not really familiar with subcycling, but looking at the code, we're overriding the averaging flag with "I" when nhtfrq = 1. So my guess is you're not actually getting an average for those variables.

    !
    ! Initialize history variables
    !
    do t=1,ptapes
      do fld=1,nflds(t)
        if (nhtfrq(t) == 1) then
           ! Override any non-I flags if nhtfrq equals 1
           tape(t)%hlist(fld)%avgflag = 'I'
        end if
        ...

[adamrher]

It's always operated like that? Yikes, noted.

It would be ideal to have this functionality -- if I outfld a variable N times in a single time-step, then "A" and nhtfrq=1 should give me the average over those N calls -- but this is not a high priority.

[peverwhee]

@adamrher just checked an it was me in fact who added that override.

So it's definitely probable it used to work that way! I'll make an issue.

[gold2718]

Is this now superseded by #1150?

[peverwhee]

@gold2718 They're two separate issues. #1150 is for subcycling averages, this one is for duplicate fields for all history variables (won't be fixed by removing the override)

[PeterHjortLauritzen]

The functionality of outputting the same field with ‘I’ and ‘A’ in the same history stream was never supported in CAM (as far as I know). Let’s add an endrun if someone tries to do that.