Build issues with izumi_nag for v3.0.1

[ekluzek]

in v3.0.1 I get these build errors for SMS_D_Ld5.5x5_amazon_rHDMA.I2000Clm60SpMizGs.izumi_nag.mizuroute-default:

Extension: /fs/cgd/data0/erik/ctsm_worktree/mizuroute/components/mizuRoute/route/build/src/mc_route.f90, line 98: Line longer than 132 characters
Warning: /fs/cgd/data0/erik/ctsm_worktree/mizuroute/components/mizuRoute/route/build/src/mc_route.f90, line 200: Unused dummy variable THIS
Extension: /fs/cgd/data0/erik/ctsm_worktree/mizuroute/components/mizuRoute/route/build/src/mc_route.f90, line 296: Line longer than 132 characters
Extension: /fs/cgd/data0/erik/ctsm_worktree/mizuroute/components/mizuRoute/route/build/src/mc_route.f90, line 300: Line longer than 132 characters
Extension: /fs/cgd/data0/erik/ctsm_worktree/mizuroute/components/mizuRoute/route/build/src/mc_route.f90, line 326: Line longer than 132 characters
Warning: /fs/cgd/data0/erik/ctsm_worktree/mizuroute/components/mizuRoute/route/build/src/mc_route.f90, line 403: Unused dummy variable T0
Warning: /fs/cgd/data0/erik/ctsm_worktree/mizuroute/components/mizuRoute/route/build/src/mc_route.f90, line 403: Unused dummy variable T1
Warning: /fs/cgd/data0/erik/ctsm_worktree/mizuroute/components/mizuRoute/route/build/src/mc_route.f90, line 405: INTEGERMISSING explicitly imported into MC_ROUTE_MODULE but not used
Time(Parsing): User 0.07, System 0.63
Error: /fs/cgd/data0/erik/ctsm_worktree/mizuroute/components/mizuRoute/route/build/src/mc_route.f90, line 192: No spacing specified for X edit descriptor
Time(Semantic Analysis): User 0.00, System 0.00
[NAG Fortran Compiler error termination, 1 error, 28 warnings]
gmake: *** [/fs/cgd/data0/erik/ctsm_worktree/mizuroute/cime/scripts/SMS_D_Ld5.5x5_amazon_rHDMA.I2000Clm60SpMizGs.izumi_nag.mizuroute-default.20260308_165904_ovtoiy/Tools/Makefile:973: mc_route.o] Error 2

[ekluzek]

We think these changes will resolve the build problems:

diff --git a/route/build/src/mc_route.f90 b/route/build/src/mc_route.f90
index 8cea9154..0a036f64 100644
--- a/route/build/src/mc_route.f90
+++ b/route/build/src/mc_route.f90
@@ -10,7 +10,6 @@ MODULE mc_route_module
 USE dataTypes,     ONLY: mcRCH           ! MC specific state data structure
 USE public_var,    ONLY: iulog           ! i/o logical unit number
 USE public_var,    ONLY: realMissing     ! missing value for real number
-USE public_var,    ONLY: integerMissing  ! missing value for integer number
 USE public_var,    ONLY: desireId        ! ID or reach where detailed reach state is print in log
 USE public_var,    ONLY: dt              ! simulation time step [sec]
 USE public_var,    ONLY: qmodOption      ! qmod option (use 1==direct insertion)
@@ -95,7 +94,7 @@ SUBROUTINE mc_rch(this,           & ! mc_route_rch object to bound this procedur
  if (nUps>0) then ! this hru is not headwater
    isHW = .false.
    do iUps = 1,nUps
-     if (.not. NETOPO_in(segIndex)%goodBas(iUps)) cycle ! skip upstream reach which does not any flow due to zero total contributory areas
+     if (.not. NETOPO_in(segIndex)%goodBas(iUps)) cycle ! skip upstream reach with zero total contributory area
      iRch_ups = NETOPO_in(segIndex)%UREACHI(iUps)      !  index of upstream of segIndex-th reach
      q_upstream = q_upstream + RCHFLX_out(iRch_ups)%ROUTE(idxMC)%REACH_Q
    end do
@@ -189,7 +188,7 @@ SUBROUTINE mc_rch(this,           & ! mc_route_rch object to bound this procedur
                          RCHFLX_out,     & ! inout: reach fluxes datq structure
                          ierr, cmessage)   ! output: error control
    if(ierr/=0)then
-     write(message,'(A,X,I12,X,A)') trim(message)//'/segment=', NETOPO_in(segIndex)%REACHID, '/'//trim(cmessage); return
+     write(message,'(A,1X,I12,1X,A)') trim(message)//'/segment=', NETOPO_in(segIndex)%REACHID, '/'//trim(cmessage); return
    endif
  end if

@@ -293,11 +292,12 @@ SUBROUTINE muskingum_cunge(rch_param,     & ! input: river parameter data struct
      write(iulog,'(3(A,1X,G12.5))') ' Qin(t-1) Q(0,0)=',Q(0,0),' Qin(t) Q(1,0)=',Q(1,0),' Qout(t-1) Q(0,1)=',Q(0,1)
    end if
 
-   ! first, using 3-point average in computational molecule, check Cournat number is less than 1, otherwise subcycle within one t
ime step
+   ! first, using 3-point average in computational molecule, check Cournat number is less than 1, 
+   ! otherwise subcycle within one time step
    Qbar = (Q(0,0)+Q(1,0)+Q(0,1))/3.0  ! average discharge [m3/s]
 
    if (Qbar>0._dp) then
-     depth = flow_depth(abs(Qbar), bt, zc, S, n, zf=zf, bankDepth=bankDepth) ! compute flow depth as normal depth (a function of 
flow)
+     depth = flow_depth(abs(Qbar), bt, zc, S, n, zf=zf, bankDepth=bankDepth) ! compute flow depth as normal depth
      ck   = celerity(abs(Qbar), depth, bt, zc, S, n, zf=zf, bankDepth=bankDepth)
      Cn   = ck*theta                    ! Courant number [-]
 
@@ -323,7 +323,7 @@ SUBROUTINE muskingum_cunge(rch_param,     & ! input: river parameter data struct
      do ix = 1, nTsub
        Qbar = (QinLocal(ix)+QinLocal(ix-1)+QoutLocal(ix-1))/3.0 ! 3 point average discharge [m3/s]
        if (Qbar>0._dp) then
-         depth    = flow_depth(abs(Qbar), bt, zc, S, n, zf=zf, bankDepth=bankDepth) ! compute flow depth as normal depth (a funct
ion of flow)
+         depth    = flow_depth(abs(Qbar), bt, zc, S, n, zf=zf, bankDepth=bankDepth) ! compute flow depth as normal depth
          topWidth = Btop(depth, bt, zc, zf=zf, bankDepth=bankDepth) ! top width at water level [m] (rectangular channel)
          ck       = celerity(abs(Qbar), depth, bt, zc, S, n, zf=zf, bankDepth=bankDepth)

[nmizukami]

Hi Erik, I now see many f90 files include lines exceeding 132 characters.
I wonder if it is possible to add compiler option-
-free
this Interprets all Fortran source files according to free-form rules.
I think 132 characters limit is coming from old day f77 (maybe because old IBM line printers has 132 character width).
Even with gnu compiler, compiling without -ffree-line-length-none would complain the line length.

I guess too long line would not be good, but not sure 132 is too restrictive or not

[ekluzek]

Hi Erik, I now see many f90 files include lines exceeding 132 characters. I wonder if it is possible to add compiler option- -free this Interprets all Fortran source files according to free-form rules. I think 132 characters limit is coming from old day f77 (maybe because old IBM line printers has 132 character width). Even with gnu compiler, compiling without -ffree-line-length-none would complain the line length.

I guess too long line would not be good, but not sure 132 is too restrictive or not

@nmizukami personally I do like having a limit and think 132 is reasonable. In order to free the restriction we'd need to change this in ccs_config and make sure we get agreement in CESM that it's a good thing to do. I brought it up in the CTSM SE group and they agreed that having a limit on lines is a good thing. And that 132 is a reasonable limit, and even shorter would be preffered. The main thing expressed was that excessively long lines interferes with readability, so having a limit enforced by one of our compilers is a good thing.

Looking it up. The original fixed-form was 72 characters. And the F90 standard increased it to 132. Different compilers allow it to go beyond with compiler options, but it looks like the standard is a 132 character limit. As such it seems like something that's good to conform to. There could be a compiler someone would want to use in the future that doesn't allow a way around the standard.

[nmizukami]

Hi Erik, Sound good for limiting the line to 132 characters. Wonder if there is any way to screen all the f90 files to identify the line exceeding this. vs code can do (I assume so)?? This would be one separate PR to do.

[ekluzek]

I did a little cshell operation to get a file to show all the longlines:

 1097  15:44   foreach file ( `find . -name '*.f90' -print` )
  1098  15:45   awk -v SH_FILE="$file" 'length($0) > 132 {print SH_FILE" :: " length($0)" chars: ln: "NR" --"$0}' $file >> longlines.txt
  1099  15:45   end

The list of files in v3.0.1 is this:

./src/accum_runoff.f90
./src/advection_diffusion.f90
./src/allocation.f90
./src/basinUH.f90
./src/data_assimilation.f90
./src/dataTypes.f90
./src/dfw_route.f90
./src/domain_decomposition.f90
./src/globalData.f90
./src/historyFile.f90
./src/histVars_data.f90
./src/hydraulic.f90
./src/init_model_data.f90
./src/irf_route.f90
./src/kwe_route.f90
./src/kwt_route.f90
./src/lake_route.f90
./src/main_route.f90
./src/mpi_process.f90
./src/ncio_utils.f90
./src/network_topo.f90
./src/nr_utils.f90
./src/pfafstetter.f90
./src/popMetadat.f90
./src/process_ntopo.f90
./src/process_param.f90
./src/process_remap.f90
./src/public_var.f90
./src/read_control.f90
./src/read_restart.f90
./src/read_streamSeg.f90
./src/standalone/get_basin_runoff.f90
./src/standalone/model_setup.f90
./src/standalone/read_remap.f90
./src/standalone/read_runoff.f90
./src/tracer.f90
./src/var_lookup.f90
./src/write_restart_pio.f90
./src/write_simoutput_pio.f90
./src/write_streamSeg.f90

with 730 total lines over 132, out of 24k lines.

[ekluzek]

The list of files is the same for the branch: nmizukami:default_setup_coupled_mode. And only 8 more lines with issues. The median is about 160 and the longest is 229 characters.

[ekluzek]

Tying a different search getting rid of comments in lines I got it down to this list of 26:

   310  15:09   foreach file ( `cat list` )
   311  15:09   sed 's/\!.*//' $file > $file.notrailingcomments
   312  15:09   awk -v SH_FILE="$file" 'length($0) > 132 {print SH_FILE" :: " length($0)" chars: ln: "NR" --"$0}' $file.notrailingcomments >> longlines.txt
   313  15:09   rm $file.notrailingcomments
   314  15:09   end

./src/accum_runoff.f90
./src/allocation.f90
./src/data_assimilation.f90
./src/dfw_route.f90
./src/historyFile.f90
./src/histVars_data.f90
./src/hydraulic.f90
./src/kwe_route.f90
./src/kwt_route.f90
./src/lake_route.f90
./src/ncio_utils.f90
./src/nr_utils.f90
./src/pfafstetter.f90
./src/popMetadat.f90
./src/process_ntopo.f90
./src/process_remap.f90
./src/read_control.f90
./src/read_restart.f90
./src/read_streamSeg.f90
./src/standalone/get_basin_runoff.f90
./src/standalone/model_setup.f90
./src/standalone/read_remap.f90
./src/standalone/read_runoff.f90
./src/write_restart_pio.f90
./src/write_simoutput_pio.f90
./src/write_streamSeg.f90

And total number of lines is 305 rather than over 700.