ncd_io_1d_log_glob is broken #24

billsacks · 2017-12-16T20:12:34Z

Bill Sacks < sacks@ucar.edu > - 2014-01-23 11:14:21 -0700
Bugzilla Id: 1906
Bugzilla CC: muszala@ucar.edu, mvertens@ucar.edu, rfisher@ucar.edu,

When trying to read in logical values with ncd_io in clm4_5_59, I get this message:

(shr_sys_abort) ERROR: ncd_io_1d_log_glob ERROR: read in bad integer value(s) for logical data

It looks like the relevant variable is never set: starting line 1216 in ncdio_pio.F90:

      allocate(idata1d(size(data))) 
      data = (idata == 1)
      if ( any(idata1d /= 0 .and. idata1d /= 1) )then
         call shr_sys_abort( subname// &
              ' ERROR: read in bad integer value(s) for logical data' )
      end if

This worked in clm4_5_52. My guess is that it got broken in clm4_5_53, which created the .in version of ncdio_pio.F90.

The text was updated successfully, but these errors were encountered:

billsacks · 2017-12-16T20:12:40Z

Bill Sacks < sacks@ucar.edu > - 2016-06-29 15:48:15 -0600

Plan is to remove this for now, since it apparently isn't used and is broken.

For now I'm just removing the ability to try to call this with a 1-d logical, rather than trying to fix the code. Fixes ESCOMP#24

Minor bug fixes, cleanup, documentation and enhancements A collection of minor bug fixes, code cleanup, documentation and enhancements, all bit-for-bit. See list of issues fixed below for details. Issues fixed (include CTSM Issue #): - Fixes #24 (ncd_io_1d_log_glob is broken) - Fixes #120 (Incorrect comments in Biogeophysics1Mod.F90 - Fixes #217 (Change some cheyenne_gnu tests to cheyenne_intel) - Fixes #245 (Put all .gitignore entries in top-level file) - Fixes #272 (Code should error on missing mxsoil_color when SOIL_COLOR is used) - Fixes #283 (Add more helpful message about need to do init_interp with wrong number of vertical layers) - Fixes #367 (For cmip6 runs: Turn on cpl hist output needed to drive a TG compset) - Fixes #412 (Fix documentation of init_interp_method) - Fixes #419 (Do not allow SOYFIXN diagnostic field with FUN) - Fixes #464 (Some lines longer than 132 characters) - Fixes #465 (Remove backwards compatibility check for snw_rds) - Fixes #467 (Increase wallclock limit for test)

I copied the implementation of ncd_io_{DIMS}d_{TYPE}_glob to ncd_io_{DIMS}d_log_glob (with appropriate adjustments for the fact that we're working with logical data). By looking at diffs between those routines, I have confirmed that the two are now essentially identical in their implementation. I also made some very minor tweaks to ncd_io_{DIMS}d_{TYPE}_glob in order to reconcile the diffs between these two routines. This fixes ESCOMP#24 for real.

CMIP6 compset modifiers; usermods for output; other output enhancements (1) Support %BGC-CROP-CMIP6DECK and %BGC-CROP-CMIP6WACCMDECK compset modifiers, so that we can turn on the necessary options (output-related and others) via new CMIP6-specific compsets. (2) Turn on carbon isotopes in CMIP6 runs (from Erik Kluzek) (3) Remove setting of CCSM_BGC=CO2A in the cmip6 usermods (4) Add usermods directories for getting typical extra output that's wanted in many cases: output_crop, output_crop_highfreq, output_bgc, output_bgc_highfreq, output_sp, and output_sp_highfreq. These can be enabled by adding something like '--user-mods-dir output_crop' on the create_newcase line (that short-hand works for an I compset; for F or B compsets, you need to provide the full path to the usermod directory). (4) Allow holes in the number of history tapes. Holes are cases where, for example, we have h0, h1 and h3 tapes, but no h2 tape (because there are no fields on the h2 tape). (This is needed for (3).) (5) Fix reading and writing of 1-d logical global arrays. This fixes #24 for real (rather than just preventing an attempt to read/write 1-d logical arrays, as was done in the previous 'fix'). (6) Add C13_NBP and C14_NBP diagnostic fields (from Keith Oleson) (7) Make a bunch of carbon isotope diagnostic fields inactive by default (8) Don't allow interpolation (use_init_interp) from a case without carbon isotopes to a case with carbon isotopes: Due to #67, interpolation from a case without carbon isotopes to a case with carbon isotopes yields incorrect initialization values for the carbon isotopes. Now that we're turning carbon isotopes on via some semi-out-of-the-box usermods (for cmip6), it is becoming more important to check to make sure someone doesn't shoot themselves in the foot this way. (9) Add tests of the new output usermods as well as of the CMIP6 compset modifiers Issues fixed: - Fixes #24 for real (ncd_io_1d_log_glob is broken) - Fixes #529 (Organize usermods_dirs to facilitate running cases with the right output)

I copied the implementation of ncd_io_{DIMS}d_{TYPE}_glob to ncd_io_{DIMS}d_log_glob (with appropriate adjustments for the fact that we're working with logical data). By looking at diffs between those routines, I have confirmed that the two are now essentially identical in their implementation. I also made some very minor tweaks to ncd_io_{DIMS}d_{TYPE}_glob in order to reconcile the diffs between these two routines. This fixes ESCOMP#24 for real.

From dev014 & dev015: CMIP6 compset modifiers, output usermods & fixes Bring in all changes from ctsm1.0.dev014 and ctsm1.0.dev015: From ctsm1.0.dev015: (1) Support %BGC-CROP-CMIP6DECK and %BGC-CROP-CMIP6WACCMDECK compset modifiers, so that we can turn on the necessary options (output-related and others) via new CMIP6-specific compsets. (2) Turn on carbon isotopes in CMIP6 runs (from Erik Kluzek) (3) Remove setting of CCSM_BGC=CO2A in the cmip6 usermods (4) Add usermods directories for getting typical extra output that's wanted in many cases: output_crop, output_crop_highfreq, output_bgc, output_bgc_highfreq, output_sp, and output_sp_highfreq. These can be enabled by adding something like '--user-mods-dir output_crop' on the create_newcase line (that short-hand works for an I compset; for F or B compsets, you need to provide the full path to the usermod directory). (4) Allow holes in the number of history tapes. Holes are cases where, for example, we have h0, h1 and h3 tapes, but no h2 tape (because there are no fields on the h2 tape). (This is needed for (3).) (5) Fix reading and writing of 1-d logical global arrays. This fixes #24 for real (rather than just preventing an attempt to read/write 1-d logical arrays, as was done in the previous 'fix'). (6) Add C13_NBP and C14_NBP diagnostic fields (from Keith Oleson) (7) Make a bunch of carbon isotope diagnostic fields inactive by default (8) Don't allow interpolation (use_init_interp) from a case without carbon isotopes to a case with carbon isotopes: Due to #67, interpolation from a case without carbon isotopes to a case with carbon isotopes yields incorrect initialization values for the carbon isotopes. Now that we're turning carbon isotopes on via some semi-out-of-the-box usermods (for cmip6), it is becoming more important to check to make sure someone doesn't shoot themselves in the foot this way. (9) Add tests of the new output usermods as well as of the CMIP6 compset modifiers From ctsm1.0.dev014: Four miscellaneous minor, bit-for-bit bug fixes: (1) Py3 pylint check and address cime issue ESMCI/cime#2822 (from Jim Edwards: #526) (2) Change uppercase DEBUG variables to lowercase debug (requested by Jim Edwards to avoid conflicting with the DEBUG CPP token) (Fixes #534) (3) Remove unnecessary line of code in LunaMod.F90 that was causing problems with some compilers due to an uninitialized variable (Fixes #322) (4) Add r8 to 0 constant to fix build issue with XLF compiler (from Jim Edwards: #531)

CMIP6 compset modifiers; usermods for output; other output enhancements (1) Support %BGC-CROP-CMIP6DECK and %BGC-CROP-CMIP6WACCMDECK compset modifiers, so that we can turn on the necessary options (output-related and others) via new CMIP6-specific compsets. (2) Turn on carbon isotopes in CMIP6 runs (from Erik Kluzek) (3) Remove setting of CCSM_BGC=CO2A in the cmip6 usermods (4) Add usermods directories for getting typical extra output that's wanted in many cases: output_crop, output_crop_highfreq, output_bgc, output_bgc_highfreq, output_sp, and output_sp_highfreq. These can be enabled by adding something like '--user-mods-dir output_crop' on the create_newcase line (that short-hand works for an I compset; for F or B compsets, you need to provide the full path to the usermod directory). (4) Allow holes in the number of history tapes. Holes are cases where, for example, we have h0, h1 and h3 tapes, but no h2 tape (because there are no fields on the h2 tape). (This is needed for (3).) (5) Fix reading and writing of 1-d logical global arrays. This fixes ESCOMP#24 for real (rather than just preventing an attempt to read/write 1-d logical arrays, as was done in the previous 'fix'). (6) Add C13_NBP and C14_NBP diagnostic fields (from Keith Oleson) (7) Make a bunch of carbon isotope diagnostic fields inactive by default (8) Don't allow interpolation (use_init_interp) from a case without carbon isotopes to a case with carbon isotopes: Due to ESCOMP#67, interpolation from a case without carbon isotopes to a case with carbon isotopes yields incorrect initialization values for the carbon isotopes. Now that we're turning carbon isotopes on via some semi-out-of-the-box usermods (for cmip6), it is becoming more important to check to make sure someone doesn't shoot themselves in the foot this way. (9) Add tests of the new output usermods as well as of the CMIP6 compset modifiers Conflicts: src/dyn_subgrid/dyncropFileMod.F90 slevis resolved conflicts (as far as I can tell) but cheyenne_intel suite returns two new FAILs: ERP_D_Ld5.f09_g17.I2000Clm50Vic.cheyenne_intel.clm-vrtlay RUN ERS_D_Ld10.f10_f10_musgs.IHistClm50SpG.cheyenne_intel.clm-collapse_pfts_78_to_16_f10 RUN I am committing the merge so as to then look at diffs that may be leading to the FAILs.

For now I'm just removing the ability to try to call this with a 1-d logical, rather than trying to fix the code. Fixes ESCOMP#24

Minor bug fixes, cleanup, documentation and enhancements A collection of minor bug fixes, code cleanup, documentation and enhancements, all bit-for-bit. See list of issues fixed below for details. Issues fixed (include CTSM Issue #): - Fixes ESCOMP#24 (ncd_io_1d_log_glob is broken) - Fixes ESCOMP#120 (Incorrect comments in Biogeophysics1Mod.F90 - Fixes ESCOMP#217 (Change some cheyenne_gnu tests to cheyenne_intel) - Fixes ESCOMP#245 (Put all .gitignore entries in top-level file) - Fixes ESCOMP#272 (Code should error on missing mxsoil_color when SOIL_COLOR is used) - Fixes ESCOMP#283 (Add more helpful message about need to do init_interp with wrong number of vertical layers) - Fixes ESCOMP#367 (For cmip6 runs: Turn on cpl hist output needed to drive a TG compset) - Fixes ESCOMP#412 (Fix documentation of init_interp_method) - Fixes ESCOMP#419 (Do not allow SOYFIXN diagnostic field with FUN) - Fixes ESCOMP#464 (Some lines longer than 132 characters) - Fixes ESCOMP#465 (Remove backwards compatibility check for snw_rds) - Fixes ESCOMP#467 (Increase wallclock limit for test)

I copied the implementation of ncd_io_{DIMS}d_{TYPE}_glob to ncd_io_{DIMS}d_log_glob (with appropriate adjustments for the fact that we're working with logical data). By looking at diffs between those routines, I have confirmed that the two are now essentially identical in their implementation. I also made some very minor tweaks to ncd_io_{DIMS}d_{TYPE}_glob in order to reconcile the diffs between these two routines. This fixes ESCOMP#24 for real.

CMIP6 compset modifiers; usermods for output; other output enhancements (1) Support %BGC-CROP-CMIP6DECK and %BGC-CROP-CMIP6WACCMDECK compset modifiers, so that we can turn on the necessary options (output-related and others) via new CMIP6-specific compsets. (2) Turn on carbon isotopes in CMIP6 runs (from Erik Kluzek) (3) Remove setting of CCSM_BGC=CO2A in the cmip6 usermods (4) Add usermods directories for getting typical extra output that's wanted in many cases: output_crop, output_crop_highfreq, output_bgc, output_bgc_highfreq, output_sp, and output_sp_highfreq. These can be enabled by adding something like '--user-mods-dir output_crop' on the create_newcase line (that short-hand works for an I compset; for F or B compsets, you need to provide the full path to the usermod directory). (4) Allow holes in the number of history tapes. Holes are cases where, for example, we have h0, h1 and h3 tapes, but no h2 tape (because there are no fields on the h2 tape). (This is needed for (3).) (5) Fix reading and writing of 1-d logical global arrays. This fixes ESCOMP#24 for real (rather than just preventing an attempt to read/write 1-d logical arrays, as was done in the previous 'fix'). (6) Add C13_NBP and C14_NBP diagnostic fields (from Keith Oleson) (7) Make a bunch of carbon isotope diagnostic fields inactive by default (8) Don't allow interpolation (use_init_interp) from a case without carbon isotopes to a case with carbon isotopes: Due to ESCOMP#67, interpolation from a case without carbon isotopes to a case with carbon isotopes yields incorrect initialization values for the carbon isotopes. Now that we're turning carbon isotopes on via some semi-out-of-the-box usermods (for cmip6), it is becoming more important to check to make sure someone doesn't shoot themselves in the foot this way. (9) Add tests of the new output usermods as well as of the CMIP6 compset modifiers Issues fixed: - Fixes ESCOMP#24 for real (ncd_io_1d_log_glob is broken) - Fixes ESCOMP#529 (Organize usermods_dirs to facilitate running cases with the right output)

fix referencing shr_kind_in

For now I'm just removing the ability to try to call this with a 1-d logical, rather than trying to fix the code. Fixes ESCOMP#24

I copied the implementation of ncd_io_{DIMS}d_{TYPE}_glob to ncd_io_{DIMS}d_log_glob (with appropriate adjustments for the fact that we're working with logical data). By looking at diffs between those routines, I have confirmed that the two are now essentially identical in their implementation. I also made some very minor tweaks to ncd_io_{DIMS}d_{TYPE}_glob in order to reconcile the diffs between these two routines. This fixes ESCOMP#24 for real.

From dev014 & dev015: CMIP6 compset modifiers, output usermods & fixes Bring in all changes from ctsm1.0.dev014 and ctsm1.0.dev015: From ctsm1.0.dev015: (1) Support %BGC-CROP-CMIP6DECK and %BGC-CROP-CMIP6WACCMDECK compset modifiers, so that we can turn on the necessary options (output-related and others) via new CMIP6-specific compsets. (2) Turn on carbon isotopes in CMIP6 runs (from Erik Kluzek) (3) Remove setting of CCSM_BGC=CO2A in the cmip6 usermods (4) Add usermods directories for getting typical extra output that's wanted in many cases: output_crop, output_crop_highfreq, output_bgc, output_bgc_highfreq, output_sp, and output_sp_highfreq. These can be enabled by adding something like '--user-mods-dir output_crop' on the create_newcase line (that short-hand works for an I compset; for F or B compsets, you need to provide the full path to the usermod directory). (4) Allow holes in the number of history tapes. Holes are cases where, for example, we have h0, h1 and h3 tapes, but no h2 tape (because there are no fields on the h2 tape). (This is needed for (3).) (5) Fix reading and writing of 1-d logical global arrays. This fixes ESCOMP#24 for real (rather than just preventing an attempt to read/write 1-d logical arrays, as was done in the previous 'fix'). (6) Add C13_NBP and C14_NBP diagnostic fields (from Keith Oleson) (7) Make a bunch of carbon isotope diagnostic fields inactive by default (8) Don't allow interpolation (use_init_interp) from a case without carbon isotopes to a case with carbon isotopes: Due to ESCOMP#67, interpolation from a case without carbon isotopes to a case with carbon isotopes yields incorrect initialization values for the carbon isotopes. Now that we're turning carbon isotopes on via some semi-out-of-the-box usermods (for cmip6), it is becoming more important to check to make sure someone doesn't shoot themselves in the foot this way. (9) Add tests of the new output usermods as well as of the CMIP6 compset modifiers From ctsm1.0.dev014: Four miscellaneous minor, bit-for-bit bug fixes: (1) Py3 pylint check and address cime issue ESMCI/cime#2822 (from Jim Edwards: ESCOMP#526) (2) Change uppercase DEBUG variables to lowercase debug (requested by Jim Edwards to avoid conflicting with the DEBUG CPP token) (Fixes ESCOMP#534) (3) Remove unnecessary line of code in LunaMod.F90 that was causing problems with some compilers due to an uninitialized variable (Fixes ESCOMP#322) (4) Add r8 to 0 constant to fix build issue with XLF compiler (from Jim Edwards: ESCOMP#531)

…SCOMP#24) Otherwise, an error for duplicate calling of `mpi_init` is thrown, since PDAF calls `mpi_init` in its main routine `pdaf_terrsysmp.F90`. Co-authored-by: Johannes Keller <jo.keller@fz-juelich.de>

…r PDAF (ESCOMP#24) 'e41dc00a7e242bcb27b83f2cbe5c528b1227ba68' into eclm

For now I'm just removing the ability to try to call this with a 1-d logical, rather than trying to fix the code. Fixes ESCOMP#24

I copied the implementation of ncd_io_{DIMS}d_{TYPE}_glob to ncd_io_{DIMS}d_log_glob (with appropriate adjustments for the fact that we're working with logical data). By looking at diffs between those routines, I have confirmed that the two are now essentially identical in their implementation. I also made some very minor tweaks to ncd_io_{DIMS}d_{TYPE}_glob in order to reconcile the diffs between these two routines. This fixes ESCOMP#24 for real.

From dev014 & dev015: CMIP6 compset modifiers, output usermods & fixes Bring in all changes from ctsm1.0.dev014 and ctsm1.0.dev015: From ctsm1.0.dev015: (1) Support %BGC-CROP-CMIP6DECK and %BGC-CROP-CMIP6WACCMDECK compset modifiers, so that we can turn on the necessary options (output-related and others) via new CMIP6-specific compsets. (2) Turn on carbon isotopes in CMIP6 runs (from Erik Kluzek) (3) Remove setting of CCSM_BGC=CO2A in the cmip6 usermods (4) Add usermods directories for getting typical extra output that's wanted in many cases: output_crop, output_crop_highfreq, output_bgc, output_bgc_highfreq, output_sp, and output_sp_highfreq. These can be enabled by adding something like '--user-mods-dir output_crop' on the create_newcase line (that short-hand works for an I compset; for F or B compsets, you need to provide the full path to the usermod directory). (4) Allow holes in the number of history tapes. Holes are cases where, for example, we have h0, h1 and h3 tapes, but no h2 tape (because there are no fields on the h2 tape). (This is needed for (3).) (5) Fix reading and writing of 1-d logical global arrays. This fixes ESCOMP#24 for real (rather than just preventing an attempt to read/write 1-d logical arrays, as was done in the previous 'fix'). (6) Add C13_NBP and C14_NBP diagnostic fields (from Keith Oleson) (7) Make a bunch of carbon isotope diagnostic fields inactive by default (8) Don't allow interpolation (use_init_interp) from a case without carbon isotopes to a case with carbon isotopes: Due to ESCOMP#67, interpolation from a case without carbon isotopes to a case with carbon isotopes yields incorrect initialization values for the carbon isotopes. Now that we're turning carbon isotopes on via some semi-out-of-the-box usermods (for cmip6), it is becoming more important to check to make sure someone doesn't shoot themselves in the foot this way. (9) Add tests of the new output usermods as well as of the CMIP6 compset modifiers From ctsm1.0.dev014: Four miscellaneous minor, bit-for-bit bug fixes: (1) Py3 pylint check and address cime issue ESMCI/cime#2822 (from Jim Edwards: ESCOMP#526) (2) Change uppercase DEBUG variables to lowercase debug (requested by Jim Edwards to avoid conflicting with the DEBUG CPP token) (Fixes ESCOMP#534) (3) Remove unnecessary line of code in LunaMod.F90 that was causing problems with some compilers due to an uninitialized variable (Fixes ESCOMP#322) (4) Add r8 to 0 constant to fix build issue with XLF compiler (from Jim Edwards: ESCOMP#531)

billsacks added this to the clm5 milestone Dec 16, 2017

billsacks added type: bug bug something is working incorrectly and removed type: bug - science labels Feb 8, 2018

billsacks added a commit to billsacks/ctsm that referenced this issue Aug 3, 2018

Remove broken ncd_io_1d_log_glob

6566eb1

For now I'm just removing the ability to try to call this with a 1-d logical, rather than trying to fix the code. Fixes ESCOMP#24

billsacks closed this as completed in 0a36ae0 Aug 4, 2018

billsacks mentioned this issue Aug 4, 2018

Minor bug fixes, cleanup, documentation and enhancements #468

Merged

billsacks mentioned this issue Oct 12, 2018

Allow holes in the number of history tapes and reorganize cmip6 usermods #536

Merged

billsacks added a commit to billsacks/ctsm that referenced this issue Feb 22, 2019

Remove broken ncd_io_1d_log_glob

aec515e

For now I'm just removing the ability to try to call this with a 1-d logical, rather than trying to fix the code. Fixes ESCOMP#24

MiCurry pushed a commit to MiCurry/CTSM that referenced this issue Sep 16, 2021

Merge pull request ESCOMP#24 from ESCOMP/mvertens/fix_refto_shr_kind2

fd70374

fix referencing shr_kind_in

slevis-lmwg pushed a commit to slevis-lmwg/ctsm that referenced this issue Aug 1, 2023

Fix for cold start ESCOMP#24 and add cold start test

79d8b5e

AGonzalezNicolas pushed a commit to HPSCTerrSys/clm5_0 that referenced this issue Jun 27, 2024

Remove broken ncd_io_1d_log_glob

66f19f6

For now I'm just removing the ability to try to call this with a 1-d logical, rather than trying to fix the code. Fixes ESCOMP#24

AGonzalezNicolas added a commit to HPSCTerrSys/clm5_0 that referenced this issue Jun 27, 2024

Merge commit USE_PDAF: mpi_init call in cime_comp_mod switched off fo…

b117258

…r PDAF (ESCOMP#24) 'e41dc00a7e242bcb27b83f2cbe5c528b1227ba68' into eclm

AGonzalezNicolas pushed a commit to HPSCTerrSys/clm5_0 that referenced this issue Jul 5, 2024

Remove broken ncd_io_1d_log_glob

da92a85

For now I'm just removing the ability to try to call this with a 1-d logical, rather than trying to fix the code. Fixes ESCOMP#24

samsrabin added simple bfb bit-for-bit labels Aug 8, 2024

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

ncd_io_1d_log_glob is broken #24

ncd_io_1d_log_glob is broken #24

billsacks commented Dec 16, 2017

billsacks commented Dec 16, 2017

ncd_io_1d_log_glob is broken #24

ncd_io_1d_log_glob is broken #24

Comments

billsacks commented Dec 16, 2017

billsacks commented Dec 16, 2017