Releases: PCMDI/cmor3_documentation
Releases · PCMDI/cmor3_documentation
Release list
CMOR 3.6.0
Features
- Added the option
--hide-progressto PrePARE to remove the number of files completed while processing files. - Added the option
--no-text-colorto PrePARE to remove the color code from the terminal output. - Added the option
--versionto PrePARE to output the version of CMOR/PrePARE in use.
Changes
- CMOR now supports Python 3.8.
- Changed
Control VocabularytoControlled Vocabulary. - Improved error messages for supported data types in pywrapper.py.
- Removed gfortran 4.8.5 requirement for OSX.
- CMOR will now allow axis values to be within 1.e-6 fractional error for
valid_minandvalid_max. cmor_CV_checkSourceIDwill throw an error ifsourceis not found in asource_idsection of the CV.- In the function
cmor_CV_checkParentExpID, ifparent_experiment_idis "" or not found in the variable file, then it will be set to "no parent" if that is the valid value forparent_experiment_idin the CV. - When
parent_experiment_idis a value other than "no parent" in the functioncmor_CV_checkParentExpID, it will check the attributesparent_activity_id,branch_method,branch_time_in_child,branch_time_in_parent,parent_time_units,parent_variant_label,parent_source_id, andparent_mip_era, and only return an error after going through all of them.
Bugfixes
- Improved how comments in the
cell_methodsattribute are handled. - Fixed error message for
false_northingincmor_set_grid_attribute - Fixed issue with PrePARE not counting errors for attribute errors.
- Fixed errors when processing the
nwdFracLutvariable in theEmontable. - PrePARE will now check if
external_variablesexists when values are present incell_measures. - PrePARE will now give an error if the
activity_idis wrong.
Closed Issues
- Python 3.8 support
- Change "Control Vocabulary" to "Controlled Vocabulary"
- in PrePARE, logfile-friendly output format
- Handling comments in "cell_methods" in PrePARE
- Copy-&-paste error in error message on
false_northingandprojection_y_coordinatein cmor_grids.c
- Error message mis-match
- Recursive directory output issue / error vs warning
- Remove fortran compiler 4.8.5 requirement for OSX
- Error that doesn't cause failure when producing mrsos for Lmon
- nwdFracLut in Emon
- Unclear why Files SKIPPED
- Wrong activity_id warning (should be error?), etc..
- add --version command line argument to PrePARE
- Error handling in cmor_CV.c
CMOR 3.5.0
Changes
- CMOR now supports Python 3.6 and 3.7.
- PrePARE will stop if CMIP6_coordinate.json, CMIP6_formula_terms.json, and CMIP6_CV.json are missing from the table directory.
- Changed missing value replacement history message to include "corresponding data".
- The UUID library ossuuid has been replaced with libuuid
- CMOR will now use the json-c library installed from conda-forge instead of building its own library.
- In the Python API, CMOR will throw an error if the number of time steps being passed to
cmor.writeis less thanntimes_passed. - If the number of timesteps written is less than the length of the time axis, CMOR will give a warning when closing.
- cdms2 has been replaced with Netcdf4 for use in PrePARE.
Bugfixes
- Fixed bug in PrePARE where it crashes when it does not find the
_FillValueattribute in a file. - Fixed error caused by variables having both a singleton dimension and a generic level.
- Fixed bugs in PrePARE where it was not processing the attribute
cell_measurescorrectly. - Fixed bug caused by
cmor.writein the Python API when the number of time steps passed is less than thentimes_passedargument. - Fixed bug when handling the
olevhalfdimension. - Fixed error caused by variables having multiple singleton dimensions.
- Fixed bug when using CMOR in Append mode for instantaneous variables.
- Fixed bug caused by multiple instances of CMOR running a file permissions test by writing a test file. Each test file will now have the instance's PID in the file name.
- Fixed bug where the frequency of the variable used in the last call to
cmor_writeis applied to the output file name of all variables when callingcmor_close_variableorcmor_close.
Closed Issues
- Update cmor3 for python3
- PrePARE missing _FillValue exit
- cmor should stop if coordinates table is not present
- singleton and generic level
- PrePARE cell_measures changed in 3.4.0
- cmor.write segfault related to timesteps
- Question about handling of missing values in cmor_variable.c
- Bug in treating olevhalf
- switch from ossuuid to libuuid
- Two or more singleton dimensions (bs550aer)
- vendorizing json-c
- Append mode for instantaneous variables
- not passing enough times as data passes
- not writing enough times should issue a warning at exit
- Builds breaking from cdms2 issue related to libnetcdf
- Falsely alerting missing write permissions cmor.c#L1683
- CMOR using incorrect frequencies for variables when closing programs with multiple frequencies.
CMOR 3.4.0
Changes
- Fixed missing out_names_tests.json in PrePARE
- More robust comparison of missing_value and _FillValue between NetCDF files and CMIP6 tables in PrePARE.
- Variable name for time bounds is extracted from the NetCDF file in PrePARE.
- Time values and their bounds must be passed through cmor_write when CMOR is in append mode. Otherwise, an error is thrown.
Closed Issues
CMOR 3.3.3
CMOR 3.3.2
CMOR 3.3.1
CMOR 3.3.0
Accelerate PrePARE from 2s to 0.1 (IPSL request). Fix PrePARE/CMOR memory issue. Call Cdunif.so library directly instead of cdms2. Rremove fixed double array to dynamic double due to number of objects increasing. *Update license REGEX to allow "/" at the end of URL (GFDL request). * Update "Conventions" REGEX to allow "CMIP-6.0" up to "CMIP-6.2". Update grid tables. Remove standard_name to vertices_latitudes and vertices_longitudes for CF-1 compliance.