Releases: GEOS-ESM/MAPL
MAPL 2.17.0
This release of MAPL has bugfixes for a few issues:
- Fix for files that don't use
MAPL_UNDEF
for_FillValue
(#1305) - Fix for ungridded dims issue in History (#1311)
Testing shows that this is zero-diff for GEOS.
From CHANGELOG.md
Fixed
- io profiler is fixed due to the change of profiler finalize
- Files read by GriddedIOMod now respect the file defined _FillValue rather than assuming it is MAPL_UNDEF
- Fix bug so that fields with an ungridded dimension of size 1 can be written via History
- Updated
components.yaml
- ESMA_env v3.11.0 (Mainly updates for NAS)
Changed
- History initialization will display chunksize for any collections that are chunked
- Removed include of
ESMC_ReturnCodes.h
. Changed someESMC_RC
codes toESMF_RC
codes
What's Changed
- Fixes #1287. Output chunksize on init by @mathomp4 in #1303
- fixes #1294 by @bena-nasa in #1305
- Bugfix/wjiang/fix io profiler by @weiyuan-jiang in #1308
- Hotfix/bmauer/fixes #1311 by @bena-nasa in #1313
- Update ESMA_env to v3.11.0 by @mathomp4 in #1315
- Fixes #1295. Clean up ESMC use by @mathomp4 in #1304
- Update CMakeLists.txt and CHANGELOG for 2.17.0 Release by @mathomp4 in #1317
- Merge Develop into Main by @mathomp4 in #1319
Full Changelog: v2.16.0...v2.17.0
MAPL 2.8.0.2
This release of MAPL is a patch on then MAPL 2.8.0 series. This update changes MAPL_SimpleBundleRead
from using MAPL_CFIORead
to the more modern MAPL_read_bundle
so that the ESMF regridding infrastructure is used.
From CHANGELOG.md
Changed
- Changed MAPL_SimpleBundleRead from using MAPL_CFIORead to the more modern MAPL_read_bundle so that the ESMF regridding infrastructure is used
Full Changelog: v2.8.0.1...v2.8.0.2
MAPL 2.16.0
This is a zero-diff (for GEOS) release of MAPL with a few bugfixes. There were fixes to splitting fields in History as well as correctly halting the program when bad chunksizes were passed into History.
From CHANGELOG.md
Fixed
- A bug in splitting the fields for History when an alias contains a single entry (#1296)
- Trap more errors in pFIO
- Free types and operation created by profiler
- Finalize profiler in MAPL_finalize call
- Fix how a retry-build is done in CI
- Fix issue with logger message and single quotes
- Fix bug with profiler reporting in programs
Changed
- Updated README.md
- Refactored MAPL_Generic.F90 - lots of changes.
- consistent indentation (emacs mode)
- extracted helper procedures in GenericInitialize
- Added new step to CircleCI to try and re-run build on failure.
- Updated CircleCI config to use new GEOS-ESM/circleci-tools orb
- Added Markdown rules to
.editorconfig
What's Changed
- Minor refactoring - indentation changed. by @tclune in #1267
- Feature/tclune/generic refactor jan05 2022 by @tclune in #1273
- Handmerge main into develop 2022-Jan-06 by @mathomp4 in #1281
- Auto GitFlow - main → develop by @github-actions in #1272
- Add retry to CI build by @mathomp4 in #1279
- initailize and finalize profiler in MAPL by @weiyuan-jiang in #1261
- Fixes #1268 - update to README.md by @tclune in #1286
- Fixes #1292. Fix for Logger Message by @mathomp4 in #1293
- Bugfix/atrayano/#1296 split field alias by @atrayano in #1297
- refactoring pfio error trapping by @weiyuan-jiang in #1290
- Add "latest version" Zenodo badge to readme by @mathomp4 in #1289
- Update CircleCI CI to use Orb by @mathomp4 in #1288
- fix bug in profiler reporting caused by recent mapl change by @mathomp4 in #1301
- Merge Develop into Main for 2.16.0 release by @mathomp4 in #1300
Full Changelog: v2.15.1...v2.16.0
MAPL 2.15.1
NOTE: Non-zero-diff for History output compared to 2.15.0 when comparing metadata and lat
and lon
. The actual fields' data themselves are zero-diff, but the metadata with those variables are different. Also, lat
and lon
were made 64-bit. Again, the same data, but now at a different precision.
This release of MAPL fixes issues with the history output from MAPL. Namely, the History output more closely follows the GEOS FP 1.2 File Specification. The changes include:
- Coordinate Variables
lon
andlat
are now 64-bit (double) in lat-lon History files rather than 32-bit (single)lons
,lats
,corner_lons
, andcorner_lats
are now 64-bit (double) in cubed-sphere History files rather than 32-bit (single)
- Variable Metadata
- Added
fmissing_value
(equal to_FillValue
akaMAPL_UNDEF
) - Added
missing_value
(equal to_FillValue
akaMAPL_UNDEF
) - Added
vmin
(equal to-MAPL_UNDEF
) - Added
vmax
(equal to+MAPL_UNDEF
) - Added
add_offset
(equal to 0.0) - Added
scale_factor
(equal to 1.0) - Added
standard_name
(equal tolong_name
)
- Added
- Global Metadata
- Added
Title
,History
,Source
,Contact
,Convention
,Institution
,References
,Filename
,Comment
- These currently have hardcoded values roughly equivalent to the GEOS FP 5.27 output
- Added
What's Changed
- Hotfix/bmauer/fixes #1269 off main by @bena-nasa in #1271
- Prepare for 2.15.1 release by @mathomp4 in #1274
Full Changelog: v2.15.0...v2.15.1
MAPL 2.8.0.1
NOTE: Non-zero-diff for History output compared to 2.8.0 when comparing metadata and lat
and lon
. The actual fields' data themselves are zero-diff, but the metadata with those variables are different. Also, lat
and lon
were made 64-bit. Again, the same data, but now at a different precision.
Changes to make MAPL 2 History output match GEOS FP file spec
- Coordinate Variables
lon
andlat
are now 64-bit (double) in lat-lon History files rather than 32-bit (single)lons
,lats
,corner_lons
, andcorner_lats
are now 64-bit (double) in History files rather than 32-bit (single)time
is now an integer if the History is output at a frequency that is not a fraction of the time unit
- Variable Metadata
- Added
fmissing_value
(equal to_FillValue
akaMAPL_UNDEF
) - Added
missing_value
(equal to_FillValue
akaMAPL_UNDEF
) - Added
vmin
(equal to-MAPL_UNDEF
) - Added
vmax
(equal to+MAPL_UNDEF
) - Added
add_offset
(equal to 0.0) - Added
scale_factor
(equal to 1.0) - Added
standard_name
(equal tolong_name
)
- Added
- Global Metadata
- Added
Title
,History
,Source
,Contact
,Convention
,Institution
,References
,Filename
,Comment
- These currently have hardcoded values roughly equivalent to the GEOS FP 5.27 output
- Added
Full Changelog: v2.8.0...v2.8.0.1
MAPL 2.15.0
This release of MAPL as a few bugfixes.
First, MAPL now allows a HISTORY.rc
file without an EXPDSC:
or EXPID:
specified (will use an empty string if not provided).
Second, this fixes an issue when build GEOS with FV_PRECISION=R8
(e.g., when building GEOSctm)
From CHANGELOG.md
Fixed
- Free types and operation created by profiler
Changed
- Refactored MAPL_Generic.F90 - lots of changes.
- Make EXPID and EXPDSC optional. Default is empty string
- Updated
changelog-enforcer
to version 3 - Compress CircleCI artifacts
- Updated VarRead_2d_r8 interface to match VarRead_2d_r4 interface in NCIO module
What's Changed
- Update to changelog-enforcer 3 by @mathomp4 in #1254
- Compress CircleCI artifacts by @mathomp4 in #1257
- make EXPID and EXPDSC optional by @weiyuan-jiang in #1256
- Feature/wjiang/free types by @weiyuan-jiang in #1260
- Update VarRead 2D_R8 interface to match VarRead 2D_R4 by @bena-nasa in #1258
- Feature/tclune/refactor generic setservices by @tclune in #1262
- Prepare for 2.15.0 Release by @mathomp4 in #1263
- Merge Develop into Main for 2.15.0 release by @mathomp4 in #1264
Full Changelog: v2.14.1...v2.15.0
MAPL 2.14.1
This is a bugfix patch release of MAPL that fixes a GNU-related bug in 2.14.0.
It also brings in an update to components.yaml
to match that in GEOSgcm.
From CHANGELOG.md
Fixed
- gfortran can not associate an allocated string. Such blocks are changed
Changed
- Updated
components.yaml
- ESMA_env v3.8.0 (Use Intel 2021.3)
- ESMA_cmake v3.8.0 (Use
-march=core-avx2
for Intel Fortran) - These are non-zero-diff for GEOS
- Updated the Intel CI image to Intel 2021.3
What's Changed
- Update to ESMA_env 3.8.0 and ESMA_cmake 3.8.0 by @mathomp4 in #1240
- change some associate block to fix GNU crash by @weiyuan-jiang in #1245
- Update CHANGELOG and CMakeLists for 2.14.1 Release by @mathomp4 in #1247
- Merge develop onto main for 2.14.1 release by @mathomp4 in #1248
Full Changelog: v2.14.0...v2.14.1
MAPL 2.14.0
This release of MAPL has many updates:
- Bugfixes for the GEOSldas (see #1226 and #1237).
- Newer-style timers from MAPL
- Updates for
mapl_tree.py
to make it work - Other refactoring
All testing with GEOS shows this to be zero-diff.
From CHANGELOG.md
Fixed
- Move out allocatable string from if condition block in
VarConn.F90
- Updates to
mapl_tree.py
to let it work in the git GEOS- Note that the interface has changed since the last time this worked, please see script help usage
- Disabled the
chname
style of running it as that does not work at the moment
Changed
- Refactored support for using DSO's for components. No change to interfaces.
- Updated MAPL to exclusively use new timers - with improved format.
Removed
- Legacy timer profiling
Deprecated
- An interface for
MAPL_AddChild
allowing specification of a DSO has
been deprecated due to non-conventional ordering of its arguments. A
new interface with conventional ordering has been introduced.
What's Changed
- Feature/wjiang/add open demo by @weiyuan-jiang in #1223
- Major refactoring of profiling by @tclune in #1222
- Feature/tclune/refactor generic by @tclune in #1225
- Fixes #1230. Update mapl_tree.py by @mathomp4 in #1231
- Added --repo option to show hierarchy of repos by @mathomp4 in #1235
- Feature/wjiang/get short name work around by @weiyuan-jiang in #1237
- Merge develop into main for 2.14.0 release by @mathomp4 in #1238
Full Changelog: v2.13.0...v2.14.0
MAPL 2.10.1
This release of MAPL is a bug-fix on 2.10.0. This is needed for GEOSldas development.
This backports a fix to ShaveMantissa.c
into the 2.10 codebase.
From CHANGELOG.md
Fixed
- Return 0 when there is no data for bit shave
Full Changelog: v2.10.0...v2.10.1
MAPL 2.13.0
This zero-diff release of MAPL is mainly for a bugfix for bit-shaving as used in the GEOSldas. In that model, there were possible grid cells with no data and bit shaving acted badly (returned error). This release fixes that.
Beyond that are other updates for IOserver and return code handling.
From CHANGELOG.md
Fixed
- Return 0 when there is no data for bit shave
- Removed tab characters from Fortran (and C) code
Added
- Add the i and j index as variables to use to generate synthetic data in ExtDataDriver.x
- Added ability to generate monthly checkpoints (fixes issue #1065)
Changed
- Changed the way how a writer is chosen. Previously, a writing processor is chosen as long as it is idle.
Now, an idle processor is chosen from a node with the most idle processors. - Changed error checking
_ASSERT
to use__RC__
macro and_VERIFY
for UserRC - Changed
_ASSERT
with.and.
conditional to separate_ASSERT
to improve error message - Changed usage of MAPL_IO subroutines in CubedSphere and LatLon Grid Factories to open command with newunit clause
- Updated
components.yaml
- ESMA_env v3.7.0 (Use MPT 2.25 at NAS on TOSS)
- ESMA_cmake v3.7.2 (Fixes FindBaselibs issue found by @sdrabenh, f2py order fix)
- Made the
MAPL_AddChildFromDSO
function system agnostic by using the CMake detected DSO suffix
What's Changed
- Fixes #1065. Added ability to generate monthly checkpoint files. by @atrayano in #1180
- Update to ESMA_env v3.6.0, ESMA_cmake v3.7.1 by @mathomp4 in #1182
- Update to ESMA_cmake v3.7.2 by @mathomp4 in #1185
- Handmerge main into develop - 2021-Nov-09 by @mathomp4 in #1191
- Removes MAPL_IO dependency from CubedSphere and LatLon GridFactories by @Gvilla1000-nasa in #1197
- new synthetic data option in ExtDataDriver.x by @bena-nasa in #1203
- Fixes #1205. Make AddChildFromDSO OS agnostic by @mathomp4 in #1206
- normal return for no data bit shaving by @weiyuan-jiang in #1211
- Changes usage of _ASSERT for SUCCESS error checking and _ASSERT with and conditional by @Gvilla1000-nasa in #1208
- Feature/wjiang/optimized writing node by @weiyuan-jiang in #1217
- Remove tab characters from Fortran and C code by @mathomp4 in #1207
- Update for MAPL 2.13.0 Release by @mathomp4 in #1218
- Merge develop into main for 2.13 Release by @mathomp4 in #1220
Full Changelog: v2.12.1...v2.13.0