Pytdlpack

Bug Fixes

- Fixed an issue in `setup.py` that when building with Intel compilers, was causing pytdlack to incorrectly read TDLPACK random-access files.

Enhancements

- Updated `TdlpackIO` module to read more bytes from the record on file in order to get grid dimension information and store it in the index.


Bug Fixes

- Fixed issue where `TdlpackRecord` objects with a very small number of stations would cause a memory overflow in the Fortran `unpack` subroutine.

- Fixed issue where some unique values were missing from the `TdlpackIO.open` `dates` or `leadtimes` instance variables.

- Fixed issue with `TdlpackFile` where `eof` was not being set to `False` on file rewind.

- Fixed issue where `number_of_stations` instance variable would not be present for `TdlpackStationRecord` objects when reading using the `TdlpackIO` module.

Bug Fixes

- Fixed issue when instantiating `TdlpackRecord` object from new data (i.e. not from file) would cause an error when trying to pack and write.
- Added try/except block when setting projstring using `Proj.definition_string()` method. This seems to be a fairly new method to the Proj object.

No major changes to Python source code with this release. This release cleans up most of the project, including documentation, issue templates, and setup.py build and install process. **This is the first release of pytdlpack to be hosted on PyPI!**

shell
pip3 install pytdlpack


setup.py has been modified such that the default Fortran compiler flags used to build tdlpack are set to GNU Fortran, however one can also build and install using Intel Fortran Compiler.

Changes:

**pytdlpack:**
- Fixed issue 8 where setting an array size would lead to a TypeError.
- Some cosmetic changes -- code/documentation cleanup

**TdlpackIO:**
- No changes with this release

_**DISCLAIMER:**_ **TdlpackIO**, is an experimental module. It is a pure-Python implementation for fast, random access-like reading of TDLPACK sequential files. With TdlpackIO, no Fortran code is called for reading. This module is a work in progress. Please do not use it to perform critical work and its implementation and usage could change with future releases.

Changes:

**pytdlpack:**
- Fixed issue where a modified plain language string would not get packed into a new TDLPACK record.

**TdlpackIO:**
- Added ability to pass multiple dates and lead times (using list syntax) to the Tdlpack.open.fetch() method.

_**DISCLAIMER:**_ **TdlpackIO**, is an experimental module. It is a pure-Python implementation for fast, random access-like reading of TDLPACK sequential files. With TdlpackIO, no Fortran code is called for reading. This module is a work in progress. Please do not use it to perform critical work and its usage could change with future releases.