search

shenyulu / easyclimate

A line of code to analyze climate
https://easyclimate.readthedocs.io/en/latest/
Other
28 stars 2 forks source link

Installation method with conda #72

Open zhangslTHU opened 1 month ago

zhangslTHU commented 1 month ago

This is a really nice Python package with (more) promising applications, and I have some questions regarding its use:

  1. Would it be possible to provide an installation method with conda? This would facilitate package management and simplify the installation process.

  2. What are the differences between windspharm-syl and windspharm? It seems they are the same, and there are some issues with windspharm as it has not been updated and maintained for a long time. Additionally, several functions within NumPy have been deprecated or removed, which may lead to warnings and errors when utilizing the windspharm. Are there any plans for future updates to windspharm-syl?

  3. According to the windspharm-conda website, windspharm is only supported for installation in Linux and Mac environments. While there are examples that it can be installed on Windows, this support is limited to versions prior to python 3.9. So i am curious about the easyclimate supporting environment.

shenyulu commented 2 weeks ago

Thank you for your kind words about the Python package! I will provide appropriate responses to your questions based on each of your points.

  1. Certainly, the conda installation method is a simplified way to install. We plan to incorporate this method in future designs.
  2. (1) There is no difference between windspharm-syl and windspharm. However, the Fortran dependency in windspharm, pyspharm, has been specially modified to pyspharm-syl to ensure compatibility with both Windows and Linux systems. (2) In fact, there are currently no major issues with the NumPy dependency in windspharm. It can be addressed after future upgrades to NumPy. You can also fork the source code repository to handle these issues in NumPy. Overall, the current problems are not severe.
  3. Fortran code can be compiled and run on both Linux and Windows. Fortran code that does not heavily rely on system libraries can be easily compiled across platforms. For example, you can use GCC (or MinGW on Windows) or Intel Fortran.