Vapor with python > 3.9

1

Greetings!

Several factors (including EOL for security patches next month) are contributing to a strong desire to stop using python 3.9, but as far as I can see, the latest vapor-python release still requires this version. I’m curious if anyone can comment on the reason for this, because if it is lack of resources and/or interest, then I’ll likely proceed with trying to migrate to 3.12 or 3.13, but if there is something more fundamental that is blocking use of an updated python, I’ll have to consider other options.

Has anyone else already made this transition?

Best regards,

Jim McCaa
Vaisala OYJ

2

Hi, the latest vapor-python release uses python 3.10. Maybe you could try reinstalling vapor in a clean environment? Would you mind sharing what your use case for Vapor is?

3

Greetings,

Thanks for the reply! I’m currently using vapor for visualizing WRF output, and this works fine with python 3.9, but I’d like to expand its usage to visualize output from some newer models by importing from xarray/zarr, and these datasets use sufficiently modern compression algorithms that backporting to 3.9 would be a big challenge.

For me, conda reports the latest available version of vapor as 3.10.0, with the following dependencies (note python 3.9). Perhaps I’m installing from the wrong channel? I have been following the instructions at Download and Installation — 3.10.1 documentation

Best regards,

Jim

vapor 3.10.0 h2bc3f7f_0
-----------------------
file name   : vapor-3.10.0-h2bc3f7f_0.tar.bz2
name        : vapor
version     : 3.10.0
build       : h2bc3f7f_0
build number: 0
size        : 6.4 MB
subdir      : linux-64
url         : https://conda.anaconda.org/ncar-vapor/linux-64/vapor-3.10.0-h2bc3f7f_0.tar.bz2
md5         : e11067f91366bd0a9f54aed3c75f6a65
timestamp   : 2025-04-29 22:18:24 UTC
dependencies:
  - anywidget
  - assimp >=5.0.1,<6.0a0
  - bqplot
  - cppyy 2.2.*
  - expat >=2.4.9,<3.0a0
  - freetype >=2.10.4,<3.0a0
  - geotiff >=1.6.0,<2.0a0
  - glm >=0.9.9.4,<1.0a0
  - hdf5 >=1.12.1,<2.0a0
  - hdf5plugin 4.*
  - ipython 8.*
  - jpeg >=9e,<10a
  - jupyter 1.*
  - libcxx >=14.0.6,<15.0a0
  - libgcc >=14
  - libstdcxx >=14
  - libtiff =4.3|4.4
  - matplotlib 3.*
  - mesalib >=21.2.5,<22.0a0
  - netcdf4 >=1.5.8,<2.0a0
  - numpy >=1.21.4,<2.0a0
  - proj >=7.2.0,<8.0a0
  - py-opencv 4.*
  - python 3.9.*
  - udunits2 >=2.2.28,<3.0a0
  - vapor-maps 1.0.*
  - wurlitzer
  - xarray 2022.*
4

Looks like 3.10.1 has not been released for conda, at least not per this source:

https://conda.anaconda.org/ncar-vapor/linux-64/

which lists the latest as 3.10.0. Since some form of 3.10.1 came out in May, this suggests it’s not going to happen… anyone had success just building the python packages outside of the conda releases?

5

Hi,

Sorry for the confusion. The current release version is 3.10 which uses python 3.9, and the current dev version is 3.10.1 which uses python 3.10. This version which uses python 3.10 will be released by December or sooner.

In the meantime, if you follow the standard Conda pkg build instructions, you should be able to run it on the vapor/conda/vapor recipe dir and get a conda pkg with the current version which supports python 3.10.