Closed ReimarBauer closed 4 years ago
Thanks for the report! Several users banded together and figured out how to get the conda-distributed version of jplephem upgraded; it was stuck on an old version.
Conda now looks to be up to 1.12: https://anaconda.org/conda-forge/jplephem
Could you try again and see if the conda-supplied version now fixes this error? Thanks!
Verified fixed. Thx :)
May be we could just add a new build number to this skyfield feedstock and add the dependency >=2.12, see https://github.com/conda-forge/skyfield-feedstock/issues/32
This would help a lot other projects.
2.12 (repeated the typo) ;)
Oh, now that I read this, your no-op update to the other feedstock makes more sense (except, I guess, that I didn't see it increment any build numbers?).
At this point, are things fixed for new installs — does installing Skyfield auto-install the new version of jplephem? In which case, a bump of the Skyfield package would help folks who had (a) installed Skyfield and gotten the old jplephem, (b) hadn't fixed it themselves, (c) hadn't uninstalled Skyfield, and (d) then run a conda-upgrade-all step after the bumped version comes out?
Did a missing save, I guess, sorry.
A new install of skyfield will use the newest build of jplephem. This is the simplest solution for our project where we only get jplephem with skyfield. Otherwise I have to add there the package with the fixation and to create a new build of our software. Later after a new release of skyfield I would have to undo this.
All having skyfield installed had in their dependency until the upgrade is triggered the old build.
But once they had an issue and try upgrade they get the newest version and not any other of jplephem.
From https://conda-forge.org/docs/maintainer/updating_pkgs.html
The build number is increased when the source code for the package has not changed but you need to make a new build. As a rule of thumb, the build number is increased whenever a new package with the same version needs to be uploaded to the conda-forge channel.
Examples for needing to increase the build number are
updating the pinned dependencies after a rerendering
Fixing wrong dependencies
When the package version changes you should reset the build number back to
Now that I see the dependency version number indeed changes in your diff, it makes perfect sense to me why a new release is needed. Thanks!
Issue: In a complete new env we can't import needed modules
Environment (
conda create -n pbenv jplephem
):Likly after the missing merges are fixed this will be solved. We found that pip install jplephem --upgrade works
see also: https://bitbucket.org/wxmetvis/mss/issues/515/modulenotfounderror-no-module-named