anaconda package (v 1.9.6) is broken (ubuntu, py37)

[Chitraketu]

The version of pyproj on anaconda is 1.9.6
This is a request to update it to version 2.1.3 (the version pip has)

The workaround is simply to install it using pip, however for some of us we are not even aware that pyproj is getting installed by anaconda so this is actually a request to update anaconda....

Thanks!

[jorisvandenbossche]

You will need to ask Anaconda about that (which you can do here: https://github.com/ContinuumIO/anaconda-issues/issues/).

But, this might not happen that fast yet: there are a lot of packages that depend on proj.4, which are not yet all updated to work with proj.4 6 (such as gdal). So they might want to wait on that (but, best ask there).
Alternatively, you can already install pyproj 2 from the conda-forge channel (https://anaconda.org/conda-forge/pyproj). But again, at this moment, you can't yet install it together with some other packages like gdal (and thus everything that depends on gdal like fiona and geopandas)

[ocefpaf]

Alternatively, you can already install pyproj 2 from the conda-forge channel (anaconda.org/conda-forge/pyproj). But again, at this moment, you can't yet install it together with some other packages like gdal (and thus everything that depends on gdal like fiona and geopandas)

BTW, conda-forge is migrating to proj4 6.0.0 and has pyproj 2.1.3. Most of the important packages were already rebuilt, including gdal 😉

See https://conda-forge.org/status/ for the migration status.

The only hiccup are the packages that depend on Qt (matplotlib, Jupyter, etc), b/c conda-forge is blocked on a new Qt build with latest openssl.

[jorisvandenbossche]

Cool!
A "migration" means that the already built packages are not yet available until all (or enough) are done? (to not break temporarily created environments in the middle of the rebuilding)

[ocefpaf]

A "migration" means that the already built packages are not yet available until all (or enough) are done? (to not break temporarily created environments in the middle of the rebuilding)

Not really. The packages are available right now. We rely on conda-build 3 package specs to provide conda with enough metadata to do the right thing. For example, b/c of the Qt issue I mentioned above, if you try:

conda create --name TEST python=3.7 gdal pyproj jupyter # or anything that depends on qt

you will get gdal 2.4.0, and pyproj 1.9.6 with proj4 5.2.0, but if you do:

conda create --name TEST python=3.7 gdal pyproj

you will get gdal 2.4.1, and pyproj 2.1.3 with proj4 6.0.0.

Things are getting more stable with conda as we move to making strict channel default and rebuild everything with conda-build 3.

[jorisvandenbossche]

Where is the time that I opened bug reports about broken environments because you updated gdal but fiona was still building .. :-)

[ocefpaf]

Where is the time that I opened bug reports about broken environments because you updated gdal but fiona was still building .. :-)

There are some other variables here that I did not mention, like bad pinning and bad package metadata (not to mention environments that are not using strict). However, those are "fixable problems" now with strict, and we are trying to identify them and fix as soon as possible.

[snowman2]

For reference: ContinuumIO/anaconda-issues#11400

[snowman2]

pyproj 2.4.1 is now available: https://anaconda.org/anaconda/pyproj