Closed scollis closed 5 years ago
It is likely that you have built the GDAL Python bindings without installing numpy-devel first.
@scollis See my comment https://github.com/ARM-DOE/pyart/pull/787#issuecomment-442337807. This is most likely not a conda-forge
issue, since numpy is pulled from defaults
channel with an rather old version (1.11.3). Unfortunately I can't see the complete conda list of that testenv
to figure out the provenience of gdal
.
AFAIK conda-forge
dropped creation of new Python2.7 packages and is based on Python3.6 and Python3.7 now. @ocefpaf can be more specific.
Please reopen if necessary.
In your env list above your gdal
is from defaults
so this is likely a bad mix of packages. If you can afford to use only conda-forge
, which most non-Windows users can, I'd recommend you to do that. If you are on Windows or need a package from defaults
try to use only defaults
, if you are on Windows and need package from both channels.... Sorry, no solution for it yet.
Perfect, thank you both. This is great information. I will try a conda-forge only env for testing. Failing this I will just turn off testing for GDAL deps in Py2.7. Thanks again, Scott
BTW just using conda-forge did not fix this.. but this is ok. I am now going to turn off our geotiff code for Py2.7
@scollis If you add gdal
to your environment-2.7.yml
then an all conda-forge
env is created.
Ooo! Ok. Good point.
From: Kai Mühlbauer notifications@github.com Reply-To: conda-forge/gdal-feedstock reply@reply.github.com Date: Wednesday, November 28, 2018 at 8:52 AM To: conda-forge/gdal-feedstock gdal-feedstock@noreply.github.com Cc: Scott scollis.acrf@gmail.com, Mention mention@noreply.github.com Subject: Re: [conda-forge/gdal-feedstock] Possible packing issue in python 2.7 (#238)
@scollis If you add gdal to your environment-2.7.yml then an all conda-forge env is created.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.
Wow. That worked. Thanks @kmuehlbauer Next time we meet it will need to be at a brewery as I owe you many beers
Wow. That worked. Thanks @kmuehlbauer
Note that if you remove defaults
(or conda-forge
) from your .condarc
you will probably get rid of this issue for good (well... until you need something from both channels :smile:).
Once conda-forge finishes the compiler migration and we start better practices for pkgs variants, using both channels together will be a better experience than it is now.
Issue: Build done incorrectly (possibly) for GDAL against numpy-devel
Environment (
conda list
):Details about
conda
and system (conda info
):See https://github.com/OSGeo/gdal/issues/1120