xhochy
commented
5 years ago @michaelaye This should be fixed, can you retest?
Closed michaelaye closed 5 years ago
xhochy
commented
5 years ago @michaelaye This should be fixed, can you retest?
michaelaye
commented
5 years ago Not here (on macOS):
$ conda update --all (py37)
Collecting package metadata: done
Solving environment: done
## Package Plan ##
environment location: /Users/klay6683/miniconda3/envs/py37
The following packages will be downloaded:
package | build
---------------------------|-----------------
arrow-cpp-0.12.1 | py37h2ddd891_0 5.5 MB conda-forge
libarchive-3.3.3 | h5c473cc_1005 1.4 MB conda-forge
libprotobuf-3.6.1 | hd9629dc_1001 3.8 MB conda-forge
libtiff-4.0.10 | hd08fb8f_1003 527 KB conda-forge
pyarrow-0.12.1 | py37hf9e6f3b_0 2.0 MB conda-forge
------------------------------------------------------------
Total: 13.3 MB
The following packages will be UPDATED:
libarchive 3.3.3-h46c8594_1004 --> 3.3.3-h5c473cc_1005
libtiff 4.0.10-h344961e_1002 --> 4.0.10-hd08fb8f_1003
parquet-cpp 1.5.1-2 --> 1.5.1-4
zstd 1.3.3-1 --> 1.4.0-ha9f0a20_0
The following packages will be DOWNGRADED:
arrow-cpp 0.13.0-py37h29c365f_3 --> 0.12.1-py37h2ddd891_0
libprotobuf 3.7.1-hfbae3c0_0 --> 3.6.1-hd9629dc_1001
pyarrow 0.13.0-py37hb37e6aa_0 --> 0.12.1-py37hf9e6f3b_0
Proceed ([y]/n)?@michaelaye Can you give this another try?
michaelaye
commented
5 years ago Conda offers me the exact same build, so why should this be improved? I get the exact same feedback as on June 10.
michaelaye
commented
5 years ago and the last commit on this feedstock was on April 30, are you pulling my leg?
xhochy
commented
5 years ago No, i've spent a considerable amount of fixing into this. The problem is not fixed by editing this feedstock but by the infrastructure consuming it.
This should be fixed by updating zstd, arrow-cpp and pyarrow.
michaelaye
commented
5 years ago So you did, thanks! I had to reinstall my conda env to make this work. Guess the previous pinning messed up the dependency tree.
michaelaye
commented
5 years ago Alas, it still isn't:
$ conda list pyarrow(py37)
# packages in environment at /Users/klay6683/miniconda3/envs/py37:
#
# Name Version Build Channel
pyarrow 0.11.1 py37hf9e6f3b_1002 conda-forge
klay6683@macl3886|~/s/dans-gdal-scripts (master|✔)
$ conda update pyarrow(py37)
Collecting package metadata: done
Solving environment: done
## Package Plan ##
environment location: /Users/klay6683/miniconda3/envs/py37
added / updated specs:
- pyarrow
The following packages will be UPDATED:
arrow-cpp 0.11.1-py37h3bd774a_1 --> 0.13.0-py37h29c365f_3
pyarrow 0.11.1-py37hf9e6f3b_1002 --> 0.13.0-py37hb37e6aa_0
thrift-cpp pkgs/main::thrift-cpp-0.11.0-hd79cdb6~ --> conda-forge::thrift-cpp-0.12.0-hdedcb34_1002
The following packages will be DOWNGRADED:
libtiff 4.0.10-hd08fb8f_1003 --> 4.0.10-h344961e_1002
zstd 1.4.0-ha9f0a20_0 --> 1.3.3-1
Proceed ([y]/n)? nMaybe the corner case is really my conda-forge based pinning?
$ conda config --show-sources (py37)
==> /Users/klay6683/.condarc <==
pinned_packages: []
report_errors: True
anaconda_upload: True
==> /Users/klay6683/miniconda3/envs/py37/.condarc <==
pinned_packages:
- conda-forge::opencv
- conda-forge::gdal
- conda-forge::numpy
channels:
- conda-forge
- defaultsThis is actually solved by updating to conda 4.7.5 that was just released, including my strict pinning to conda-forge!
$ conda list zstd (py37)
# packages in environment at /Users/klay6683/miniconda3/envs/py37:
#
# Name Version Build Channel
zstd 1.4.0 ha9f0a20_0 conda-forge
klay6683@macl3886|~/D/s/c/p4catalog (master|✚2)
$ conda list pyarrow (py37)
# packages in environment at /Users/klay6683/miniconda3/envs/py37:
#
# Name Version Build Channel
pyarrow 0.13.0 py37he1943e6_2 conda-forge
klay6683@macl3886|~/D/s/c/p4catalog (master|✚2)
$ conda update --all (py37)
Collecting package metadata (current_repodata.json): done
Solving environment: done
# All requested packages already installed.
Issue: Previous versions of
zstdhad no dependencies. The recently (for versions >1.3.3) added dependencies forcecondato downgradepyarrowfrom 0.13 to 0.12.1. Is that intended?Environment (
conda list):Details about
condaand system (conda info):