conda-forge / great-expectations-feedstock

A conda-smithy repository for great-expectations.
BSD 3-Clause "New" or "Revised" License
4 stars 12 forks source link

2 possible package resolutions #76

Open juanfcocontreras opened 3 years ago

juanfcocontreras commented 3 years ago

Since the last update, I have a conflict to update:

Collecting package metadata (current_repodata.json): done
Solving environment: |
Warning: 2 possible package resolutions (only showing differing packages):
  - conda-forge/noarch::great-expectations-0.7.11-py_0, conda-forge/noarch::pyfiglet-0.8.post1-py_0, conda-forge/noarch::pypandoc-1.5-pyh9f0ad1d_0, conda-forge/noarch::requests-2.24.0-pyh9f0ad1d_0
  - conda-forge/noarch::jsonpatch-1.24-py_0, conda-forge/noarch::jsonpointer-2.0-py_0, conda-forge/noarch::requests-2.23.0-pyh8c360ce_2, conda-forge/osx-64::great-expectations-0.12.4-py37hc8dfbb8done
jcampbell commented 3 years ago

Interesting -- the command run was specifically to update Great Expectations?

juanfcocontreras commented 3 years ago

No, I was doing:

conda update --all

And every time I run that command it switches from one resolution way to the other one.

If I remove great-expections the update all command works without any issue.

xhochy commented 3 years ago

Warnings shouldn't be an issue. This just shows you that the solving algorithm of conda has found two solutions that are equal to its cost measures.

Did you get an actual error?

juanfcocontreras commented 3 years ago

Warnings shouldn't be an issue. This just shows you that the solving algorithm of conda has found two solutions that are equal to its cost measures.

Did you get an actual error?

My problem is that every few days I run the command conda update --all to have all the packages updated.

Right now every time I run that command, great-expectations version changes from version 0.7.11 to 0.12.4 (and its dependencies), and vice versa (at the next execution of the command).

I don't know whether to consider it an error or not, but it is certainly not a desired behavior nor does it happen with the rest of the installed packages.

juanfcocontreras commented 3 years ago

I think the problem is with dependency requests. Recently, version 2.24 of requestshas been added to conda-forge.

In version 0.7.11 of great-expectations the restriction is: - requests >=2.20

While in recent versions of great-expectations the restriction is: - requests >=2.20,<2.24

I think that the problem is that in one execution conda updates requests to version 2.24 and downgrades great-expectations to version 0.7.11 while in the next execution conda updates great-expectations to version 0.12.4 while downgrades requests to version 2.23.

The question would be if actually great-expectations is incompatible with version 2.24.

jcampbell commented 3 years ago

@juanfcocontreras I just created this PR: https://github.com/great-expectations/great_expectations/pull/2020 to try to update our version requirements.

As I recall, the issue with requests 2.24 was actually caused by an incompatibility with the snowflake driver for SQL. For users not using that driver, I suspect using 2.24 would be fine. See: https://github.com/great-expectations/great_expectations/pull/1623.

I don't know whether snowflake has addressed the issue: https://github.com/snowflakedb/snowflake-connector-python/issues/324

juanfcocontreras commented 3 years ago

@juanfcocontreras I just created this PR: great-expectations/great_expectations#2020 to try to update our version requirements.

As I recall, the issue with requests 2.24 was actually caused by an incompatibility with the snowflake driver for SQL. For users not using that driver, I suspect using 2.24 would be fine. See: great-expectations/great_expectations#1623.

I don't know whether snowflake has addressed the issue: snowflakedb/snowflake-connector-python#324

Great! Thank you!

juanfcocontreras commented 3 years ago

It seems that the PR has some issues. Could you check it, @jcampbell ?

Thanks in advance!