Skip to content

Conversation

@alasdairtran
Copy link
Contributor

This PR pins the Python version to prevent Conda from downgrading (e.g. from 3.7 to 3.6). It also updates two packages, numba and h5py, to make them compatible with Python 3.7. Fixes issue #768

@alasdairtran alasdairtran changed the title Upgrade ython in scipy notebook to 3.7 Upgrade Python in scipy-notebook to 3.7 Dec 22, 2018
@parente
Copy link
Member

parente commented Dec 24, 2018

Build fails consistently after 10 minutes of solving conda dependencies in the tensorflow image with no output. We can try merging and see what happens with the build on Docker Cloud, and revert if it causes trouble there.

@alasdairtran alasdairtran force-pushed the upgrade-python-in-scipy-notebook-to-3.7 branch from a9806d2 to fa09b38 Compare December 25, 2018 01:22
@alasdairtran
Copy link
Contributor Author

Build fails consistently after 10 minutes of solving conda dependencies in the tensorflow image with no output. We can try merging and see what happens with the build on Docker Cloud, and revert if it causes trouble there.

Actually I think Tensorflow hasn't officially supported Python 3.7 yet. We can track the issue here.

Let's wait for Tensorflow 1.13 to be released before merging my PR.

@parente parente added the tag:Upstream A problem with one of the upstream packages installed in the docker images label Dec 28, 2018
@parmentelat
Copy link

my understanding is that tensorflow was released as 1.13 a short while ago
is there anything that needs to be done explicitly to get this back on track ?

@parente
Copy link
Member

parente commented Mar 7, 2019

Resolved the merge conflict using the nice new GH editor for that, and kicked off a build to see where we stand.

@parente
Copy link
Member

parente commented Mar 7, 2019

SpecsConfigurationConflictError: Requested specs conflict with configured specs.
  requested specs: 
    - keras=2.2
    - tensorflow=1.12
  pinned specs: 
    - python=3.7
Use 'conda config --show-sources' to look for 'pinned_specs' and 'track_features'
configuration parameters.  Pinned specs may also be defined in the file
/opt/conda/conda-meta/pinned.

Next step is for the PR to update versions.

@parmentelat
Copy link

Hey folks; I'd love to see this come through, is there anything I can do to help out on this one ?

@parente
Copy link
Member

parente commented Mar 12, 2019

@parmentelat The work to be done to get a conda-forge build of tensorflow 1.13 w/ Python 3.7 support is tracked in conda-forge/tensorflow-feedstock#66. You can give that a read and see if there's a place where you can lend a hand.

@panmona
Copy link

panmona commented Apr 18, 2019

The referenced PR was merged 2 days ago ✨

@parente parente force-pushed the upgrade-python-in-scipy-notebook-to-3.7 branch from 93f1a29 to f067ea1 Compare April 21, 2019 20:50
@parente
Copy link
Member

parente commented Apr 21, 2019

Rebased on master and rebuilding with tf 1.13

@parente
Copy link
Member

parente commented Apr 22, 2019

Build is green. Thanks for the contributions and patience everyone!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

tag:Upstream A problem with one of the upstream packages installed in the docker images

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants