Skip to content

Sane numpy#1

Merged
souravsingh merged 9 commits intoconda-forge:masterfrom
kain88-de:sane-numpy
Aug 25, 2017
Merged

Sane numpy#1
souravsingh merged 9 commits intoconda-forge:masterfrom
kain88-de:sane-numpy

Conversation

@kain88-de
Copy link
Copy Markdown
Contributor

switch to a ABI compatible numpy configuration and add python 3.4 to build matrix.

@conda-forge-linter
Copy link
Copy Markdown

Hi! This is the friendly automated conda-forge-linting service.

I wanted to let you know that I linted all conda-recipes in your PR (recipe) and found some lint.

Here's what I've got...

For recipe:

  • The recipe license should not include the word "License".

@peterjc
Copy link
Copy Markdown
Contributor

peterjc commented Aug 8, 2017

I assume this is part of a general conda-forge numpy update effort (the details of I do not understand and am not commenting on).

Good catch on checking importing of Bio.Cluster and Bio.KDTree in the tests which are both dependent on compiled C code.

Since you are working on this, can you also add Bio.Affy, Bio.phenotype, Bio.cpairwise2, Bio.trie to the import tests as well (again C code, but not all using NumPy itself).

Cross reference bioconda/bioconda-recipes#5465 for fixing the same oversight in the older BioConda recipe which this conda-forge recipe was based on.

@kain88-de
Copy link
Copy Markdown
Contributor Author

@ocefpaf there seems to be a numpy/python issue on windows with python 3.5.

I assume this is part of a general conda-forge numpy update effort (the details of I do not understand and am not commenting on).

Yes the idea is that we can update numpy without having to rebuild all packages depending on numpy when a new numpy version is released.
see: conda-forge/conda-forge.github.io#415

@peterjc
Copy link
Copy Markdown
Contributor

peterjc commented Aug 8, 2017

@kain88-de Understood, thank you. We're trying to do something similar with the Biopython wheels for PyPI where we build against the oldest numpy as possible for each Python, so our wheels should work with any more recent numpy: https://github.com/biopython/biopython-wheels

@ocefpaf
Copy link
Copy Markdown
Member

ocefpaf commented Aug 8, 2017

@ocefpaf there seems to be a numpy/python issue on windows with python 3.5.

That is my bad. When I wrote the docs I was sure I saw numpy 1.8 for Python 3.5 in defaults. We need to use 1.9 there 😒

I am working to get numpy for Windows on conda-forge so we can have control over this.

Comment thread recipe/meta.yaml
- numpy x.x
- numpy 1.8.* # [not (win and (py35 or py36))]
- numpy 1.9.* # [win and py35]
- numpy 1.11.* # [win and py36]
Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@kain88-de I guess that this is the current combination. I don't really want to spread this at the moment. I'd rather work fast to get older versions of numpy in the channel so we can simplify this ASAP.

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I can wait for numpy 1.8 on windows. I rather have a correct long-term solution for the maintainers. I also changed these to see if I can break things in the build ;-)

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Cool. We'll have a meeting next Friday and I'll try to ask for a numpy 1.8 and 1.9 on defaults for both py35 and py36.

@souravsingh
Copy link
Copy Markdown
Contributor

@kain88-de Is the patch ready for merging or should we wait until NumPy 1.8 and 1.9 is available in the default channel?

@kain88-de
Copy link
Copy Markdown
Contributor Author

@ocefpaf how are the default channels looking?

@souravsingh depends on you. If we wait you don't need to change this in the future.

@ocefpaf
Copy link
Copy Markdown
Member

ocefpaf commented Aug 14, 2017

@ocefpaf how are the default channels looking?

Promising. We discussed that last meeting and people are looking into it. I'll know more next week.

@kain88-de
Copy link
Copy Markdown
Contributor Author

@ocefpaf any change on this?

@ocefpaf
Copy link
Copy Markdown
Member

ocefpaf commented Aug 25, 2017

@ocefpaf any change on this?

Nope. I'll ask again today but maybe we should not wait for that...

@kain88-de
Copy link
Copy Markdown
Contributor Author

@souravsingh You can merge this if you like.

@souravsingh souravsingh merged commit 1b4c162 into conda-forge:master Aug 25, 2017
@kain88-de kain88-de deleted the sane-numpy branch August 25, 2017 15:22
@kain88-de
Copy link
Copy Markdown
Contributor Author

The merge is currently failing because there is an issue with the anaconda token.

@kain88-de
Copy link
Copy Markdown
Contributor Author

conda-forge/conda-forge.github.io#426 (comment)

looks like we can wait until there pops up a PR to fix the issue.

@ocefpaf
Copy link
Copy Markdown
Member

ocefpaf commented Aug 25, 2017

The short-term solution is to re-submit the recipe to staged-recipes. It won't change the recipe here, only the tokens.

@souravsingh
Copy link
Copy Markdown
Contributor

@ocefpaf Do I now have to submit a PR to staged-recipes with the new recipe? If so, What would happen to this feedstock repository?

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

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

5 participants