Add a recipe for pydata/sparse #5631

hameerabbasi · 2018-04-11T13:07:21Z

This is to add a recipe for pydata/sparse.

As an aside, can we install the requirements for py.test and run it as a command, if possible?

Edit: And how do I specify minimum versions?

conda-forge-linter · 2018-04-11T13:07:29Z

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

I just wanted to let you know that I linted all conda-recipes in your PR (recipes/sparse) and found it was in an excellent condition.

synapticarbors · 2018-04-12T16:03:58Z

recipes/sparse/meta.yaml

+    - pip
+  run:
+    - python
+    - numpy >= 1.14


remove the space between the logical sign and the version number, i.e:

- numpy >=1.14

synapticarbors · 2018-04-12T16:05:27Z

recipes/sparse/meta.yaml

+  description: |
+    sparse implements sparse multidimensional arrays on top of Numpy
+    and scipy.sparse. It tries to mimic Numpy's ndarray API.
+  doc_url: https://sparse.pydata.org/en/0.3.0/


Maybe drop the version number from the doc_url? I'm afraid it might not get updated when the recipe does since it's buried far down in the recipe.

I could, but since this package is currently not mature I expect the information on there to get out of sync with what's in the recipe quite quickly, which is why I added a link to the release version.

synapticarbors · 2018-04-12T16:06:03Z

recipes/sparse/meta.yaml

+extra:
+  recipe-maintainers:
+    - hameerabbasi
+    - mrocklin


@mrocklin -- are ok with being a co-maintainer of the sparse recipe on conda-forge?

Yes, I am ok with being a co-maintainer of the sparse package on conda-forge.

Additionally lets ping @jakevdp to see if he is interested.

Sure, sounds good

synapticarbors · 2018-04-12T16:08:16Z

recipes/sparse/meta.yaml

+    - scipy >= 0.19
+    - numba >= 0.37
+
+test:


You can run pytest here using the command section of test:

https://conda.io/docs/user-guide/tasks/build-packages/define-metadata.html#test-commands

You'll need to add pytest and any other testing-only requirement in the requires section:

https://conda.io/docs/user-guide/tasks/build-packages/define-metadata.html#test-requirements

Also being able to run the test suite depends on the test files being included in the source distribution.

hameerabbasi · 2018-04-12T16:27:39Z

Well, this is weird. Do I need to add all requirements to the test section? It seems it can't import Numpy, even though Numpy is in the run section. Should I move it to build?

synapticarbors · 2018-04-12T16:33:31Z

Anything in the run section should automatically be available in the test section. I think you need numpy in the build section since in the setup.py you import the package which will necessitate all of the run dependencies. Doesn't look like that happens once you switch to the version that uses versioneer.

synapticarbors · 2018-04-12T16:41:20Z

You're going to want the requirements under both build and run; they'll just duplicated.

jakevdp · 2018-04-12T16:50:37Z

Looks like the MANIFEST.in in the package needs to be updated: it specifies LICENSE.txt rather than LICENSE.rst, and so the license file is not included in the distribution available on PyPI https://github.com/pydata/sparse/blob/master/MANIFEST.in

hameerabbasi · 2018-04-12T17:34:40Z

I pushed 0.3.1 to PyPI with an updated MANIFEST.in. Unfortunately, it seems RTD isn't picking up the tag/building documentation.

hameerabbasi · 2018-04-13T06:57:51Z

Hmm. It seems I need to run py.test in another directory... Can someone point me how to cd into the package directory?

mrocklin · 2018-04-13T14:14:43Z

recipes/sparse/meta.yaml

+    - pytest-cov
+    - pytest-flake8
+  commands:
+    - py.test


Perhaps py.test sparse ?

I may be mistaken, but I think the preference is to not run full test suites as part of a conda-forge recipe, because it eats up too many CI resources.

See the user guide here: https://conda-forge.org/docs/testing.html#should-a-recipe-run-all-of-a-package-s-tests

Fair enough. I'll just take out the test and we'll do the testing on our own CI.

hameerabbasi · 2018-04-15T16:24:12Z

@synapticarbors I apologize for this, but could you either merge or re-trigger Travis? The failure seems to be related to not being able to download the package, and it downloaded fine in the other CI services.

jakevdp · 2018-04-15T16:42:39Z

I restarted the Travis build

hameerabbasi · 2018-04-15T16:50:21Z

Another intermittent error. 😐

Edit: And a different one at that. Best to wait a bit I guess. Travis might be having issues.

jakirkham · 2018-04-16T04:43:54Z

Anything in the run section should automatically be available in the test section. I think you need numpy in the build section since in the setup.py you import the package which will necessitate all of the run dependencies. Doesn't look like that happens once you switch to the version that uses versioneer.

This shouldn't be needed any more now that it is not importing the package, correct?

hameerabbasi · 2018-04-16T06:32:07Z

Unfortunately, the release version is still doing that. I'm hesitating on releasing now because we made some backwards incompatible changes (feature removals) I need to clean up and add back with new logic.

jakevdp · 2018-04-16T13:41:23Z

Looks good to me. 👍 for merge

jakirkham · 2018-04-16T13:53:44Z

FWIW was able to build locally when dropping basically everything from build except python and pip.

jakevdp · 2018-04-16T14:19:26Z

That makes sense: I think numpy, scipy, and numba can be removed from build_deps. If you look at the v0.3.1 setup.py file you can see that it doesn't import those or import anything that uses those.

hameerabbasi · 2018-04-16T14:31:02Z

Ah, you're both right. I removed those from the build deps, and I also removed numba as it was introduced as a dependency after 0.3.1.

jakirkham · 2018-04-16T14:50:29Z

Do you still want to wait on this or is it good to merge?

hameerabbasi · 2018-04-16T14:52:45Z

+1 to merge from me. 😄

jakirkham · 2018-04-18T18:56:51Z

Sorry for the delay. Thanks also for your patience. This is now in a feedstock with builds triggered.

Add initial sparse recipe.

c2d1213

synapticarbors reviewed Apr 12, 2018

View reviewed changes

hameerabbasi added 2 commits April 12, 2018 18:17

Add fixes suggested by @synapticarbors.

a16e04a

Fix version in doc_url.

d941537

Move requirements to build.

67788f0

hameerabbasi force-pushed the sparse-recipe branch from fa9996c to 67788f0 Compare April 12, 2018 16:42

hameerabbasi added 3 commits April 13, 2018 08:40

Bump to version 0.3.1.

3302e03

Add @jakevdp as co-maintainer of recipe.

a93a367

Wrong SHA256.

ea5c3af

mrocklin reviewed Apr 13, 2018

View reviewed changes

hameerabbasi added 2 commits April 15, 2018 18:13

Correct Numpy required version.

9b4b427

Take out unneeded tests.

44fece8

hameerabbasi added 2 commits April 16, 2018 16:29

Remove unnecessary build deps.

258f3b2

Remove numba as not needed in 0.3.1.

1dc9f3b

jakirkham merged commit 7ce0cf9 into conda-forge:master Apr 16, 2018

hameerabbasi mentioned this pull request Apr 16, 2018

Conda-forge package? pydata/sparse#115

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add a recipe for pydata/sparse #5631

Add a recipe for pydata/sparse #5631

hameerabbasi commented Apr 11, 2018 •

edited

Loading

conda-forge-linter commented Apr 11, 2018

synapticarbors Apr 12, 2018

synapticarbors Apr 12, 2018

hameerabbasi Apr 12, 2018

synapticarbors Apr 12, 2018

mrocklin Apr 12, 2018

jakevdp Apr 12, 2018

synapticarbors Apr 12, 2018

hameerabbasi commented Apr 12, 2018

synapticarbors commented Apr 12, 2018

synapticarbors commented Apr 12, 2018

jakevdp commented Apr 12, 2018

hameerabbasi commented Apr 12, 2018

hameerabbasi commented Apr 13, 2018

mrocklin Apr 13, 2018

jakevdp Apr 13, 2018

jakevdp Apr 13, 2018 •

edited

Loading

hameerabbasi Apr 15, 2018

hameerabbasi commented Apr 15, 2018 •

edited

Loading

jakevdp commented Apr 15, 2018

hameerabbasi commented Apr 15, 2018 •

edited

Loading

jakirkham commented Apr 16, 2018 •

edited

Loading

hameerabbasi commented Apr 16, 2018 •

edited

Loading

jakevdp commented Apr 16, 2018

jakirkham commented Apr 16, 2018

jakevdp commented Apr 16, 2018 •

edited

Loading

hameerabbasi commented Apr 16, 2018

jakirkham commented Apr 16, 2018

hameerabbasi commented Apr 16, 2018

jakirkham commented Apr 18, 2018

Add a recipe for pydata/sparse #5631

Add a recipe for pydata/sparse #5631

Conversation

hameerabbasi commented Apr 11, 2018 • edited Loading

conda-forge-linter commented Apr 11, 2018

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

hameerabbasi commented Apr 12, 2018

synapticarbors commented Apr 12, 2018

synapticarbors commented Apr 12, 2018

jakevdp commented Apr 12, 2018

hameerabbasi commented Apr 12, 2018

hameerabbasi commented Apr 13, 2018

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jakevdp Apr 13, 2018 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

hameerabbasi commented Apr 15, 2018 • edited Loading

jakevdp commented Apr 15, 2018

hameerabbasi commented Apr 15, 2018 • edited Loading

jakirkham commented Apr 16, 2018 • edited Loading

hameerabbasi commented Apr 16, 2018 • edited Loading

jakevdp commented Apr 16, 2018

jakirkham commented Apr 16, 2018

jakevdp commented Apr 16, 2018 • edited Loading

hameerabbasi commented Apr 16, 2018

jakirkham commented Apr 16, 2018

hameerabbasi commented Apr 16, 2018

jakirkham commented Apr 18, 2018

hameerabbasi commented Apr 11, 2018 •

edited

Loading

jakevdp Apr 13, 2018 •

edited

Loading

hameerabbasi commented Apr 15, 2018 •

edited

Loading

hameerabbasi commented Apr 15, 2018 •

edited

Loading

jakirkham commented Apr 16, 2018 •

edited

Loading

hameerabbasi commented Apr 16, 2018 •

edited

Loading

jakevdp commented Apr 16, 2018 •

edited

Loading