drop support for python 2.7 by lheagy · Pull Request #111 · simpeg/discretize

lheagy · 2018-10-24T20:37:33Z

add an f-string to the copyright in the init.py
remove testing on python 2.7 on both travis and appveyor
include requires_python >= 3.6 in the setup.py
I suggest we move to version 0.4.0 with this pr

leouieda · 2018-10-24T20:47:19Z

It's such a liberating feeling, isn't it?

banesullivan · 2018-10-24T21:10:46Z

As an outsider who's been thinking about incorporating discretize with some of my projects, I would still need Python 2.7 compatibility.

I've been toying with the idea of trashing all my PVGeo file IO for GIF Tensor/Tree Meshes and other discretize-friendly data formats. Then simply use the VTK interfaces already in discretize (I would also create VTK interfaces where they are missing in discretize and ensure the existing ones are good to go).

Still having Python 2.7 support would allow me to make PVGeo algorithms that call back to discretize and have a graphical user interface for discretize directly in ParaView. This would look similar to GMSGDataExchange/omf#27 which unfortunately won't be happening because the VTK interface in omf would have to be separated.

Me needing Python 2.7 support would only last until ParaView upgrades its Python to 3.6 which I believe is scheduled for Summer 2019.

Don't let this hold back progress though!

prisae · 2018-10-24T21:19:12Z

What about having v0.3.3 on bug-fix-support till the end of 2019, but v0.4.0 and future ones are Python 3 only? This would be the same as numpy/scipy will be anyway, just two months earlier. Or do you think you will need new features in discretize as well @banesullivan?

Edit: It would obviously be v0.3.x with only bug-fix-support, as any bug-fix would increase the version number. So you could fix the dependency to v0.3, and v0.4 and following are Python 3 only.

banesullivan · 2018-10-24T21:31:29Z

@prisae, that would work fine!

I could then set PVGeo's dependency as v0.3.3 and any VTK interfaces needed for discretize that are not currently implemented, I will add directly in PVGeo. Then once ParaView upgrades to Python 3, I would migrate those VTK object interfaces into discretize so classes like CurvilinearMesh, CylMesh, etc would have toVTK() methods like the _toVTRObj method in MeshIO.

lheagy · 2018-10-26T20:53:21Z

Hi @banesullivan, thanks for chiming in! We can definitely continue to support in terms of bug fixes in the mean-time.

P.S. I added you to the SimPEG org 🎉. I am thrilled that you are interested in interfacing to discretize - having accessible 3D viz is something that would have a very big impact!

banesullivan · 2018-10-26T21:27:57Z

Wow, thanks @lheagy!!! I am definitely interested interfacing discretize into my 3D viz projects in @OpenGeoVis so this will be the perfect way to start collaborating!

codecov · 2018-10-26T23:58:33Z

Codecov Report

Merging #111 into master will decrease coverage by 1.56%.
The diff coverage is 40.35%.

@@            Coverage Diff             @@
##           master     #111      +/-   ##
==========================================
- Coverage   72.99%   71.43%   -1.57%     
==========================================
  Files          20       20              
  Lines        4503     4589      +86     
==========================================
- Hits         3287     3278       -9     
- Misses       1216     1311      +95

Impacted Files	Coverage Δ
discretize/CylMesh.py	`91.55% <ø> (-0.02%)`	⬇️
discretize/utils/coordutils.py	`83.33% <ø> (ø)`	⬆️
discretize/utils/__init__.py	`100% <ø> (ø)`	⬆️
discretize/utils/meshutils.py	`74.19% <ø> (-0.81%)`	⬇️
discretize/utils/interputils.py	`80.55% <ø> (-0.53%)`	⬇️
discretize/utils/matutils.py	`92.88% <ø> (-2.06%)`	⬇️
discretize/Tests.py	`78.53% <100%> (-0.11%)`	⬇️
discretize/View.py	`18.25% <100%> (-0.12%)`	⬇️
discretize/InnerProducts.py	`97.42% <100%> (ø)`	⬆️
discretize/TensorMesh.py	`90.36% <100%> (-0.03%)`	⬇️
... and 8 more

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 7bc916a...f14e177. Read the comment docs.

prisae

I think this would be a good point to clean-up the code base from stuff like from __future__ import print_function, as those are mostly obsolete now.

prisae · 2018-10-27T00:09:44Z

setup.py

+    version="0.4.0",
    install_requires=[
        'numpy>=1.7',
        'scipy>=0.13',


How about taking the opportunity and crank up the scipy-requirement? Ideally at 1.0? But that might be too high. Anyway, 0.13 is a bit low, and leaving Python 2 behind might be a good point to adjust.

good call >= 1?

It might be a bit pushy. But it was released a year ago, and it was a big release (many changes), so it might make it easier...

prisae · 2018-10-27T00:10:06Z

setup.py

@@ -78,5 +78,6 @@ def configuration(parent_package='', top_path=None):
    platforms=["Windows", "Linux", "Solaris", "Mac OS-X", "Unix"],
    use_2to3=False,


use_2to3: obsolete?

- remove imports from __future__ - update scipy requirement - only specify additional requirements for development in the requirements_dev.txt (e.g. don't include requirements described in the setup.py)

prisae

Looks good to me now.

…nto no_more_py27

add python 3.7 to testing

banesullivan · 2018-10-28T21:46:20Z

Would it be possible merge #114 to v0.3.x and deploy on PyPI before this bump to 0.4.0?

lheagy · 2018-10-28T21:51:32Z

Yep, we can plan to do that @banesullivan. I am just not sure what is causing the upstream errors on python 2.7... it is failing on the pip install (I am sort-of hoping other groups are running into this and it will resolve itself, but it has been close to a week...)

prisae

@lheagy , you replaced absolute imports by relative imports for the Py2->Py3 switch in most files. However, not in

__init__.py
DiffOperators.py
InnerProducts.py
View.py

Forgotten or is there a particular reason?

lheagy · 2018-10-30T14:54:45Z

Thanks @prisae - definitely forgotten :)

…nto no_more_py27

lheagy · 2018-12-08T00:10:52Z

discretize/BaseMesh.py

+        ):
+        return self.reshape(x, xType, outType, format)
+
+    def reshape(self, x, xType='CC', outType='CC', format='V'):


I think we should also rethink input variable names here (e.g. format should not be used). What about

def reshape(self, x, x_loc='CC', out_loc='CC', out_format='V')

I think "loc" is a bit more descriptive (and even "location"). I am still not sure about "format". (It should also take "vec", "vector", "matrix" to let the user be a bit more flexible in the input).

Thoughts?

cc @rowanc1

prisae · 2020-08-26T17:21:37Z

Talking about Python 3.5 over in #219 it seems odd that this PR is still open. Could we merge this @lheagy ?

prisae · 2020-11-18T15:25:26Z

Most of this was implemented in #227 instead.

lheagy added 3 commits October 24, 2018 13:25

drop support for python 2.7

760fea4

ensure python >=3.5 in the setup.py, test on python 3.5 on travis

17f6b30

require python 3.6

ba52721

lheagy mentioned this pull request Oct 24, 2018

drop support for python 2.7 simpeg/simpeg#731

Closed

Bump version: 0.3.3 → 0.4.0

08e0b9d

prisae requested changes Oct 27, 2018

View reviewed changes

clean-out of python 2.7 specific code

2b2dd42

- remove imports from __future__ - update scipy requirement - only specify additional requirements for development in the requirements_dev.txt (e.g. don't include requirements described in the setup.py)

prisae approved these changes Oct 27, 2018

View reviewed changes

lheagy added 4 commits October 26, 2018 17:34

Merge branch 'master' into no_more_py27

137902d

docstring cleanup

f86c801

Merge branch 'no_more_py27' of https://github.com/simpeg/discretize i…

0f7705d

…nto no_more_py27

Update .travis.yml

151a968

add python 3.7 to testing

Update .travis.yml

b562c5b

prisae reviewed Oct 30, 2018

View reviewed changes

lheagy added 6 commits October 30, 2018 07:54

Merge branch 'master' into no_more_py27

54c5070

light cleanup of the docs

293cf26

Merge branch 'no_more_py27' of https://github.com/simpeg/discretize i…

5996e59

…nto no_more_py27

use relative imports

667dc4e

Merge branch 'master' into no_more_py27

5af81f1

travis debugging

67f2952

lheagy added 2 commits December 7, 2018 15:59

add missing comma

61a8a5d

rename r--> reshape, light docstring cleanup

2d411fb

lheagy commented Dec 8, 2018

View reviewed changes

lheagy added 4 commits December 7, 2018 16:29

use warnings package to raise warning

1ce6e7d

favor exceptions over assertions

ae14152

adding docs, light cleanup

e5c49de

use in testing

f14e177

prisae mentioned this pull request Sep 28, 2020

Restructure #227

Merged

jcapriot closed this Nov 17, 2020

jcapriot deleted the no_more_py27 branch August 19, 2021 19:39

		@@ -78,5 +78,6 @@ def configuration(parent_package='', top_path=None):
		platforms=["Windows", "Linux", "Solaris", "Mac OS-X", "Unix"],
		use_2to3=False,

Conversation

lheagy commented Oct 24, 2018 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

leouieda commented Oct 24, 2018

Uh oh!

banesullivan commented Oct 24, 2018 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

prisae commented Oct 24, 2018 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

banesullivan commented Oct 24, 2018

Uh oh!

lheagy commented Oct 26, 2018

Uh oh!

banesullivan commented Oct 26, 2018

Uh oh!

codecov bot commented Oct 26, 2018 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

prisae left a comment

Choose a reason for hiding this comment

Uh oh!

prisae Oct 27, 2018

Choose a reason for hiding this comment

Uh oh!

lheagy Oct 27, 2018

Choose a reason for hiding this comment

Uh oh!

prisae Oct 27, 2018

Choose a reason for hiding this comment

Uh oh!

prisae Oct 27, 2018

Choose a reason for hiding this comment

Uh oh!

prisae left a comment

Choose a reason for hiding this comment

Uh oh!

banesullivan commented Oct 28, 2018

Uh oh!

lheagy commented Oct 28, 2018

Uh oh!

prisae left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

lheagy commented Oct 30, 2018

Uh oh!

lheagy Dec 8, 2018

Choose a reason for hiding this comment

Uh oh!

prisae commented Aug 26, 2020

Uh oh!

prisae commented Nov 18, 2020

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

lheagy commented Oct 24, 2018 •

edited

Loading

banesullivan commented Oct 24, 2018 •

edited

Loading

prisae commented Oct 24, 2018 •

edited

Loading

codecov bot commented Oct 26, 2018 •

edited

Loading

prisae left a comment •

edited

Loading