Initial attempt at python 3 support#468
Conversation
|
No attempt yet to do anything other than compile. |
Codecov Report
@@ Coverage Diff @@
## master #468 +/- ##
==========================================
+ Coverage 43.06% 43.06% +<.01%
==========================================
Files 299 299
Lines 98215 98215
==========================================
+ Hits 42294 42295 +1
+ Misses 55921 55920 -1
Continue to review full report at Codecov.
|
|
Thank you!!! |
|
Yes, we should definitely compile and test against python 3. Python 2 is going to be around for at least as long as RHEL 7 is, so I'd like to keep support for both for the moment. Though I think that the default should move to python 3. |
|
For the Debian package, I played around with this and found two more places to change. For simplicity, I just added them to @opoplawski's branch -- feel free to edit or remove them if required. If there are no objections, I intend to upload a new GDL version with these patches to Debian, so that the version that will come with Buster already is a Python 3 module. I am a bit in hurry with that, since we will have a freeze soon, and this change needs review by our ftp-masters (which also takes time). I will upload the patches for the tests ASAP. |
|
I just uploaded the updated tests that work for Python 3. I'll take the opportunity to document here what does not work yet. I use the following GDL function for tests:
|
4cd310a to
0489305
Compare
|
…and I just rebased to the current master so that it could be cleanly merged. In that case,I would open the problems shown above as issues so that they don't get lost. |
|
The tests fail on travis with: +1 for opening separate tickets for each issue. Thanks! |
|
The problem is IMO here that Python 3 is not (completely) installed and the Python 2.7 lib is enforced in |
Python 2 is dead from 2020-01-01, so it makes sense to use Python 3 as default.
I also re-checked all test parameterizations that were commented out, and enabled those which work now.
Will look at it |
… are commented out in travis.yml
|
@olebole I was proposed to solve a conflict. Rebasing was not an option. |
|
@slayoo I've successfully tested this patch on my branch https://github.com/GillesDuvert/gdl/tree/test-new-python, which is in sync with current GDL master. |
This will close #91