Conversation
IAlibay
left a comment
IAlibay
left a comment
There was a problem hiding this comment.
Congrats everyone :)
What was the decision on dropping py3.5 too?
| CONDA_DEPENDENCIES="" | ||
| INSTALL_HOLE="false" | ||
|
|
||
| - env: NAME="python 2.7" |
There was a problem hiding this comment.
Can't remember the exact issue but I think we said we would drop 3.5 too.
There was a problem hiding this comment.
I thought dropping py2.7 and py3.5 would be a 2.0 thing, and that 1.xx releases would be just bug fixes, based on that same comment?
There was a problem hiding this comment.
It looks like everyone else is moving on, I'd like to just ditch python 2. If we find anything nasty that we fix I could do a backport/LTS thing, but I'd rather just force everyone to move to py 3, anaconda makes this possible enough.
There was a problem hiding this comment.
I was about to tag that comment! ;) I think this discussion is suited for the dev list, but if major packages are already dropping support for 3.5 (see #2541 (comment)) probably we should drop it too.
There was a problem hiding this comment.
I thought dropping py2.7 and py3.5 would be a 2.0 thing, and that 1.xx releases would be just bug fixes, based on that same comment?
My view here (probably to be discussed on the dev list) is that it would be nice to increase the number of releases to include small bug fixes. The reason is that many issues are solved relatively quickly, but make it on master after a very long time (and therefore most users don't benefit from them).
For this reason, I think we should start dropping Python 2 and 3.5 support with 1.0.x releases (as done here) and tidy up the code as much as possible on the way to 2.0 (possibly with type hints, and other cool stuff).
There was a problem hiding this comment.
In https://www.mdanalysis.org/2019/11/06/roadmap/ we said 1.x will be the last version of MDAnalysis that fully supports Python 2.7
This means that 1.x supports and tests 2.7 so we cannot remove Python 2.7 tests. Drop it for 2.0.
I'll start a discussion on the dev list instead of an issue thread...
package/MDAnalysis/version.py
Outdated
|
|
||
| #: Release of MDAnalysis as a string, using `semantic versioning`_. | ||
| __version__ = "0.20.2-dev0" # NOTE: keep in sync with RELEASE in setup.py | ||
| __version__ = "1.0.0-dev0" # NOTE: keep in sync with RELEASE in setup.py |
There was a problem hiding this comment.
Tests are failing because this tag is different from the one in MDAnalysisTests/__init__.py.
E AssertionError: MDAnalysis release 1.0.1-dev0 must be installed to have meaningful tests, not 1.0.0-dev0
E assert '1.0.0-dev0' == '1.0.1-dev0'
E - 1.0.1-dev0
E ? ^
E + 1.0.0-dev0
E ? ^
orbeckst
left a comment
orbeckst
left a comment
There was a problem hiding this comment.
1.0.1 must retain Python 2.7 – in fact, all of 1.x has to, according to https://www.mdanalysis.org/2019/11/06/roadmap/ and semantic versioning.
This needs to be 2.0.0.
1.0.x will be backports of serious bugs
IAlibay
left a comment
IAlibay
left a comment
There was a problem hiding this comment.
Unless there's some disagreement on the mailing list, lgtm.
Codecov Report
@@ Coverage Diff @@
## develop #2729 +/- ##
===========================================
+ Coverage 91.22% 91.80% +0.57%
===========================================
Files 176 15 -161
Lines 24115 2025 -22090
Branches 3160 0 -3160
===========================================
- Hits 22000 1859 -20141
+ Misses 1492 166 -1326
+ Partials 623 0 -623 Continue to review full report at Codecov.
|
| SETUP_CMD="${PYTEST_FLAGS} --cov=MDAnalysis" | ||
| PIP_DEPENDENCIES="${PIP_DEPENDENCIES} setuptools<45.0.0" | ||
|
|
||
| - env: NAME="asv check" |
There was a problem hiding this comment.
Note that ASV is currently running under 2.7 so we need an issue updating that to 3.x
6 participants
Start of the next release