-
Notifications
You must be signed in to change notification settings - Fork 33
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Migration to Python 3 #311
Conversation
:( |
In the error message from CMake
|
I see, the mpich wrapper contains |
Ubuntu seems to be fine. The issue is with Fedora MPICH |
Oh, the fedora docker build was disabled, see espressopp/buildenv#27 |
Hey there, I was trying to compyle the branch with intel compilers and python 3.7 and i got one strange boost error I could not get rid of
Add the installation prefix of "boost_python" to CMAKE_PREFIX_PATH or set any clues? |
…ong parameter for the compiler
…ong parameter for the compiler
I think this is because I hardcoded in required boost components python38; I have made a small fix to CMakeList.txt to select properly boost component based on the founded version of Python (hopefully this works). |
It's still failing, even with rebuilt fedora image. I have temporally switched off build of fedora_mpich with clang |
With codecov step the issue is with the log size on travis. The log contains enormouse number of
lines. This results in the error: I do not get why codecov analysis, e.g. |
Codecov Report
@@ Coverage Diff @@
## master #311 +/- ##
========================================
Coverage ? 0.0%
========================================
Files ? 424
Lines ? 27724
Branches ? 3187
========================================
Hits ? 0
Misses ? 27724
Partials ? 0
Continue to review full report at Codecov.
|
Any progress with this PR? |
From my side, go ahead! |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I'm testing on Intel Compiler 19.0.5.281 and Boost 1.74, and everything seems to be fine except for the following test with -O2
and -O3
optimization. This could be just the result of more aggressive optimization from the Intel compiler as it doesn't show up in -O1
and w/o any optimization flag. If we don't need to address this now, then the PR is already fine from my side.
21/66 Test #21: MTSAdResS ..........................***Failed 1.27 sec
..F.
======================================================================
FAIL: test_AT_CG_templates_FAdResS (__main__.TestMTSAdResS)
----------------------------------------------------------------------
Traceback (most recent call last):
File "/home/vance/espressopp/py3/espressopp/testsuite/AdResS/MTSAdResS/test_MTSAdResS.py", line 242, in test_AT_CG_templates_FAdResS
self.assertAlmostEqual(after[0], 9.364194, places=5)
AssertionError: 9.36420869641006 != 9.364194 within 5 places (1.4696410060466292e-05 difference)
----------------------------------------------------------------------
Ran 4 tests in 0.056s
FAILED (failures=1)
Cubicity check passed -> powered by HeSpaDDA
HeSpaDDA message: Size Lenghts are eq. while ordering axis with preference on X, Y and Z!
Cubicity check passed -> powered by HeSpaDDA
HeSpaDDA message: Size Lenghts are eq. while ordering axis with preference on X, Y and Z!
Cubicity check passed -> powered by HeSpaDDA
HeSpaDDA message: Size Lenghts are eq. while ordering axis with preference on X, Y and Z!
Cubicity check passed -> powered by HeSpaDDA
HeSpaDDA message: Size Lenghts are eq. while ordering axis with preference on X, Y and Z!
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I have tested with 2 versions of python3 and it seems to work fine. The only complementary PR, I see that might be required is a bit more of documenting that we reached Py3. As an idea I would suggest to add some compiling/installing requirements for a couple of Dist. perhaps the ones we have in the CI. However, this can be a PR coming up in the next days. @jkrajniak go ahead!
I have migrated the code base to work with Python 3. I have tested with Python 3.8; Locally all tests passed without issue. I have also run simulation from examples/lennard_jones in both python 2 and 3. The results are identical. Nevertheless, it would be good if someone could do separate tests.
TODO:
I think the best idea would be to move the current
master
to branchpy2
and makemaster
python3 only compatible.Fix #227