@schneiderfelipe sorry but there seem to have been conflicts merged in recently - possibly mine.

You need to merge or rebase manually - let me know if you need help with that.

This is failing and I don't know why. Errors are not due to decimal places as they didn't change after the last commit.

Failures are due to dvb_ir.out ORCA logfiles, but the newest version (4.0) is the only one that doesn't fail. In summary, errors happen with (2 failed tests each):

• ORCA2.6/dvb_ir.out, associated with class OrcaIRTest_old_coordsOK (inherits from OrcaIRTest)
• ORCA2.8/dvb_ir.out, associated with class OrcaIRTest_old (inherits from OrcaIRTest too)
• ORCA2.9/dvb_ir.out, associated with class OrcaIRTest
• ORCA3.0/dvb_ir.out, associated with class OrcaIRTest as well

ORCA4.0/dvb_ir.out presents no error, but is associated with class GenericIRTest, which doesn't test for entropy/enthalpy/freeenergy values. Furthermore, OrcaIRTest inherits from GenericIRTest.

This is clearly inconsistent. Is there a reason for this? What should we do?

This is failing and I don't know why. Errors are not due to decimal places as they didn't change after the last commit.

Failures are due to dvb_ir.out ORCA logfiles, but the newest version (4.0) is the only one that doesn't fail. In summary, errors happen with (2 failed tests each):

• ORCA2.6/dvb_ir.out, associated with class OrcaIRTest_old_coordsOK (inherits from OrcaIRTest)
• ORCA2.8/dvb_ir.out, associated with class OrcaIRTest_old (inherits from OrcaIRTest too)
• ORCA2.9/dvb_ir.out, associated with class OrcaIRTest
• ORCA3.0/dvb_ir.out, associated with class OrcaIRTest as well

ORCA4.0/dvb_ir.out presents no error, but is associated with class GenericIRTest, which doesn't test for entropy/enthalpy/freeenergy values. Furthermore, OrcaIRTest inherits from GenericIRTest.

This is clearly inconsistent. Is there a reason for this? What should we do?

Well, I think the cause of this is the improved accuracy in your checks. We've changed the coordinates and such over the years so that unit test logfiles for different packages are more aligned. So I'm not surprised we would see slight differences for these older logfiles which are now in the regression suite.

You can adjust the test class attributes in a custom test class and assign it to the old unit test in the regression suite. See this for an example: https://github.com/cclib/cclib/blob/master/test/regression.py#L2927. I know this is a little clumsy, and we need to improve the usability, plus that file is getting mighty large... and that will eventually get improved.

P. S. Trying to add more traceback information to the regression suite run on Travis in #834. Also, for some reason I don't see the Travis links/icons in github PRs - do you guys see them?

Well, I think the cause of this is the improved accuracy in your checks.

I don't think so. This is the Travis build for 1ed29be (prior to the change in accuracy). Here is the Travis build for the last commit (96e33a2), including the improved accuracy. The errors are the same.

P. S. Trying to add more traceback information to the regression suite run on Travis in #834.

Thanks. Let's see what the traceback info gives us.

Also, for some reason I don't see the Travis links/icons in github PRs - do you guys see them?

I don't see anything as well.

Well, I think the cause of this is the improved accuracy in your checks.

I don't think so. This is the Travis build for 1ed29be (prior to the change in accuracy). Here is the Travis build for the last commit (96e33a2), including the improved accuracy. The errors are the same.

Hmm.. need to look into this then.

P. S. Trying to add more traceback information to the regression suite run on Travis in #834.

Thanks. Let's see what the traceback info gives us.

Unfortunately my PR broke Travis so I reverted it. Need to figure out how to pass an additional flag using pytest.

Also, for some reason I don't see the Travis links/icons in github PRs - do you guys see them?

I don't see anything as well.

Seems there was some kind of migration from travis-ci.org to travis-ci.com. Removed and re-added the app integration and works again now from the new domain.

Unfortunately my PR broke Travis so I reverted it. Need to figure out how to pass an additional flag using pytest.

I think adding something like

[pytest]
addopts = --tb=long

to pytest.ini might work.

Seems there was some kind of migration from travis-ci.org to travis-ci.com. Removed and re-added the app integration and works again now from the new domain.

Yes, I see it now. Thanks!

Unfortunately my PR broke Travis so I reverted it. Need to figure out how to pass an additional flag using pytest.

I think adding something like

[pytest]
addopts = --tb=long

Unfortunately not. I'm trying to add use a flag that is defined in regression.py.

Unfortunately not. I'm trying to add use a flag that is defined in regression.py.

Sorry, I thought that was the pytest --tb flag!

Codecov Report

Merging #812 (bb453f5) into master (937fdbf) will not change coverage.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##           master     #812   +/-   ##
=======================================
  Coverage   87.28%   87.28%           
=======================================
  Files          69       69           
  Lines       12778    12778           
=======================================
  Hits        11153    11153           
  Misses       1625     1625           

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 937fdbf...bb453f5. Read the comment docs.

@berquist, can this be merged? 🥳