Description

bnavigator

opened on Nov 18, 2019

Issue body actions

My recent pull requests enhanced some unit tests, but coverall reports a decrease in coverage.

I believe this happens because it counts the results from the travis jobs without slycot. Most of the unittests are skipped without slycot.

Activity

1. bnavigator commented on Nov 27, 2019

   

   bnavigator

   on Nov 27, 2019

   ContributorAuthor

   More actions

   See the reported decrease for my last contribution to #345 (commit d26d639):

   It is compared to the master branch : "3875 of 4625 relevant lines covered (83.78%)"
   but now counts many more lines : "26201 of 32361 relevant lines covered (80.96%)"

   Something is broken with the coveralls setup.
2. murrayrm commented on Dec 28, 2019

   

   murrayrm

   on Dec 28, 2019

   Member

   More actions

   It looks like slycot may not be getting installed properly in the runs where it is supposed to be => code that relies on slycot is no longer being tested. Looking to see what is wrong...
3. murrayrm

   self-assigned this

   on Dec 28, 2019
4. murrayrm

   added this to the 0.8.3 milestone on Dec 29, 2019
5. murrayrm commented on Dec 30, 2019

   

   murrayrm

   on Dec 30, 2019

   Member

   More actions

   Fixed in PR #353.
6. murrayrm

   closed this as completedon Dec 30, 2019
7. bnavigator commented on Jan 6, 2020

   

   bnavigator

   on Jan 6, 2020

   ContributorAuthor

   More actions

   Coverage still does not count the correct line numbers or compares to a wrong reference. See #366
8. bnavigator commented on Jan 28, 2020

   

   bnavigator

   on Jan 28, 2020

   ContributorAuthor

   More actions

   I don't think this is fixed by #353. With my recent PR #370, coverall reports

   3952 of 4670 relevant lines covered (84.63%)

   but on the master build it is compared to

   34471 of 41958 relevant lines covered (82.16%)

Issue actions