codecov[bot]
commented
5 years ago Codecov Report
Merging #158 into master will not change coverage. The diff coverage is
100%.
@@ Coverage Diff @@
## master #158 +/- ##
=======================================
Coverage 87.86% 87.86%
=======================================
Files 14 14
Lines 898 898
=======================================
Hits 789 789
Misses 109 109| Impacted Files | Coverage Δ | |
|---|---|---|
| fitgrid/__init__.py | 100% <100%> (ø) |
:arrow_up: |
Continue to review full report at Codecov.
Legend - Click here to learn more
Δ = absolute <relative> (impact),ø = not affected,? = missing dataPowered by Codecov. Last update a619fc6...8e9c844. Read the comment docs.
Knits TravisCI together with conda
.travis.yml
install:
conda build condamakes a fitgrid tarball from the meta.yaml, then does aconda install fitgrid -c localinto a fresh conda env to run pytestsdeploy: a new script provider uploads converted tarballs to the Anaconda Cloud kutaslab channel on all and only github release versions tagged vN.N.N
Pros: fitgrid pytests run in a conda env as installed by conda and github. All and only github release versions tagged vN.N.N are automatically packaged and uploaded to Anaconda Cloud with tarball build strings matching the release vN.N.N and git commit short hash (this is important).
Cons: TravisCI builds are slower in the install stage, the before_deploy stage triggers on every build not just master. The slowdown is a nuisance but not intractable. Not much help for the install stage since the point is to pytest fitgrid in whatever environment conda feels like building that day. Possible fixes for the deploy stage are to make conda uploads conditional on vN.N.N tags (but this means no upload dry run reports) and/or move the before_build stage sphinx setup to a script provider for gh_pages (this works in testing on kutaslab/spudtr); a workaround is simply to temporarily comment out deployment while working on dev branches.