show code coverage with Codecov.io#4587
Conversation
Current coverage is 88.68% (diff: 100%)
|
std/numeric.d
Outdated
| } | ||
|
|
||
| @safe unittest | ||
| unittest |
|
@wilzbach Nice job :) Let me know if you have any questions about Codecov configuration. |
f6ce3ec to
721bbce
Compare
.travis.yml
Outdated
| # test code coverage | ||
| - (cd dmd && make -f posix.mak) | ||
| - (cd druntime && make -f posix.mak) | ||
| - (cd phobos && make -f posix.mak $(find std etc -name "*.d" | sed "s/[.]d$/.test/" | grep -vE '(std.encoding|net.curl)' ) |
There was a problem hiding this comment.
Encoding's aren't registered properly for single tests - see e.g. #4493 and the reported bug for more infos.
|
Travis is passing finally 🎉, so this experiment is ready. As mentioned earlier there are some weird bootstrapping issues with You can preview the automatic coverage report here. In case you are unfamiliar with The behavior of |
Hi all, the idea for this automation is simple, let's ensure (1) that no PR decreases the code coverage, (2) help reviewers judging code coverage improvements PRS, (3) provide simple tools for contributors to see which parts are uncovered.
You can preview this here.
That being said, this won't be perfect in the near future as Travis can only run Linux and Mac OS builds, so special versioned paths will not be treated right. The idea here is to get an easy 90% solution that already helps reviewers in most cases (=a lot better than 0%). In the future we might add Codecov reporting to the AutoTester.
Now to this PR:
make -f posix.mak std.test etc.testentropyin std.numeric (and thus schwarzSort in std.algorithm.sorting)utf-16lein std.encodingiso-8859-1 not recognizedfor std.net.curldruntimeanddmdare needed, we now have to movephobosin its own folder and then clone the other repos. It makes the Travis script a bit ugly, better ideas are of course welcome.(if everything goes well, CodeCov should comment on this PR)