It's not a big deal. At first I thought it was changes I made, but than I noticed that it reported a drop of 4% coverage on my unchanged master branch. It's more of a curiosity than anything else, though, also a little disappointing, b/c I had just hit 92% coverage! Dropped to 88% after 1.17 was released... 
bluejekyll/trust-dns | Coveralls - Test Coverage History & Statistics
Hmm, it looks like the drop on master was from build #495 to build #499. The total number of "relevant lines" dropped significantly!
Looking at the biggest drop, authority.rs, it looks like a lot of lines that are just closing } are no longer marked at all, e.g. before and after.
1 Like
I was concerned that maybe there was a regression in the tests that were recognized as doc tests or something. But the same numbers are there before and after 1.17.
I guess it is the relevant lines. Did 1.17 bring any changes, like a reduction in generated code or anything? It's a mild curiosity.
On a related note, I have been thinking of trying to make coverage reports easier to generate in Rust. Has anyone looked at a cargo based tool for Rust integrated with llvm-cov?
1 Like