The test coverage could be better. Use cargo llvm-cov to find areas of the code that aren't covered (see relevant section of our docs on how to use this).
In most cases, it probably makes sense to use patched examples rather than writing traditional unit tests for it, but the main thing is that the tests are targeted, as opposed to many of the regression tests we have now (not all though!).
I just ran cargo llvm-cov and it seems most of the places that aren't covered are error paths for input files, which should be eminently testable 😄. Hopefully you should be able to write tests using the simple example, but, if that proves difficult, we may have to do #1080 first.
It's fine not to have 100% coverage for this module. This issue is just about improving things.
The test coverage could be better. Use
cargo llvm-covto find areas of the code that aren't covered (see relevant section of our docs on how to use this).In most cases, it probably makes sense to use patched examples rather than writing traditional unit tests for it, but the main thing is that the tests are targeted, as opposed to many of the regression tests we have now (not all though!).
I just ran
cargo llvm-covand it seems most of the places that aren't covered are error paths for input files, which should be eminently testable 😄. Hopefully you should be able to write tests using thesimpleexample, but, if that proves difficult, we may have to do #1080 first.It's fine not to have 100% coverage for this module. This issue is just about improving things.