You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
PLeVasseur
changed the title
Increase code coverage for unit tests of up-transport-vsomeip
Increase code coverage for unit tests of up-transport-vsomeip to 90%
Jul 31, 2024
I evaluated tarpaulin and llvm-cov unit test coverage tools.
I prefer llvm-cov for the following reasons:
The syntax highlighting was more intuitive.
The coverage summary appears more comprehensive using default settings, i.e. function, line and region coverage percentages.
It uses LLVM source-based code coverage. This is theoretically faster and more accurate due to the fact that Rust by default uses LLVM as its backend compiler.
Capable of generating html and lcov report standard formats.
It offers some additional nice-to-have features such as branch coverage, which can be enabled using a nightly Rust build.
Perhaps, one major setback of tarpaulin is that it relies on kcov and ptrace system calls to collect debug info to calculate coverage. Firstly, kcov and ptrace are only available on Linux.
Secondly, this method is known to produce less accurate results compared to LLVM's native, efficient coverage instrumentation.
I observed some discrepancies when comparing the reports, which led me to believe that LLVM's instrumentation is indeed more precise.
No description provided.
The text was updated successfully, but these errors were encountered: