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
AFAIU, the provided link only helps in showing code coverage data. Looking at the lit documentation directly, they don't seem to be able to report code coverage metrics. While a package like coverage.py might be helpful, it does not work in this case as lit is it's own executable.
We might be able to run a sufficiently smart pytest program, that basically does what lit does atm (crawl over files, run the right commands), while also being introspectable. Though I feel that it might actually be easier to just expand our pytest suite to a point where we won't need the filecheck tests to get a reasonable high coverage.
Though I feel that it might actually be easier to just expand our pytest suite to a point where we won't need the filecheck tests to get a reasonable high coverage.
What I am not 100% sure about is whether this is actually possible/reasonable, in particular around xdsl-opt, as this is really intended to be run from the command line. This becomes particularly true for things around the mlir-converter. I guess we can write some weird tests to circumvent this though...
Maybe something with https://llvm.org/docs/CommandGuide/llvm-cov.html can be done.
The text was updated successfully, but these errors were encountered: