[Checkers] checkers test failures

Matt Papi mpapi at csail.mit.edu
Thu May 29 12:24:34 EDT 2008


I just made a handful of changes to the checkers and, as I was about
to commit them, I noticed that there are many new failures in the
checkers test suite. I spent a while trying to figure out which
commits introduced the failures, and here's what I found:

- r2501/2502 breaks 8 Javari tests.
- r2557, which is logged as documentation improvements, breaks a total
of 30 framework, Basic checker, and IGJ tests, seemingly for the same
reason (missing message files). The current dev release was made after
this commit -- does it have these failures? (I didn't check.)
- r2569, which is logged as code clean-up, fixes some of what's broken
in 2557, but introduces a bunch of NonNull failures; some Javari, IGJ,
and framework failures still remain.
- One of the commits after r2569 seems to fix some of the Javari
failures, but there are still failures in NonNull, Javari, IGJ, and
the framework.

It goes without saying that running the test suite before committing
is always a _really_ good idea :). Maybe some kind of continuous
integration tool (or at least a test-running commit hook or nightly
build script), if easy to set up, could be helpful?

My top priority for today is addressing comments/getting things in
shape for the code review. After that, Mahmood and I can coordinate on
getting this stuff fixed for the next release.

- Matt



More information about the checkers mailing list