Nick Mathewson
2017-09-26 16:14:28 UTC
So, here's some good news about unit test coverage (as measured by
'make check') in 0.3.2:
Our (line) test coverage rate is up, from 52.88% to 54.72%.
But here is some bad news:
The number of uncovered lines has still increased (from 30860 to
31608), even though covered lines have increased more rapidly.
If you're interested in tracking down where we need coverage, I have
used our "cov-diff" tool to generate a cleaned diff from the old
coverage to the new coverage. You can download it (it's big!) from
https://people.torproject.org/~nickm/volatile/coverage-diff.xz
The format is: it is a diff between normalized gcov output files. To
find new or modified un-covered lines, look for lines that begin with
the regex
/^\+ *\#/
("starts with a plus, then some spaces, then a #").
Any lines like this are good candidates for additional testing IMO.
peace,
'make check') in 0.3.2:
Our (line) test coverage rate is up, from 52.88% to 54.72%.
But here is some bad news:
The number of uncovered lines has still increased (from 30860 to
31608), even though covered lines have increased more rapidly.
If you're interested in tracking down where we need coverage, I have
used our "cov-diff" tool to generate a cleaned diff from the old
coverage to the new coverage. You can download it (it's big!) from
https://people.torproject.org/~nickm/volatile/coverage-diff.xz
The format is: it is a diff between normalized gcov output files. To
find new or modified un-covered lines, look for lines that begin with
the regex
/^\+ *\#/
("starts with a plus, then some spaces, then a #").
Any lines like this are good candidates for additional testing IMO.
peace,
--
Nick
Nick