Extract coverage data in an info-file created by lcov to separate
.gcov files, one for each source file.
info2gcov [-q] [-h] <info-file>
info2gcov will write .gcov-files where the source files are. It will
silently skip any files it cannot write to.
I wanted to use the Emacs cov-mode but it only could parse gcov files.
My tests produces separate coverage data files because I want to run them in
parallel (see sidebar). gcov could not handle this, neither could gcovr
for some reason.
Using lcov to produce separate .info files allowed me to aggregate all
those into a total.info.
This script added the final piece, extract aggregated coverage data to
separate .gcov-files.
If you have --coverage it is not possible to run tests in parallel because
the writing of coverage data to the .gcda-files will clash.
The only solution is to create separate coverage data sets by using the
GCOV_PREFIX to set a directory where data will be stored (instead of in the
original object directory).
Then you have to merge all the .info-files into one with
lcov --add-tracefile test1.info <...> -o total.info