The libc-scraper for sigkit is broken.
The awk in run.sh "needs" to be $4, not $3.
parallel's -j argument should probably be 0, not 3.
run.sh should probably use an absolute or relative path to process-deb.sh, instead of just assuming it will be in $PATH (unless an install script is shipped).
It should probably be documented that process-deb.sh depends on batch_process.py being in ~/sigkit/ (unless an install script is shipped).
pkl_blacklist in merge_ubuntu.py should probably be a list literal instead of a dictionary literal.
merge_ubuntu should probably operate on more than just bionic in main.
The documentation suggests merge_ubuntu.py uses all the cpus available on the machine. The cpu_factor, which is passed to the ProcessPoolExecutor constructor only uses the square root of the number of CPUs.
convert_siglib.py should probably use os.path.basename and not indexing on ".", to catch relative paths.
I'm happy to do a PR for some of these, if some decisions are made about the right way to handle them.
The libc-scraper for sigkit is broken.
The awk in run.sh "needs" to be $4, not $3.
parallel's -j argument should probably be 0, not 3.
run.sh should probably use an absolute or relative path to process-deb.sh, instead of just assuming it will be in $PATH (unless an install script is shipped).
It should probably be documented that process-deb.sh depends on batch_process.py being in ~/sigkit/ (unless an install script is shipped).
pkl_blacklist in merge_ubuntu.py should probably be a list literal instead of a dictionary literal.
merge_ubuntu should probably operate on more than just bionic in main.
The documentation suggests merge_ubuntu.py uses all the cpus available on the machine. The cpu_factor, which is passed to the ProcessPoolExecutor constructor only uses the square root of the number of CPUs.
convert_siglib.py should probably use os.path.basename and not indexing on ".", to catch relative paths.
I'm happy to do a PR for some of these, if some decisions are made about the right way to handle them.