Tags: mmitrevs/infer
Tags
[CLOpt] replace uses of Str with Core.String Summary: Running `infer --help` with spacetime showed that `Str` was a hog. And indeed perf is 74% better now: ``` before: bash -c 'for i in $(seq 100); do infer --help > /dev/null 2>/dev/null; done' 28.02s user 0.62s system 98% cpu 29.205 total after: bash -c 'for i in $(seq 100); do infer --help > /dev/null 2>/dev/null; done' 6.34s user 0.64s system 92% cpu 7.557 total ``` Running 101D8 on many files one after the other is now 64% faster too: ``` $ cd infer/tests/codetoanalyze/c/errors $ # before $ rm -fr infer-out && time bash -c 'for src in */*.c; do infer -a capture --continue -- clang -c $src; done' bash -c 7.77s user 0.66s system 97% cpu 8.647 total $ # after $ rm -fr infer-out && time bash -c 'for src in */*.c; do infer -a capture --continue -- clang -c $src; done' bash -c 2.35s user 0.56s system 93% cpu 3.119 total $ time infer -a capture -- clang -c */*.c infer -a capture -- clang -c */*.c 0.54s user 0.20s system 99% cpu 0.737 total ``` Reviewed By: mbouaziz Differential Revision: D4875803 fbshipit-source-id: cfcfa69
do not print infer.py usage message on wrong arguments Summary: On wrong arguments (or on no arguments at all), `infer` would spew the error message of `infer.py`, which makes no sense. Make the python code swallow error messages and exit with a special code on errors coming from command line parsing so that the OCaml side is in charge of printing usage messages. Reviewed By: cristianoc Differential Revision: D3731594 fbshipit-source-id: fe49cda
update clang command-line flags that we filter out Summary:Open-source clang has caught up a bit with apple's clang, so we don't need to filter as many compilation flags as we used to. Reviewed By: akotulski, martinoluca Differential Revision: D3212553 fb-gh-sync-id: 5638dc8 fbshipit-source-id: 5638dc8
PreviousNext