On Sun, May 21, 2017 at 1:10 PM, Paul Eggert <egg...@cs.ucla.edu> wrote: > Jim Meyering wrote: >> >> two of the three instances it flagged in grep >> were odd enough that I feel they benefit from the added attention > > > Anybody reading those two instances of the code would already know what's > going on; it's pretty obvious, it's already commented, and to my mind the > extra attention gets in the way more than it helps. > > I agree that it would be better to advise the compiler directly about the > optimization that's needed there, if GCC etc. ever get around to doing that. > (I'm pretty sure the optimization advice is still needed, by the way: static > compilers typically don't have enough information, and even dynamic ones > would have troubles with those two.)
I feel that the false positives should be sufficiently rare that it's worth leaving it on by default. It might catch me doing something stupid even before I run a test :-) Contrast with options like -Waggregate-return that coreutils, grep, diffutils, gzip and others disable via configure.ac. If you'd prefer a file-wide directive or some other syntax in kwset.c, I'll be happy to adjust.
