Todd C. Miller <[email protected]> wrote: > On Wed, 20 Nov 2019 10:17:09 -0700, "Theo de Raadt" wrote: > > > Todd C. Miller <[email protected]> wrote: > > > > > On Wed, 20 Nov 2019 07:38:46 -0700, "Theo de Raadt" wrote: > > > > > > > Kernel environment cannot use userland includes. > > > > > > Some other systems have sys/stdbool.h, we could as well if we wanted > > > to. The simplest approach is to move include/stdbool.h -> > > > sys/sys/stdbool.h and make /usr/include/stdbool.h a link to > > > sys/stdbool.h as we do for stdarg.h and stdint.h. > > > > But is it really neccessary? Has int really caused everyone that > > much harm? > > From a readability standpoint, it is nice to be able to use true > and false. You can use those with int as well as with bool.
I don't think the type of readability stdbool.h brings results in 1 fewer bug in the resulting code, also I think adding it to a body of code late risks introducing bugs. More to the point, whenever I see codebases which mixes true/false, 0/-1, and 0/1, I don't pick up a vibe of "better readability". And it gets even worse when one see that stdbool.h has 3 variations internally which are not 100.0% compatible. I don't see any value bringing it to the kernel.
