thanks. much lower priority for me (and i'm volunteering to fix it), but... what should i do about the mac here? #if around the linux-only/mac-only signals (SIGPOLL is linux-only, SIGEMT is mac-only, for example).
also -- something i didn't know until just now -- it turns out that the mac doesn't have real-time signals. i've attached a patch, but it adds #ifs to lib.c which doesn't currently have them, so i don't know if you'd prefer me to move those bits to portability.c? (i also didn't know whether you'd prefer big "linux" vs "macOS" #ifs or per-constant #ifs, but assumed the former given the complaints i've heard about GNU code...) On Mon, Jun 10, 2019 at 1:21 PM Rob Landley <[email protected]> wrote: > > On 6/10/19 10:23 AM, enh via Toybox wrote: > > ping? > > Applied. > > > -*- > > > > i did think of a way we could maybe add a test for the real-time > > signals: collect the value from `kill -l SIGRTMAX` and check that > > `kill -l SIGRTMAX-1` is `$((sigrtmax - 1))`. (you might think that > > SIGRTMIN/SIGRTMAX are constants, but SIGRTMIN varies based on how many > > real-time signals the libc implementation has taken for itself, and on > > MIPS at least SIGRTMAX is different. i think the BSDs differ from > > Linux too.) > > They're compile time constants, I know they vary by architecture. > > > SIGHUP is probably (?) the same everywhere so we could have similar > > for `kill -l HUP` and `kill -l 1`. > > I checked the kernel source with "grep -r SIGHUP arch/*/include include | grep > define" and yes, they are. COuldn't tell you about macos. > > > given that `kill -l` was broken even for the easy cases before my > > patch, i've at least attached the trivial HUP/1 test... > > Also applied. > > Rob
0001-Fix-mac-build.patch
Description: Binary data
_______________________________________________ Toybox mailing list [email protected] http://lists.landley.net/listinfo.cgi/toybox-landley.net
