On Dec 9 11:52, Christian Franke wrote: > Corinna Vinschen wrote: > > I would prefer that SCHED_BATCH gets its own, single value. > > There's no good reason to add another ifdef for that. Why > > not just #define SCHED_BATCH 6? > > The idea was to keep the non-Cygwin value in sync with Linux. > https://github.com/torvalds/linux/blob/fac04ef/include/uapi/linux/sched.h#L111 > Of course we could drop this idea and use 6.
I see, but I wonder if Linux binary compatibility for flag values is really important. From the readability POV it's bad enough that we have two different values for SCHED_OTHER. And SCHED_SPORADIC is already breaking Linux binary compat since it clashes with SCHED_ISO (which, funny enough, isn't documented in sched(7), afaics). Maybe you'd like to ask on the newlib ML if somebody prefers Linux binary compat for SCHED_BATCH? Corinna
