| Message ID | 6f25e6c1-c176-629f-321c-9b404c5d6de6@landley.net |
|---|---|
| State | Changes Requested |
| Headers | show |
>>>>> "Rob" == Rob Landley <rob@landley.net> writes: > My rule of thumb is that once a change is 7 years old, you can rely on > the installed base to have it. With that in mind, would you like to use > nproc to check how many processors are available? > http://git.savannah.gnu.org/gitweb/?p=coreutils.git;a=commit;h=74cf4cb26dcecd36eb45dc00dbd4587d9dc24a2f Why? What is the advantage? Purely cosmetical? Are there any well known distributions providing nproc and not getconf? > diff --git a/package/Makefile.in b/package/Makefile.in > index afd5d3a..5eed804 100644 > --- a/package/Makefile.in > +++ b/package/Makefile.in > @@ -12,7 +12,7 @@ HOSTMAKE := $(shell which $(HOSTMAKE) || type -p $(HOSTMAKE) || echo make) > # If the number of processors is not available, assume one. > ifeq ($(BR2_JLEVEL),0) > PARALLEL_JOBS := $(shell echo \ > - $$((1 + `getconf _NPROCESSORS_ONLN 2>/dev/null || echo 1`))) > + $$((1 + `nproc 2>/dev/null || echo 1`))) > else > PARALLEL_JOBS := $(BR2_JLEVEL) > endif > _______________________________________________ > buildroot mailing list > buildroot@busybox.net > http://lists.busybox.net/mailman/listinfo/buildroot
Hello, On Fri, 30 Dec 2016 14:14:46 -0600, Rob Landley wrote: > My rule of thumb is that once a change is 7 years old, you can rely on > the installed base to have it. With that in mind, would you like to use > nproc to check how many processors are available? > > http://git.savannah.gnu.org/gitweb/?p=coreutils.git;a=commit;h=74cf4cb26dcecd36eb45dc00dbd4587d9dc24a2f This patch lacks your Signed-off-by line, so we can't apply it as-is. Best regards, Thomas
Hello, On Fri, 30 Dec 2016 21:46:43 +0100, Peter Korsgaard wrote: > > My rule of thumb is that once a change is 7 years old, you can rely on > > the installed base to have it. With that in mind, would you like to use > > nproc to check how many processors are available? > > > http://git.savannah.gnu.org/gitweb/?p=coreutils.git;a=commit;h=74cf4cb26dcecd36eb45dc00dbd4587d9dc24a2f > > Why? What is the advantage? Purely cosmetical? Are there any well known > distributions providing nproc and not getconf? Well, the simplification provided by Rob looks useful to me. It's purely cosmetic, but we also do tons of other cosmetic changes. The big question is whether nconf is available even in the really old RHEL distros that some of our users have. Best regards, Thomas
>>>>> "Thomas" == Thomas Petazzoni <thomas.petazzoni@free-electrons.com> writes: > Hello, > On Fri, 30 Dec 2016 21:46:43 +0100, Peter Korsgaard wrote: >> > My rule of thumb is that once a change is 7 years old, you can rely on >> > the installed base to have it. With that in mind, would you like to use >> > nproc to check how many processors are available? >> >> > http://git.savannah.gnu.org/gitweb/?p=coreutils.git;a=commit;h=74cf4cb26dcecd36eb45dc00dbd4587d9dc24a2f >> >> Why? What is the advantage? Purely cosmetical? Are there any well known >> distributions providing nproc and not getconf? > Well, the simplification provided by Rob looks useful to me. It's > purely cosmetic, but we also do tons of other cosmetic changes. True, but for stuff depending on the details of the host OS we need to be careful, so it is interesting to know if this change is NEEDED somewhere, or just a nice-to-have cleanup thing. > The big question is whether nconf is available even in the really old > RHEL distros that some of our users have. Indeed, this was the kind of details I was missing in the commit description.
diff --git a/package/Makefile.in b/package/Makefile.in index afd5d3a..5eed804 100644 --- a/package/Makefile.in +++ b/package/Makefile.in @@ -12,7 +12,7 @@ HOSTMAKE := $(shell which $(HOSTMAKE) || type -p $(HOSTMAKE) || echo make) # If the number of processors is not available, assume one. ifeq ($(BR2_JLEVEL),0) PARALLEL_JOBS := $(shell echo \ - $$((1 + `getconf _NPROCESSORS_ONLN 2>/dev/null || echo 1`))) + $$((1 + `nproc 2>/dev/null || echo 1`))) else PARALLEL_JOBS := $(BR2_JLEVEL) endif