On Sat, Jun 14, 2014 at 10:05:48PM -0500, Bruce Dubbs wrote:
> Dan McGhee wrote:
> >
> >This concept and "-j" have now become important to me and I'd like to
> >learn some more before I start.
> 
> man make
> 
> https://www.gnu.org/software/make/manual/html_node/Parallel.html#Parallel
> 
 I'm going off on a tangent here, and this is for beyond-BLFS,

 One of the packages I build on my desktops is mediainfo (originally
this was to let me check the durations of the streams in a video
when I was having difficulty getting youtube to accept it - ISTR
that was NOT the problem, and I now used it to check the tags in my
flac files).

 With recent versions, this has a bash script which appears to be
trying to use all the cores during the build (and I cannot see
anything _obviously_ wrong in what it is doing).  For me, on a box
with 4 cores (real or hyperthreaded) the desktop becomes unresponsive
(even the clock stops updating for long periods of time), but the
build does eventually complete and the system is again usable.  But
on my box with only two cores it does not complete in the amount of
time I have been prepared to wait.

 My guess is that for some reason the script decides there are far
more cores than actually exist, and that is what brings the box to
its knees (ISTR the code is C++, so everything probably assumes that
memory is an infinite resource).  In this case, I use a sed to
force it to use regular "make" (and therefore -j1 because the script
which runs the compile ignores $MAKEFLAGS) instead of the Zen_Make
function in the tarball's script.

 From this, I draw two conclusions :

1. Running too many jobs in parallel _will_ bring a system to its
knees.

2. Some packages are just _different_. ;-)

ĸen
-- 
Nanny Ogg usually went to bed early. After all, she was an old lady.
Sometimes she went to bed as early as 6 a.m.
-- 
http://lists.linuxfromscratch.org/listinfo/lfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page

Reply via email to