On 2015-Aug-20 10:08:37 -0500, "Matthew D. Fuller" <fulle...@over-yonder.net> wrote: >On Thu, Aug 20, 2015 at 08:32:07AM +0100 I heard the voice of >Mathieu Arnold, and lo! it spake thus: >> >> I'm not exactly sure what you're all complaining about. The port >> just works fine, it does take a few seconds for make to iterate over >> all options, but it's because the port has way too many options. > >It's a pretty sizable difference... > >% svn co -r394572 [ Mk, print] >% cd print/ghostscript9 >% time make -V MAINTAINER >doc...@freebsd.org >0.903u 0.023s 0:00.92 100.0% 834+264k 82+0io 6pf+0w > >% ( cd ../../Mk && svn up -r394573 ) >Updating '.': >U bsd.options.mk >U bsd.port.mk >Updated to revision 394573. > >% time make -V MAINTAINER >doc...@freebsd.org >81.908u 0.070s 1:21.99 99.9% 848+264k 82+0io 5pf+0w
On my E3-1230v2 system (not a slouch), the difference is 1s to 110s - which is in line with Matthew's results. A two-orders-of-magnitude slowdown in make startup is unreasonable and I formally request r394573 be backed out. I can see the benefits of the functionality but the current cost is too high. -- Peter Jeremy
pgpl7TlEjyf80.pgp
Description: PGP signature