just how much autotools got faster, or: just forget about dolt

[Dan Nicholson]

On Sun, Nov 23, 2008 at 6:51 AM, Ralf Wildenhues <Ralf.Wildenhues at gmx.de> wrote:
> For all dolt junkies, and whoever claims autotools and esp. libtool
> are performance hogs: yes, they are still not the fastest thing in
> the world.  But we are getting better:
>
> As an example, I built the ompi/ subdirectory of OpenMPI on a two-way
> GNU/Linux system using bash, with
> 1) Autoconf 2.61, Automake 1.10.1, Libtool 1.5.26,
> 2) git master versions of same tools,
>
> doing
> a) a full 'make -j3',
> b) measuring just build system overhead, which in case of (2) is pretty
> well approximated with
>    \time make -k -j3 LIBTOOLFLAGS=-n am__mv=:
> and in case of (1) involves placing a fake 'mv' program early in $PATH.

Just for a data point, here is the same experiment on xserver git with
git/old autotools and with/without dolt. This was basically single run
of each test, so I'm sure there's tons of statistical error. I did do
this while the box was otherwise idle, though.

old autotools (full build):
real    2m59.809s
user    3m25.916s
sys     1m52.775s

old autotools + dolt (full build):
real    2m3.060s
user    2m50.078s
sys     0m38.099s

old autotools (dry run):
real    1m13.512s
user    0m46.539s
sys     1m23.739s

old autotools + dolt (dry run):
real    0m6.719s
user    0m8.574s
sys     0m2.389s

git autotools (full build):
real    1m58.631s
user    2m58.349s
sys     0m30.855s

git autotools + dolt (full build):
real    1m51.437s
user    2m48.725s
sys     0m26.904s

git autotools (dry run):
real    0m16.292s
user    0m21.050s
sys     0m5.069s

git autotools + dolt (dry run):
real    0m7.106s
user    0m8.859s
sys     0m2.615s

So, it seems that the autotools have improved dramatically. Well done.

This is a Fedora 9 system, so "old" autotools is autoconf-2.61,
automake-1.10.1 and libtool-1.5.24. I had to patch dolt a bit to get
the dry run behavior. Basically, just prepend the actual command with
: when -n was passed.

--
Dan