Reduced header file inclusions

Regis Boudin regis at boudin.name
Wed Apr 11 17:05:13 MDT 2007


Hi Jens,

Thanks for the new patch :)

On Wed, 2007-04-11 at 22:23 +0200, Jens Seidel wrote:
> On Tue, Apr 10, 2007 at 04:35:36PM +0200, Jens Seidel wrote:
> > On Tue, Apr 10, 2007 at 06:56:03AM -0700, Robby Stephenson wrote:
> > > Thanks for the patch, Jens.
> 
> I checked now most other header files and attached a new patch.

Juste tested here with g++ 3.3, 3.4, 4.0, 4.1, 4.2 and 4.3, all happy to
build, so all clear on the GCC compatibility front :)

> I did not checked src/translators/btparse/ and src/translators/pilotdb/
> as these directories are probably from external sources. I wonder
> whether you synchronize these with upstream? Do you report bugs (such as
> the wrong memory access in the pilotdb exporter)?

btparse looks abandonned upstream. The patches Robby submitted years ago
were never applied, there has been no release since it was copied into
the tellico tree either. Sometimes I have a look at it and clean up the
various warnings left, which are harmless anyway... No idea about
pilotdb, though.

> I'm not sure how both patches affect tellico. But at least in theory
> the compile should be faster and include graphs smaller (no need to
> rebuild large parts if a header file changed; code parser such as
> doxygen will be faster).

To get a rough idea, I just compiled with and without the patch, with
and without the --enable-final option passed to configure The results
running "time make >/dev/null 2>/dev/null" are :

* without enable-final :
+ without patch :
real    4m52.304s
user    4m22.264s
sys     0m20.989s
+ with patch :
real    4m53.394s
user    4m22.156s
sys     0m21.013s

* with enable-final :
+ without patch :
real    2m51.778s
user    2m36.998s
sys     0m6.484s
+ with patch :
real    2m41.322s
user    2m35.018s
sys     0m6.296s

Tested on an up-to-date Debian Sid, with and Athlon64 X2 3800+ and 1GB
of RAM, each test done after an "echo 3 >/proc/sys/vm/drop_caches". I
also used the --diable-rpath and --enable-new-ldflags all the time, but
I doubt it would bring a difference between the cases.

So, overall, not much gain in time apparently ; but having cleaner
includes is always a good thing anyway :)

Regis




More information about the tellico-users mailing list