On Mon, May 26, 2014 at 04:12:48PM +1000, Graeme Lee wrote:
> >Based on your diffs, I end up with the diff below. For now, only
> >test-built and packaged on amd64; I'll do some testing with some
> >of my old .ly fikes next weekend. Testbuilds (and real functional
> >tests) on other platforms (at least i386) are welcome, of course.
[...]
> I've tested it now on both i386 and amd64 platforms
>
> I ran a 50+ page score and a 30+ page Violin Etude book (among other things)
> on both platforms running current as of the 22nd of May.
Thanks for testing and for your initial diff. I'll run some light
tests on my own scores and then put it in.
> Of note is that
> the amd64 platform uses double the memory footprint to compile the document
> and hits the default ulimit on larger works.
IIRC (it's a long time since I had a *close* look at the lilypond
soures), there are tons of complicated (C++) classes with lots of
pointers to other complicated classes, and there are a lot of
connections between those classes and SCM objects, also done by
pointers. So it's not a big surprise that it consumes twice the
memory on a 64 platform compared to a 32 bit platform.
Ciao,
Kili
