[fluxus] 3m (was: Re: building CVS on Ubuntu Feisty)
Alex Norman
alex at neisis.net
Thu Apr 26 11:19:26 PDT 2007
Hmm, so I updated to the svn version of plt, I did configure --enable-shared
and I did make 3m && make install-3m
then I compiled and installed fluxus
but when I try to run fluxus I get this:
load-extension: couldn't open
"/usr/local/lib/plt/collects/fluxus-0.12/compiled/native/x86_64-linux/fluxus-engine.so"
I have
/usr/local/lib/plt/collects/fluxus-0.12/compiled/native/i386-linux/
I am on a 64 bit box, I don't know where this i386-linux thing comes from, if it
is something in the plt install or in some environment variable [i don't see
anything about it] or the fluxus install,...
any ideas?
thanks,
-Alex
On 0, Dave Griffiths <dave at pawfal.org> wrote:
> I'm so glad to hear at least someone's got it working!!!
>
> > Just in case. (Someone correct me if im wrong here...)
> >
> > You also need to build mzscheme3m with
> >
> > > ./configure --enable-shared
>
> yes
>
> > and build the 3m thing (whats that?):
> >
> > > make 3m
>
> I *think* the svn version builds this by default now.
>
> 3m is a garbage collector the plt people have recently switched to using,
> which is more accurate than their previous one. I'm quite glad they've
> done this, it would affect fluxus scripts used for things like
> installations, as it fixes problems where memory would not be freed over
> long time periods.
>
> Gory detail:
>
> 3m is short for "moving memory manager" and allows allocated chunks of
> memory to be shifted around to reduce fragmentation. This doesn't directly
> change anything to do with how Scheme scripts are written (neither does
> the entire switch to plt either btw), but it makes writing the C++
> interface to Scheme *much* harder - hence it took me a long time to fully
> understand the situation, and track down all the bugs.
>
> I've been using 0.12 + 3m solidly for a couple of months now - including
> developing and performing with al-jazari, so I'm as confident as I ever am
> about it's stability.
>
> cheers,
>
> dave
>
>
>
More information about the Fluxus
mailing list