On Tue, 2010-06-01 at 13:24 +0100, Claude Heiland-Allen wrote: > On 01/06/10 09:55, Dave Griffiths wrote: > > On Sun, 2010-05-30 at 07:44 +0200, gabor papp wrote: > >>> some random memory corruption or race condition is happening - are there > >>> any tips for debugging fluxus with valgrind? > >> as far as i remember you need to use the > >> –vex-iropt-precise-memory-exns=yes parameter in valgrind otherwise you > >> get false alarms. i also had to suppress opengl driver errors, but you > >> might not get those with nvidia. > > > > Seems like we introduced something recently, or it could be an > > interaction between specific versions of plt scheme and fluxus, see > > also: > > http://www.reddit.com/r/programming/comments/c9epm/programming_music/c0r38o2 > > > > Claude - what version of PLT scheme are you running? > > I was lazy and just installed from package manager, is it still > essential to compile PLT from source? Fluxus seems to work, apart from > the random segfaults...
I have a feeling that we've changed some things to work with newer versions of plt scheme which crash in older versions. It may be a coincident, but Kassen found a bug that caused crashes in version 4.2.4 which I reported and fixed locally in my PLT source, perhaps these versions have similar problems. > I just checked the package lists, Puredyne 9.10/9.11 DVDs have: > > plt-scheme 4.1.5-1 Right, I need to downgrade plt to this version and see if that's the problem. cheers, dave
