On 9/20/05, Paul Brossier <[EMAIL PROTECTED]> wrote: > On Tue, Sep 20, 2005 at 10:18:18AM +0200, Jean de Largentaye wrote: > > On 9/20/05, Paul Brossier <[EMAIL PROTECTED]> wrote: > > > severity 328871 wishlist > > > retitle 328871 please update to 0.9.3.1 > > > thanks > > > > > > On Mon, Sep 19, 2005 at 10:41:54PM +0200, Jean de Largentaye wrote: > > > > Ok, I'm working on getting it to work, and I have, without much trouble. > > > > > > yes me too. with libode, and either pyrex 0.9.3 or 0.9.3.1; the c files > > > produced are almost identical (additionnal (void *) casts in 0.9.3.1), > > > and compile pyode with the same errors. > > > > Wait.. I'm not following you here. What I undertand from what you're > > writing is that, in your case, whether using debian's 0.9.3 or the > > upstream 0.9.3.1, you get the same errors, and it doesn't build. Or do > > you mean the same warnings and it *does* build? > > yes, both compiled with warnings. > > > In my case, it wouldn't build with deb's 0.9.3-3 (with the cast > > errors) , but it did with 0.9.3.1 (without the cast errors). :) > > What is the situation? > > i do not get the error you see, at least not on powerpc. could you post > them here please? this would confirm the (voild*) cast is required. > > cheers, paul >
Well the error does not appear... *Danger Will Robinson: lack of rigor detected!* I hadn't noticed the ode_[no]trimesh.c files were already present in the distributed package. These were generated by a vanilla 0.9.3 pyrex... And these were the files that provoked the error. But once I purged these files, (and modified the setup script to use the right pyrexc binary...), it used the system's pyrex to generate them and it created working source code. So it seems I was wrongfully attributing the compilation trouble to deb's Pyrex, when it actually came from PyODE's distribution. I will notify them. I'm awfully sorry for the trouble :(. John
