[Fluxus] nuke button crash

[Dave Griffiths]

On Thu, 2008-09-04 at 10:27 +0100, Dave Griffiths wrote:
> >> This is now fixed in cvs, but I'm trying to track down another elusive
> >> crash I got last night - which is proving to be difficult to spot.
> > are you sure you committed the changes? i don't see anything changed
> > apart from ribbon-prim.scm and poly-tools.ss.
> >
> > could you explain how you found the cause for the f6 crash, please? i
> > tried a couple of things, but i was unable to do anything with the
> > extensive list of scheme calls which came from the backtrace.
> 
> Yeah, I should have elaborated. In brief I fixed the symptom not the
> cause, there was a syntax error inside (poly-for-each-face).
> 
> It's my understanding that the fluxus module scheme code is compiled to
> bytecode when you start up fluxus. The new version of PLT does some
> further optimisation, or more likely has better error detection on the
> code it compiles. It was trying to print out an error that
> poly-for-each-face had a misplaced paren. However at the time during
> reset, there is no message handler, so it falls over trying to print the
> warning. Phew! :)

In trying to fix the crash underlying this, I've fixed a bug where the
output stopped printing in the repl after resetting with F6. I'll ask on
the PLT list about this crash.

> It was late and I was doing other things, so I just fixed the error and
> swept it under the carpet, along with the issue of why it didn't crash on
> startup, or where the error message goes to at startup...
> 
> I'm a bit more worried about this other crash as it happened while I was
> in the middle of livecoding something fairly gfx intensive[1] and wasn't
> even running anything (every-frame) so I think it's a crash inside the
> renderer, which we haven't had for some time.

I haven't seen this again, despite running the script for hours and
hours in the debugger. I'm inclined to think it may have been a form of
the bug above now, as I haven't come across anything else. Things feel
pretty stable otherwise.

cheers,

dave