Re: [palito-dev] last intro test core dumping

[Tom Barnes-Lawrence]

On Sat, Jun 28, 2003 at 12:03:27PM +1000, Michael Devey wrote:
> 
> and I'm about to go jetsetting myself.

 Hope you have a nice time!

> If palito is crashing at the end of the all unit test intro try making
> tree1.flush an empty file.  It's worked on every debian machine I've
> had it crash on so far.
> 
> yes that's lame I know but I haven't found out why it's crashing yet.
> incidently other unit files as a replacement may allow the game to
> run (from memory not platform01 or the palmtree one - they may all
> be sharing something in common which highlights this bug on some
> platforms)

 It hadn't occurred to me it would be the contents of the files that
would make a difference. I tried moving tree1.flush away and replacing
it with an empty file, and yes that worked.

Then I tried moving it back, and doing the same with palmtree01.flush
(I had investigated what was happening with gdb a few days back) instead.
This works too!! Now that's odd.
I tried moving the plamtree back, and now replaced the metal_maker file
with an empty file instead. Sure enough, this works as well!


 Running the x11 version under gdb, when it crashes, some flush functions
were being called by intro_draw()- specifically, intro_draw was doing

 cpu_exec(cl->code,0)
when *(cl->code) was
{filename = 0x80725b0 "palmtree01.flush", itrArray = 0x8071ec8, 
  varArray = 0x40029860, varCount = 10}

(note that it isn't tree1.flush here!)
and *(cl->code->itrArray) was
{type = 82, val = {real = 3.08285662e-44, integer = 22, 
    string = 0x16 <Address 0x16 out of bounds>}, line = 1, column = 1}

Don't know if that's any news to you. I'd imagine not.

Seeing what happens when you blank any of those files I tried just
confuses the hell out of me. Nothing I can come up with makes any
sense there.

  Tomble