Path: chuka.playstation.co.uk!news
From: "Jon Prestidge (alias Moose)" <zmoose@premier.co.uk>
Newsgroups: scee.yaroze.programming.3d_graphics
Subject: OT buffer puzzle
Date: 19 Jun 1997 11:03:22 GMT
Organization: PlayStation Net Yaroze (SCEE)
Lines: 67
Message-ID: <01bc7ca1$2688a700$LocalHost@default>
NNTP-Posting-Host: client86ac.globalnet.co.uk
X-Newsreader: Microsoft Internet News 4.70.1155

Hello, this (as many things do) is perplexing me......

I'm sure this next section of code where stuff for the ordering tables are
set-up will look familiar - it seems to be in many yaroze demos on the web
site....

   #define OT_LENGTH	9
   static GsOT Wot[2];			/* Handler of OT */
   static GsOT_TAG wtags[2][1<<OT_LENGTH];
   .
   .
   .
   Wot[0].length = OT_LENGTH;
   Wot[0].org = wtags[0];
   Wot[1].length = OT_LENGTH;
   Wot[1].org = wtags[1];

   GsClearOt(0, 0, &Wot[0]);
   GsClearOt(0, 0, &Wot[1]);


It just did not look right to me -- ie I would have thought the Wot[n].org
would have had to have been set using a "&" before the wtags[n] to get it's
address -- but there again I'm not a C guru and so I assumed that there
must be some C-weirdness where arrays are automatically by reference... and
after all that bit of code seems to be widely used -- so it must work...
mustn't it?!?!   Anyway I included it in a program I was starting to write.
  When going throught the program on the debugger (MetroWerks debugger
V1.6) I noticed that the address that the pointer Wot[n].org was being set
to was 9, which just so happend to be the constant OT_LENGTH!  Obviously
address 9 can't be right can it?  It certainally was not even remotely like
the address of the wtags array.  So anyway I amended the code to be as
follows...

   #define OT_LENGTH	9
   static GsOT Wot[2];			/* Handler of OT */
   static GsOT_TAG wtags[2][1<<OT_LENGTH];
   static GsOT_TAG *temp_GsOT_TAG_pointer;
   .
   .
   .
   Wot[0].length = OT_LENGTH;
   temp_GsOT_TAG_pointer = &wtags[0][0];
   Wot[0].org = temp_GsOT_TAG_pointer;

   Wot[1].length = OT_LENGTH;
   temp_GsOT_TAG_pointer = &wtags[1][0];
   Wot[1].org = temp_GsOT_TAG_pointer;

   GsClearOt(0, 0, &Wot[0]);
   GsClearOt(0, 0, &Wot[1]);


... and this looked much better under the debugger... the pointer
Wot[n].org now pointed to the same address as wtags[n].   I probably could
have gotten away with just putting  Wot[n].org = &wtags[n][0]  but I wanted
to see things happen in the debugger. 

My program works OK with my amended version of theOT set-up.
I can't say whether or not it would have worked with the code the normal
way because the program didn't work then anyway for other reasons.

I'd be very interested on people's comments... could it be the debuger
giving-out wrong data,  or have I got the wrong idea about pointers and
stuff,  or have I lost my marbles... or .... other?

Jon