I have had a sneaking suspicion that this "pet bug" of mine has not left
us completely yet:

* Macro bug (job000146)
  (POV can dereference a deallocated pointer if you return a local from a
  macro. Causes POV to crash.)
  Reported in:
"Known Bug" for Rune
http://news.povray.org/3BB4ED11.8EC69CEC@hotmail.com
strings operations crash with both compiles
http://news.povray.org/e3piut8anmf4cli4pnfmstadgc3nleqvb2@4ax.com


But in the last few betas it has surfaced so rarely that it has been
very difficult to chase it.

I'm running POV-Ray v3.5.beta.RC4.icl.win32 in Windows 98 on a 100 MHz
Pentium with 48 MB of RAM.


Here is the shortest code I could find that leads to this error message:
"The POV-Ray core rendering code caused an access violation exception."


#macro B(P, Q)
  #local G = dimension_size(P, 1);
  #local H = dimension_size(Q, 1);
  P
#end

#declare N = B(array[1] { 0 }, B(array[1] { 0 }, array[1] { 0 }))


I also found that another error message is triggered by some similar code:

#macro A()
  #local T = array[1] { 0 }
  T
#end

#macro B(P, Q)
  #local G = dimension_size(P, 1);
  #local H = dimension_size(Q, 1);
  A()
#end

#declare N = B(array[1] { 0 }, B(array[1] { 0 }, array[1] { 0 }))


Here's the error message:

  #local T = array[1] { 0 }

  T <----ERROR

Parse Error: Cannot pass uninitialized identifier as macro parameter.
Initialize identifier first.



Can anyone else confirm this ?


Tor Olav

Post a reply to this message

On Thu, 09 May 2002 03:42:00 +0200, Tor Olav Kristensen
<tor### [at] hotmailcom> wrote:
> Can anyone else confirm this ?

At first run it reported uninitialized identifier Q inside first macro. At
second run it crashed.

POV 3.5 RC4 icl on WinNT Sp 6 PII 233 with 128 MB

ABX

Post a reply to this message

In article <3CD### [at] hotmailcom> , Tor Olav Kristensen 
<tor### [at] hotmailcom>  wrote:

> But in the last few betas it has surfaced so rarely that it has been
> very difficult to chase it.

Yes, because the bug would only appear in a very specific case.  While I did
not get a crash, I got an incorrect error message and was able to track it
down based on that.

    Thorsten

____________________________________________________
Thorsten Froehlich, Duisburg, Germany
e-mail: tho### [at] trfde

Visit POV-Ray on the web: http://mac.povray.org

Post a reply to this message

> 
> On Thu, 09 May 2002 03:42:00 +0200, Tor Olav Kristensen
> <tor### [at] hotmailcom> wrote:
> > Can anyone else confirm this ?
> 
> At first run it reported uninitialized identifier Q inside first macro. At
> second run it crashed.
> 
> POV 3.5 RC4 icl on WinNT Sp 6 PII 233 with 128 MB

Thank you Wlodek.


Tor Olav

Post a reply to this message

Thorsten Froehlich wrote:
> 
> In article <3CD### [at] hotmailcom> , Tor Olav Kristensen
> <tor### [at] hotmailcom>  wrote:
> 
> > But in the last few betas it has surfaced so rarely that it has been
> > very difficult to chase it.
> 
> Yes, because the bug would only appear in a very specific case.  While I did
> not get a crash, I got an incorrect error message and was able to track it
> down based on that.

Thorsten, does this mean that it is fixed now ?


Tor Olav  =)

Post a reply to this message

> Thorsten, does this mean that it is fixed now ?

Sure, next time it will crash your OS as well ;-)

    Thorsten

Post a reply to this message