hello. i don't use the windows version much anymore but i was trying out the
3.6 release this weekend. now, this is on WinME, which I know sucks, so put
your jokes away ;) it came with my laptop and i leave it installed for the
occasional game.

anyway, i've had some issues with the parsing phase(i think) of povray 3.6
when using a mesh2 (exported from wings). pov crashes when I make a change
to the mesh2 object in povray such that the resulting mesh2 has an incorrect
syntax, then try to render it. i'd expect a parse error, not a pvengine
crash. the resulting dialog box says something like (to paraphrase)
"pvengine has caused something awful to happen, press OK or DEBUG". please
note that wings exports correctly and pov will render the image as is. it's
only if i screw up the syntax that it crashes.

Pressing "debug" doesn't do anything. it simply redraws the dialog box. so
i'm forced to click "OK" and pov finishes crashing. Now, i'd like to have
debug work. I'd assume that it would open up a stack trace, or some kind of
debugger where i might get some juicy information to pass along to the pov
team. I couldn't replicate this bug on a different win2k machine, and I
assume not many poeple run WinME.

So my question is, does anyone have a suggestion to get this "debug" button
to work? I believe the last time I ever tried using such a thing was in
win98 when I had Visual Studio installed. I no longer have Visual Studio. Or
can someone suggest a debugger that I can use to run pov with to see what
happens? Free, trial or shareware please. What does the Pov team use in such
a situation?

thanks for your time. any other suggestions on tracking this down would be
appreciated. i'll post real details when I have some. I want to make sure
it's 100% reproduceable and not caused by some other programs running,
before sending people on a wild goose chase.

-r

Post a reply to this message

In article <40cf0a0b$1@news.povray.org> , "Ross" <rli### [at] everestkcnet>
wrote:

> Pressing "debug" doesn't do anything. it simply redraws the dialog box. so
> i'm forced to click "OK" and pov finishes crashing. Now, i'd like to have
> debug work. I'd assume that it would open up a stack trace, or some kind of
> debugger where i might get some juicy information to pass along to the pov
> team. I couldn't replicate this bug on a different win2k machine, and I
> assume not many poeple run WinME.

That dialog and debug button are not from POV-Ray!  They are from the
system, more precisely, you have some development tools (most likely from
Microsoft) installed, and those make the system offer you to debug any
application that crashes.  However, that only makes sense for your own
applications.  Usually that button does nothing elsewhere.  And, as said, it
is not a dialog or button controlled by the application!

    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

Thorsten Froehlich wrote:

>>I'd assume that it would open up a stack trace, or some kind of
>>debugger where i might get some juicy information to pass along to the pov
>>team. 
> 
> They are from the system, more precisely, you have some development
> tools (most likely from Microsoft) installed, and those make the
> system offer you to debug any application that crashes.

And quite rarely is such debugger output of any significant value when 
trying to find the reason for the crash. A lot valuable is a detailed 
report of the crash: what did you actually do, what software/hardware do 
you use etc.

Severi S.

Post a reply to this message

"Thorsten Froehlich" <tho### [at] trfde> wrote in message
news:40cf35bc@news.povray.org...
> In article <40cf0a0b$1@news.povray.org> , "Ross" <rli### [at] everestkcnet>
> wrote:
>
> > Pressing "debug" doesn't do anything. it simply redraws the dialog box.
so
> > i'm forced to click "OK" and pov finishes crashing. Now, i'd like to
have
> > debug work. I'd assume that it would open up a stack trace, or some kind
of
> > debugger where i might get some juicy information to pass along to the
pov
> > team. I couldn't replicate this bug on a different win2k machine, and I
> > assume not many poeple run WinME.
>
> That dialog and debug button are not from POV-Ray!  They are from the
> system, more precisely, you have some development tools (most likely from
> Microsoft) installed, and those make the system offer you to debug any
> application that crashes.  However, that only makes sense for your own
> applications.  Usually that button does nothing elsewhere.  And, as said,
it
> is not a dialog or button controlled by the application!
>
>     Thorsten
>

 right, i understand that the debug dialog box is not a povray thing. i do
not have any MS development tools installed however. i might have at one
point, and that could be left over somehow. i would bet that is the case.

in any event, do you have a recommendation on tracing what happens that
causes the crash? or should I just make a step by step documentation of how
to reproduce the crash and submit a formal bug report?

thanks

Post a reply to this message

> in any event, do you have a recommendation on tracing what happens that
> causes the crash? or should I just make a step by step documentation of how
> to reproduce the crash and submit a formal bug report?

the easiest way for us to fix it is if you can put a reproducable scene file
(as small as physically possible considering it has a mesh in it) into a zip
file and post it in a binaries group of this server or on the net where we can
get at it, then let me know. I'll then try to track the problem down, provided
it is reproducable on any of the platforms I have available here.

I can't suggest any way to get useful debugging on your platform without the
aid of something like visual studio. If Intel has a trial version of their
debugger (don't recall if they do or don't) then that may or may not help.

-- Chris

Post a reply to this message

"Chris Cason" <new### [at] deletethispovrayorg> wrote in message
news:40cff7b3@news.povray.org...
> > in any event, do you have a recommendation on tracing what happens that
> > causes the crash? or should I just make a step by step documentation of
how
> > to reproduce the crash and submit a formal bug report?
>
> the easiest way for us to fix it is if you can put a reproducable scene
file
> (as small as physically possible considering it has a mesh in it) into a
zip

Well it seems that the size of the mesh might be key in reproducing the
problem. i cannot reproduce it with a mesh of a single face forming a
triangle. nor a 6 faced box mesh, just a large mesh containing thousands (i
think something like 4500) vertices. and I think the problem comes from
povray trying to print out the parse error message in the "Messages" tab. it
tries to print out the vertices for this mesh, and might be running out of
memory somehow? i don't know how, i have severeal hundred megabytes disk
space for virtual memory and 128M of physical memory. i only speculate that
it's a memory thing because when it tries to print out a few lines of the
beginning of the error message (before it tries to print out the vertice
list) it gets really slow, then eventually fails when it tries to print the
vertice list.

in 3.5, the error messages are much more succint and it doesn't try to
display the list of vertices. i tried turning off all error messages in 3.6
by adding -GA in the toolbar command line options area but it still tried
printing the error and subsequently crashed. Is -GA not what I want if the
goal is to turn off the message? I have not tried redirecting the messages
to a file yet, maybe sometime this weekend. My attention span faded and I
went back to working on a real scene.

so i hesitate to post a substantially large mesh (honestly i don't know how
big this mesh file is in bytes) especially since it might not be
sufficiently large depending on how much memory you have, or anything like
that. basically here is a shell of a pov file indicating what I do, camera
and lights ommited i've changed them enough to know they don't cause or
prevent a crash:



#declare t1 = texture {pigment {rgb 1}}
#declare some_big_mesh = mesh2 {
 vertex_vectors {...}
 normal_vectors {...}
 texture_list {
//  1,                  // this is what works
  0,                    // this, a parse error, causes a crash.
  texture {t1}
 }
 face_indices {...} /* needs to be of the form: face_indices
{number_of_faces, <a, b, c>, texture_index, ... }*/
 normal_indices {...}
}

object {some_big_mesh}


Again, this is a bug that occurs on parsing of an incorrect file. so it may
not be too important on a priority list.

Computer specs:
WinME
AMD Mobile Duron 800MHz
128MB RAM

Post a reply to this message

alright. i downloaded wings3d at work to test it out this mesh2 bug thing to
see if it happens on Win2k professional. it is reproduceable. i made a
sphere and subdivided it a few times to get 7680 polygons. the resulting
exported scene was just under 1MB

exported to pov, rendered, and it renders normally. inserted the parse error
as described in my previous post, and pov slowed down a bit in it's error
message, but it succesfully generated the message without crashing. I tried
rendering the same file with the same incorrect pov code again and pov
crashed saying "pvengine.exe has generated errors and will be closed by
Windows. You need to restart the program. An error log is being created."

the log is created in drwtsn32.log, and attached here (hopefully thats ok, i
can delete this post or would not mind if the server admin deletes it if
they deem it neccesary)

win2k machine has 256MB RAM. i think the other details are in the log file.

thanks, and sorry if this is a bad place to post an error log.

Post a reply to this message

Attachments:
Download 'poverror.log.txt' (17 KB)

On Fri, 18 Jun 2004 12:58:46 -0400, "Ross" <rli### [at] everestkcnet> wrote:
> alright. i downloaded wings3d at work to test it out this mesh2 bug thing to
> see if it happens on Win2k professional. it is reproduceable. i made a
> sphere and subdivided it a few times to get 7680 polygons. the resulting
> exported scene was just under 1MB

From what you are describing it suggests me such scene is possible with
HF_macros. And indeed I can cause strange behaviour there

#version 3.5;

#include "shapes.inc"

camera {
  location  <0.0, 0.5, -4.0>
  direction 1.5*z
  right     x*4/3
  look_at   0
}

light_source { <-30, 30, -30> , 1 }

plane { y, -1 pigment { rgb 1 } }

object{
  HF_Sphere(function{0}, yes, yes, 100, yes, "eksport.inc", 0, 1, 0)
  #include "eksport.inc"             
  pigment { red 1 }
}

Above scene creates file "eksport.inc" with size over 1 MB. It renders fine.
Now if I modify macro HFCreate_ in shapes.inc from text:
  #write(_HFMACRO_OUTPUT_FILE,"face_indices {"
to
  #write(_HFMACRO_OUTPUT_FILE,"face_indices {k"
to force parse error POV-Ray behaved strange. In one run it crashed. In other
run it hanged in rendering mode with neither parse error nor crash nor
rendering.

HTH, Chris

ABX

Post a reply to this message

"ABX" <abx### [at] abxartpl> wrote in message
news:gk96d09247au438uo9esrfns1hi11datll@4ax.com...
>
> From what you are describing it suggests me such scene is possible with
> HF_macros. And indeed I can cause strange behaviour there


yay, i'm not crazy! ;)

Post a reply to this message