R: R: [Plib-devel] Update on 3ds loader

[Paolo L. <p.l...@ci...>]

Wolfram Kuss wrote:

> Are you (Per/Paolo) working under Windo$?

Yes, NT4 at office and 98 at home. I use the full Cygnus dev-sys (gcc =
2.9.2).=20

> Paolo, shall I send you the file?

Yes, please.
To all: what about to start using the anonymous FTP space for test =
cases? (even if I got problems to go to =
ftp://plib.sourceforge.net/pub/plib/)

For now you could try to put such initialization in function =
ssgLoad3ds()
:
  mat_tfnames =3D new char*[MAX_MATERIALS];	// this line already exists
  for ( int _i=3D0; _i < MAX_MATERIALS; _i++ )
    mat_tfnames =3D (char *)NULL;

This is implemented in experimental version of the loader at =
ftp://ftp.cira.it/paolo/plib/ssgLoad3ds.cxx that you can test in the =
meanwhile.
I'm not able to test such modification since my model, that I'll manage =
to distribute soon as test case, is fully textured.

To Per:
- this version also set node name for VTable nodes to the actual object =
name read in the file - previously the related material name was given =
to the node (I did it, sorry). This is useful for later navigation in =
the graph, e.g. to set/change properties to nodes for which the 3DS name =
is known (in my case: don't lit the far cylinder showing the sky =
texture);
- thanks for the strdup() - I didn't realize how useful it is, but, once =
discovered, I'm using it even too often!

-----Messaggio originale-----
Da:	Wolfram Kuss [SMTP:w_...@rz...]
Inviato:	venerd=EC 7 aprile 2000 18.20
A:	pli...@li...
Oggetto:	Re: R: [Plib-devel] Update on 3ds loader

My problem is not that something is loaded too often
or some texture is not found, but that it crashes because it is=20
dereferencing an invalid pointer.

I am working under (on??) Windows NT.
It crashes while loading chevy.3ds because it tries to
reference an undefined pointer (the hex address is cdcdcdcd,
which is Visual C++ way of saying bad pointer).

I set a break point in parse_material and another one in=20
parse_mapname.
The first three times it finds a breakpoint is in
parse_material. Then it goes into parse_mapname.
num_materials is three.

mat_tfnames is obviously a array of pointers. The array is defined.
However, the pointers are not.

MAYBE the first three materials are un-textured,
the texturenames are therefore not defined and when it=20
finds the first mapname it tries to compare it to the undefined,
previous ones?

Are you (Per/Paolo) working under Windo$?

Paolo, shall I send you the file?

Per, have you looked at the newest CVS-code, perhaps=20
it is different from the code you have? Do you use MSVC 6?
Do you do a debug build?

Bye bye,
Wolfram.

_______________________________________________
plib-devel mailing list
pli...@li...
http://lists.sourceforge.net/mailman/listinfo/plib-devel