Hi Robert -- Sorry, looks like your forward to me got buried in my inbox.
Fortunately I saw this one.
I'm not as familiar with the loader code as I once was. It was originally
written by Brede, of course, and I haven't looked at it closely since I wrote
the exporter 4 or 5 years ago. I have no idea why it's using strings to parse
the palette record.
I'd say that empty OpenFlight files are pretty corner-case (no one has ever ran
into this issue until now), so special-casing it in such a way that doesn't
interfere with the mainstream code path would probably be the best approach. I
didn't look at the patch, but Nathan's description of it sounds like the right
thing to do.
Hope this helps,
-Paul
On 3/29/2012 2:00 AM, Robert Osfield wrote:
Hi Nathan,
You submission did make it on to osg-submissions and I reviewed it but
wasn't clear on the intention of original code and the wasn't
comfortable with the modification so I posted a query about the
original function to Paul Martz but haven't yet had a reply.
Robert.
On 28 March 2012 22:20, Nathan Monteleone<[email protected]> wrote:
nbmont wrote:
Hi,
Saw this in vs2010 when I tried to load an .flt file freshly made in creator
but with nothing in it. You get a string index out of range exception in
PaletterRecords.cpp, line 73 where it does
uint32 paletteSize = in.readUInt32();
// Entries in vertex pool found by offset from start of this record.
const int RECORD_HEADER_SIZE = 4;
const int OFFSET = RECORD_HEADER_SIZE+sizeof(paletteSize);
std::string buffer(paletteSize,'\0');
in.read(&buffer[OFFSET], paletteSize-OFFSET);
The problem is that for an empty vertex palette, OFFSET equals paletteSize, so
OFFSET is past the end of buffer. This code normally doesn't have a problem
because it's reading 0 bytes into this location anyway, but iterator debugging
catches the out of range index.
A simple if (OFFSET< buffer.size()) fixes the problem. I attached an amended
file with the fix, plus a test model that reproduces the problem.
I know it's a little silly to be loading empty .flt files, but I can't control
what people try to load and *technically* it's a valid file.
Hope that's helpful -- thanks for all the wonderful work on OSG.
Nathan
bump -- did this make it to the e-mail list?
------------------
Read this topic online here:
http://forum.openscenegraph.org/viewtopic.php?p=46657#46657
_______________________________________________
osg-submissions mailing list
[email protected]
http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org
_______________________________________________
osg-submissions mailing list
[email protected]
http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org
_______________________________________________
osg-submissions mailing list
[email protected]
http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org
