Vis data cannot be created for any of my STRIFE1.WAD files for MAP04 -- neither in game, nor with glVis 1.5 after glBSP 2.20 "-v2" / "-v3". a) 28,372,168 bytes (April 1996) - version 1.0 b) 28,377,364 bytes (May 1996) - should be patched to version 1.3 c) 28.116.953 bytes = b) after applying ZenNode 1.2.1 Please tell me which information you might need to verify it before I start to upload any files (I have own webspace in case you need some bigger files). The same files worked well in previous versions of Vavoom, and with the original STRIFE.EXE.

It's a bug in latest version of glBSP which seams to create some bad nodes. Try previous version.

You are probably right, glVis 1.5 after glBSP 2.10c works well. So I hope Andrew Apted will still be available, his last reply in a DoomsDay board thread sounded not too promising. On the other hand: Are you preparing for glBSP v5 nodes?

I already added support for them in Vavoom.

[quote="Janis Legzdinsh":10rozrys]It's a bug in latest version of glBSP which seams to create some bad nodes. Do you have more information, like what in particular is wrong with the GL nodes? P.S. MAP04 must be in the full version of STRIFE which I don't have. That means I can't debug this problem right now.

It also happens with Hexen's map02 I think.

Repeating my reply to the SourceForge bug: I started debugging this but ran out of steam. The problem has an equal chance of being in glVIS itself, which is not my program, and I'm not willing to spend more time debugging someone else's program. I'm marking this bug "Rejected" until a specific fault in glBSP (e.g. seg XXX in map YYY is bad) can be reported.

So I hope Janis (or someone else with good C knowledge) can support you here, Andrew. As I wrote, Version 2.10c of glBSP does not provoke a problem in glVis, therefore the created nodes must differ. But I am not experienced enough in C programming and debugging to compare both tools (I prefer Delphi). __ P.S.: According to "Beyond Compare 2", the GWA files created by glBSP 2.10c and 2.20 (using the same command line options) differ remarkably. Not only by additional BUILDER / TIME / CHECKSUM comments, but also in the binary data.

Fixed.

<!-- s8) --><img src="{SMILIES_PATH}/icon_cool.gif" alt="8)" title="Cool" /><!-- s8) --> Thanks!