Re: an attempt at using the new id tools

Bernd Kreimeier (Bernd.Kreimeier@NeRo.Uni-Bonn.DE)
Thu, 27 Jun 1996 20:35:25 +0200 (MET DST)

Date: Thu, 27 Jun 1996 20:35:25 +0200 (MET DST)
From: Bernd Kreimeier <Bernd.Kreimeier@NeRo.Uni-Bonn.DE>
Message-Id: <199606271835.UAA15863@marvin.nero.uni-bonn.de>
To: quake-dev@gamers.org
Subject: Re: an attempt at using the new id tools

>error with Solaris 2.5 qbsp

I am compiling qbsp from *.cc now (introduced a few
header files by splitting bsp5.h, removed one redundancy
already - there is nothing like plowing your way through
somebody else's code :). It does not abort with a Bus
Error under Solaris 2.5.

>It is in tjunc.c/FindEdge() (if this is not
>a bogus), in the VectorSubtract macro (do not use macros,
>my mother always told me). I will leave this to those
>who prefer a C version :-).

No warranty: as soon as I made it a function with proper
prototype, I got notified that it is called with float* here
instead of double* (as anywhere else, it seems). This might
well end up in accessing memory that the process does not
own, hence a Bus error. However, just replacing all floats
in bspfile.h with doubles did not fix the Solaris problem
with the C version. I'd rather have inline and const...

To check the output, now, I'll have to get that mipmap WAD,
too. The dm1.prt and dm1.h2 files are identical (Linux qbsp.c
run compared with Solaris qbsp.cc run). The *.h1 are not.
Hmmm.

b.