behave itself a little more often. I can find no explanation why and it
seems like it should possibly do nothing, but it SEEMS to help.
Thanks for the suggestion Tim.
How view height changes in QW are passed from server to client HAS CHANGED
and you will need to update both client and server for things not to act
weird. You may also need to distclean before rebuilding..
protection.
sv_timekick: This cvar controls the number of times a player has to
be caught "cheating" before they get kicked. sv_timekick shows up in
serverinfo if changed from the default. If sv_timekick < 1, timecheat
detection is disabled. Default is 3.
sv_timekick_fuzz: This cvar affects how strict the protection is. The
higher the number, the more "fuzz" is applied, and the less strict the
detection code is. Raise this if your players are being kicked for packet
loss and lag. Default is 10, giving a fuzz factor of about 1 percent.
sv_timekick_interval: This cvar controls how often, in seconds, the time
tally is counted. Lowering this value increases the chance of false
positives, but helps to minimize the amount of damage a timecheating
player can cause. Default is 30 seconds.
Also, there was a bug in configure.in that caused glx not to be built at all on
my system.
NOTE: I changed SRC_DIR to srcdir (GNU `standard'), though I'm beginning to
wonder if that was so good.
are done (sorry if this steps on your toes with the view.c merge Deek) and
I'm almost positive flymode will now work. Even though view offset is
done, it won't.
The reason for this is that cl.stats[STAT_FLYMODE] is pretty much going to
ALWAYS be 0 on a standard server. Since 0 tells us that we're not flying,
this is fine. cl.stats[STAT_VIEWHEIGHT] is also going to be 0, but it
should be 22 for normal views. I could always assume this value is an
offset from 22, but that just seems lame to me. I'll either do it anyway
or we'll have to find a good opportunity in the connect cycle to set the
cl.qfserver qboolean to true.
I'm thinking about using an info key value for this, but we'd be better
served I think by coordinating with QSG to up the protocol version across
all engines.
gl_model.c had it as 512, but qw_server/model.c and common/model.c had it as
256. That wasn't the real problem, what it was was hipnotic (obviously) has
more than 256 models in "Research Facility", and so uquake-svga bobmed
Bill Currie
Joseph Carter
Loring Holden
Pontus Lidman
Jeff Teunissen
Anton E. Gavrilov
Marcus Sundberg
Zephaniah E. Hull
Dan Olson
Daniel McGregor