THIS THREAD IS NO LONGER ACTIVE, PLEASE USE THE FOLLOWING URL INSTEAD: http://ktx.qw-dev.net

Current known bugs:

Reported bugs to be verified/reproduced:
+ End of game damage stats seem to be broken since last dmgfrags modification, they might include all damage done, target pented or not

Fixed bugs:
+ Autotrack as in ktpro

Current Accepted Feature Requests:
+ Implement race mode

Submited feature requests:
- Workaround the fact that ezquake developpers named some command speed, speed2...: aliaslist in client should show which "cmd cm number" to type to switch to speed or speed2

Added features:
+ MVD demos and streams now contains autotrack information
+ KTX now supports VWEP protocol extension to display current weapon model carried by the player to other instead of a generic weapon
+ End game statistics now have % + direct hits for both rl and gl
+ Instagib mode (instead of dmm8)

End stats damage bug reported by Renzo.

Implement dmm8 mode

Be able to vote for the map speed in some sort of workaround, maybe not kt'x area. but anyway

Stop calculating lg accuracy after enemy has died, maybe not even a worthy feature..

I have no idea what dmm8 is

You mean player speed? the one capped at 320 since forever? The one we all overcome by hoping like aroused rabbits?

If it's already there, it's not that big so we will keep it...

nope not player speed but the map speed :F http://ftp.unsix.net/pub/games/quake/id1/maps/speed.bsp

dmm8 is rail and instakill on buckshot and boomstick (and ofcourse the rail is instakill too) u can chek it out in ktpro if there's some server up with it..

Indeed... or need to hit really hard on ezquake devs X(

Well, ktx-svn already has instagib mode, maybe we can wait and see if dmm8 is still needed

add autotrack such as ktpro has it!

Please, please, please fix the sv_login 1 command.

I am trying to get a semi-public server going with sv_login 1 and sv_forcenick 1 but both of these commands do not work like they should.

sv_login 1 fails with a hung authentication session when running the KTX mod or KTPro, but not when running other mods like Rocket Arena.
sv_forcenick 1 fails to force the proper nick on the first connection to the server. Typing 'reconnect' once connected allows the server to force your nick.

The sv_forcenick issue is probably a MVDSV bug.

If these commands can be made to work it will help me get a no-anons server going that we can use to collect accurate statistics and run league games.

anyone who got the latest KTX qwprogs.qvm ? , i can only get a qwprogs.so-file (using MVDSV 0.28-SVN)
and is there any new KTX mappacks for download? i have notices some issues with map-errors for instance with dm5a.bsp

Done by qq.
Needs testing.

Indeed.

Why? .so/.dll is considerably better performing (cpu% wise) than .qvm.

And more insecure and unstable afaik.

Native support is unstable compared to virtual machine?
That's new.

Why more insecure?

Not really, its due to the nature of the memory protection in the qvm.
Sure, any bug is still serious... but references to null pointers that would crash a native vm harmlessly reference a small chunk of memory in the vm.
This means that a qvm can _never_ segfault... at least not inside the qvm itself.
This makes it more stable, as you can at least just change map.


If disconnect (or anyone else, sorry disconnect!) can get a back door in ktpro by exploiting a mod console command, just imagine what he can do if he can change the return offset inside a native progs by exploiting any buffer overflows that might exist inside the native code. Whilst you can overwrite the return offset inside a qvm, doing so is fairly safe, as the server will just bail out. Worst case it'll segfault on a dodgy read access, but this is much better than allowing a remote exploit, for server admins at least.

But yeah, qvms run an order of magnitude slower than native code, slower than qc too (for normal use). Either way the major overhead is within the physics code, and in the mvdsv qvm implementation this is still handled inside the engine, so its really not all that significant however you look at it... Unless of course you manage to exploit an infinite loop somewhere in the code... Again, with a qvm there's something you can do about that, though I'm not sure if mvdsv does or not. I suspect not, so that's fairly moot.

So really, qvm vs native is a question of whether you trust the gamecode coders or not.

Thanks for the lesson, interesting reading

And also if you intend to run more than few servers (that are actually used and not idling) on fairly good computers.

feature: dunno if this should be ktx or mvdsv job, but some admins would really appreciate the possibility of denying more than X connections per IP, where X would be possible to set
what for? mostly as a weapon for script kiddies like goq who run 10 bots playing on ffa servers or joining 4on4 server with bots just to get admin priviliges from an elect

Already opened: http://sourceforge.net/tracker/index.php?func=detail&aid=1775137&group_id=118599&atid=681549

how about a proper calculator for rocket and grenade damage? seeing ones own accuracy there would be nice =)
and keeping rl and gl separate would be best =)
showing just direct hits is not what we waaaant! =)
I think most players are working on their lg accuracy and like seeing it but nobody knows their rl accuracy.. so..

Easy, will display both

Hey deurk, any progress with the sv_login and sv_forcenick bugs? AU players are inches from a QW utopia free from unnameds and fakenicked players, but we need those bugs fixed

I'm also interested in any additions to KTX that will allow for more granular statistics to be exported in log files. Timestamps for all log entries would seem a logical progression. We could then calculate how many hours or days (weeks?) players have spent in a server, how many games they have played, longest 'streak', etc. People just love stats. The more information we can export to .txt, the more interesting statistics we can gather.

I mentioned on SF but also exporting the cells fired vs cells hit (used to calculate LG%) at the end of each match would allow admins to build accurate stats on players over time.

I'm also all for Rille's suggestion of RL%. That would be sweet. Hell, when not just export all weapon stats (hit/fired) at the end of each match?

Oh and another thing, how about RADIUS support for use with sv_login?

We could set up QW authentication servers (not unlike Valve's old WON servers) that server admins could reference to ensure no fakenicks are being used in league servers.

...hmm that might be a bit ambitious.

Detailled statistics have been implemented in SVN in the form of an xml output.
An external project is beeing started to provide a parser for those statistics.
I will only slightly change end of game stats for % and stuff but will not add detailled info there.

If you want to know more or help, contact me on #ktx

Not yet... Will try to give it a look this week-end.