zturtleman / ioq3ztm Goto Github PK

The dynamic lights (dlight) in quake3 limit light effect using a radio on x/y, 
but not z. This causes all light to be cylinders which effect all heights. So a 
rocket explosion at the bottom of the pillar lights up the whole vertical area, 
which looks kind of weird.

My initial tests of using a 3D texture to solve this, seem to work fine. Still 
needs some clean up. I plan to try to get this (optionally, disabled by 
default) in ioq3.

Static win32 and win64 libs for Ogg and Vorbis were replaced with full source 
(so it's easier to compile on Mac and other platforms). The Code::Blocks engine 
project needs to be updated to use the local Ogg and Vorbis.

Unlagged is a lag-compensation "game" modification for hit-scan weapons.

It's present in the WolfET GPL release.

Currently the crosshair for player 1 is in the "Game Options" menu. It should 
be possible to select the crosshair for the other three players as well. It 
doesn't seem like they should be under game options. Maybe put in player setup 
menus?

What steps will reproduce the problem?
1. Run spearmint
2. load map
3. add splitscreen player

Well the models for players usually refresh when the scores are shown.(I can't 
remember what variable it is to turn this off) But i think it would be a good 
idea to either remove this. As getting everyone to hit the score button or 
waiting for them to die at the same time is a bit of an inconvenience. Or make 
local clients automatically load model changes.

Clean up and merge support for using TrueType Fonts for all text drawing from 
Turtle Arena.

Pros;
  * The QVM compiler does not meet Debian Free Software Guidelines, whereas LLVM does.
  * LLVM supports C++, which would allow the game logic to be written C++.

Neutral;
  * The LLVM bitcode is less limited (and thus less secure) then QVM, but no less so than DLLs (which is what newer id Software games have).
  * As far as I know the LLVM bitcode would not have memory protection, but I am interested in removing it anyway (see issue 19).
  * I think the LLVM bitcode could call any non-static internal engine function (`trap_*` functions would not be necessary), which I think the DLLs can do as well.

Cons;
  * This makes compiling game logic more complicated. (Either people need to install clang and llvm-link, or I need to include the source in the code repo and add to build system.)


It seems to me using LLVM bitcode would be like having cross-platform DLLs 
(which are easier to load from pk3s?).

brightness doesn't seem to change. Not sure if this is just an ubuntu linux 
issue

Port Code::Blocks projects from Turtle Arena, allows building using a IDE on 
Windows and Linux.

Team Arena's UI always get servers from client. If you change master server and 
don't refresh the list, server for the wrong master are shown.

Cache servers for each master server in client, then have Team Arena's UI get 
correct cached servers.

May also want to change q3_ui to not cache servers, instead always get from 
client like missionpack ui. It would make the two UIs more consistant and use 
less RAM.

(Look into including in ioq3 as well.)

Turtle Arena adds a hardcoded bot named "Random". Adding it will select one of 
the other bots at random. Easier to just add a bot, "addbot Random".

Exactly what the title says

Currently cg_drawGun affects all local clients. Should there be per-client 
cvars?

Well first of all rend2 does not work what-so-ever. This might have been a 
previous issue but I'm not sure so i figured I’d put it. Also I get crashes 
and freezes when i try changing the video resolution with opengl11 in 
fullscreen, there is also spaces between the local clients screens. If you need 
anymore info let me know.

There are various functions called from CG_AddCEntity in cg_ents.c, such as 
CG_General and CG_Item, which are only called once per-frame and so can't use 
cur_lc/cur_ps in them!

What steps will reproduce the problem?
1. Downloading subversion code
2. Trying to compile code
3. Fails with DED_CC code/sys/sys_win32.c

What version of the product are you using? On what operating system?

Windows 64 Bit
&
Linux Cross compiling for 32 and 64 bit (same failure)

Please provide any additional information below.

attached output

Switch to ZeniMax's GPLv3+ (with additional terms) license used by RTCW SP/MP, 
Wolf:ET, and Doom 3. This will allow including bug fixes and features from them.

I specifically plan to include bug fixes and features from Wolf:ET.

I would like to port the Wolf:ET game code to ioq3ztm, allowing Q3A and Wolf:ET 
to both be played on the same engine. I imagine it wouldn't be hard to do the 
same for RTCW MP.

Need to fix linear fogging in Rend2. Probably going to need to make changes 
GLSL fog shader(s).

I would like to remove (or at least change the behavior of) the QVM memory 
protection. Currently it requires allocating extra unused RAM (it's more than 
64MB in Turtle Arena) and doesn't allow memory to be shared by game, cgame, and 
ui (which would benefit Turtle Arena, less memory usage and faster load times).

Removing the memory protection would make QVMs less safe, but not more so than 
using DLLs (which were used by the Wolfenstein id Tech 3 games and later id 
Software games).

What steps will reproduce the problem?
1. Change sv_maxClients while in splitscreen
2. Select "Restart Arena" from pause menu
3. Game crashes

What version of the product are you using? On what operating system?
r301 linux

NOTE: It only crashes like this when in splitscreen.

I noticed that players were not joining teams autocratically.

In line 889 of ui_startserver.c wait needs to be at least 9 to fix this.

All the splitscreen clients should be treated the same (think of them as 
different control layouts / multiplayer profiles). There shouldn't be a "main 
client."

Currently the client executable does all of the communication to the server for 
all splitscreen clients through the main client connection. It would be vary 
difficult to switch to a different client, on client and server.

Maybe the "client connection" could be separated from "client player". The 
"client connection" would not be dependent on a "client player". It would store 
info needed for the net traffic, pure authentication, and pointers to "client 
players".

Another option would be to actually have the client executable communicate the 
extra local clients separate from main client. This would increase network 
traffic, use more memory (to hold additional snapshot?), and just seems like a 
big hassle though.

There are still some issues with capsule collision.

I think players can get stuck in each other sometimes.

Players landing on players often don't get groundEntityNum set. Due to the 
sphered bottom and top?

Running into a crouching player makes runner move up. Trying to step up on the 
player?

There may be other issues. I wonder if just using a vertical cylinder would fix 
some or all of the issues.

HUD in widescreen viewports always draws aspect correct, placing things at 
left, center, and right. Add top, center, bottom positions and fix scaling.

WolfET foliage support was added to opengl1 renderer, it should be added to 
Rend2 as well.

Currently there can only be 1024 game entities. Increasing the limit to 4096 
shouldn't have that big of negative effect on the game, other than increased 
RAM usage.

Turtle Arena is stable with the increased limit, but as far as I know it has 
not been tested with more than 1024 game entities.

(Doom3 supports 4096 game entities too by the way.)

Support rotated, non-vertical, capsules for entity and player collision. Useful 
for walking on walls, spheres, and stuff. Games that do this: Prey, Super Mario 
Galaxy, etc.

Add support for disabling splitscreen at compile time. Useful for derivative 
games which do not want splitscreen or for platforms where it doesn't make 
sense (iOS, Android, etc).

This means vms will need to get max supported splitscreen players from client, 
instead of at compile time.

Having Key_GetCatcher separate for each local client would be useful if each 
local client could open menu in game separately. Though this would mean the 
menu could only use the players control binds, so not useful for chat or 
setting their name. Might be useful for team and player model select.

sv_public controls whether the server sends heartbeats to master servers, 
whether it replies to "getinfo" requests, and whether it is possible for 
clients to connect.

sv_public 1 - send heartbeats to master servers
sv_public 0 - no heartbeats
sv_public -1 - don't reply to "getinfo", causes server browsers to think server 
not running
sv_public -2 - disallow clients to join server

The changes also allows non-dedicated client hosting a server to advertise on 
master server.

Currently the HUD in ioq3ztm is always aspect correct, but fixed in the 4:3 
area in the center of the screen. In Turtle Arena the HUD is aspect correct in 
widescreen, but use the whole screen by selecting left/center/right placement 
for all drawing.

If a function needs cur_ps/cur_lc in cg_events.c it should take an argument! 
Events only happen once, not for each local client. Using cur_ps/cur_lc is 
never good there.

Abusing the cur_* variables is not good. The cur_* variable are not really that 
good to begin with...

Was still trying to use ioquake3.x86_64 for ioquake3.app

Attached corrected make-macosx.sh

Also the make-macosx.sh file doesn't have the right chmod on it from source to 
be able to run.

HUD is only drawn for last splitscreen player when using Rend2 renderer.

maps won't start

Requesting servers from master dpmaster.deathmask.net...
0 servers parsed (total 0)
Loading level q3dm1...
********************
ERROR: CL_ParseServerMessage: Illegible server message
********************
----- Server Shutdown (Server crashed: CL_ParseServerMessage: Illegible server 
message) -----
---------------------------

This is with r366

Rend2 dylibs should probably be added to the renderer objects in the mac build 
script

Integrate support for building for / running on Android, based on Kwaak3.

Support rendering Vector Graphics (possibly loaded from SVG images). For 
example, this could allow UI to look clear at all resolutions.

I would like to keep vector data in memory instead of rendering to pixels on 
load, allowing clear drawing at any size. The main goal is static images, but 
may look into animated SVG as well.

Requires research and testing.

playerState_t and entityState_t are game specific and should not be used in 
client and server.

Thoughts about solving this:

playerState_t and entityState_t networking needs to be moved to game/cgame. See 
MSG_WriteDeltaPlayerstate, MSG_ReadDeltaPlayerstate, MSG_WriteDeltaEntitystate, 
MSG_ReadDeltaEntitystate.

Server needs to read/write some parts of playerState_t and entityState_t, such 
as viewheight and clientNum. Might be a good idea to add a game VM call to 
get/set them instead of trying to manage multiple structs or casting points to 
different types. Example: viewOrigin = (float*)VM_Call(GAME_GET_PS_VAR, 
clientNum, GETPS_VIEWORIGIN);

svc_baseline cannot be sent in svc_gamestate as cgame isn't loaded yet (and 
cannot be loaded as svc_gamestate contains fs_game and list of pure pk3s to 
use). So baseline must be sent later, or possibly removed?

snapshot_t in cg_public.h includes playerState_ts. It could be replaced with 
void*, but it must be a data struct and not a pointer as cgame QVM it limited 
to 32 bit pointers and the client could be 64 bit. It will have to be removed 
and done another way. Extend CL_GetSnapshot to have cgame give location for 
playerState_ts to be copied into.

Enabling r_fastsky resulting in only the last viewport being seen in 
splitscreen, others are black. The renderer is clearing the whole screen before 
drawing each view.

There is not quite working capsule (aka cylinder) trace/collision code, it 
isn't used anywhere currently. It would be nice to merge capsule fixes from 
Tremulous so that capsules could be used instead of bounding boxes for various 
entities.

Use for players by default. If not to much work, add a cvar to choose bbox or 
capsule?

What steps will reproduce the problem?
1. Attempt to compile under VS2010

What is the expected output? What do you see instead?
I expected it to compile without changes, but all profiles fail to compile.

What version of the product are you using? On what operating system?
r114. Windows 7 x64

Please provide any additional information below.
After changing line 2462 in files.c in the quake3 project from "const int 
MAX_EXTS = 8;" to "#define MAX_EXTS 8", the "Debug TA" and "Release TA" 
profiles compiled successfully, but "Debug" and "Release" did not. I would 
recommend changing this line if it doesn't create errors with MinGW in order to 
support VS2010, and renaming the "Release" and "Debug" profiles to "Release 
(old)" and "Debug (old)", respectively, or outright deleting those profiles.

There are only two small glitches.

1. Player 2-4 have no crosshair.
2. Player 1 has a crosshair, but crosshair is not centered with 2 Players.

Reported by Tobias Kuehnhammer.

Need to support multiple listeners in the base and OpenAL sound systems. Sounds 
should play relative to all players.

Added framework for multiple listeners. OpenAL just ignores splitscreen sounds. 
Base sound, while it does play splitscreen sounds, could be better.

clc_moveLocal was added to allow client to send user commands for splitscreen 
local clients without breaking q3 network protocol. Now that I've dropped 
backward compatibility, there is no reason for them to be separate.

Merge arrow key support for Team Arena UI from Turtle Arena.

See "#ifdef IOQ3ZTM // ARROWS" and the like.

(Look into including in ioq3 as well.)

I compiled the code with no errors, but when i started the engine it gave me a 
user interface error.

MacOSX 10.6.8


Attached crash log