What changed in EDuke32, build by build

related to the definitions of RANDOM_NEG in bunny.cpp, ripper.cpp and ripper2.cpp. Do so in a way that isn't re-introducing compiler warnings. This partially fixes compatibility with demos made for SW 1.2. Additionally, replace the 3 separate definitions of RANDOM_NEG with a common one within game.h.

sintable and radarang. sintable[512] was different in a 32-bit build.

functional with existing demo files, the code is known to be incompatible with demos made for SW 1.2.

brought back in r7873, this should restore the behaviors of DOS v1.2.

Fixes tiled automap rendering in SW, without breaking the status bar as in the case of r8766.

as done in r8766, since this breaks status bar output in tiled overhead map with Polymost. Do add comment about this for now.

Fixes non-deterministic white rectangles rendering bug in PCExhumed.

corresponding variable in rooms.cpp is used if DEBUG is defined to nonzero.

later. This also turns out to fix a Polymost-specific bug revealed in SVN r8711. Further remove a call to PicInView which does nothing.

a bit more usable in multiplayer, hiding the engine's console.

toggling MultiPlayQuitFlag on instead of QuitFlag in multiplayer games if a player tries to close the window. Further update MNU_QuitCustom to use CON_Quit. - In RunLevel, this fixes quit of game for all players. - In MenuLevel, this leads to transmit of PACKET_TYPE_MENU_LEVEL_QUIT. However, the work isn't complete, since the game isn't shut down on the peers' sides, leading to possibly undesired side-effects. Currently known issue: - If the master closes the window, the scores won't be shown for the master. Reason is that waitforeverybody will terminate the app.

use p->TeamColor instead of User[p->PlayerSprite]->spal, since it's possible the player left before the map was loaded.

quitevent in the game, and further call getpackets within StatScreen.

all players after level change. This is done by adding calls to getpackets within the BonusScreen and StatScreen while loops. It is related to a change from the DOS versions of 1997. Basically, dosendpackets immediately sends all packets via the commit driver under DOS, while in the mmulti port in use here, sending might be postponed. This was leading to a problem with the game entering waitforeverybody before showing the stats. In particular, it initially attempted to send a PACKET_TYPE_PLAYER_READY message, which could get postponed in the manner described above. In case it received PACKET_TYPE_PLAYER_READY messages from all other peers before the time arrived for sending pending packets, though, waitforeverybody would return, eventually leading to a call to StatScreen/BonusScreen, without sending the pending message until the user in question continued. Other peers would have to wait for the given player to confirm level change before they could see their own stat screens.

processing multiplayer params or waiting for players. It's been decided that Control won't call TerminateGame in this case, one implication being that voidsw.cfg is not updated.

Thanks to LittleTijn for testing and inspecting.

mostly from JFSW commits 7a95dee807c735de6588372e268af7a5c7cdc533 and 6137b48d6d432c34d94d08fea527b89539cc3adb. In particular, this makes Master/Slave usable for 2-player sessions, with SW's networking code behaving as if these were actually Peer-2-Peer sessions. Thanks to JonoF for his work on the support.

on the frame rate

params, but leave them for documentation. Note that currently, the only use of ComboMissileSeek is in the saveable_weapon_code array.

the functions' signatures (often due to being used as callbacks/hooks)

Note that the passed type is always ct_mainmenu.

with (possible) side-effects is written to an unused var.

used if previously commented out code was uncommented (including a few instances of uninitialized var reads).

packet of type PACKET_TYPE_MESSAGE. This reverts a tiny portion of JFShadowWarrior commit 6b01a4703d55ae3e131d8c11903ff2dcd78c15d8.

video mode change in USE_OPENGL=0 build.

in USE_OPENGL=0 builds, and further make setrendermode usable.

with minor changes.

as well as the wrong aspect ratio in 3D for Ken-Build.

1. Fix USE_OPENGL=0 builds. 2. Don't ignore the user's choice of BPP. 3. If eduke32.cfg has Polymer set to 1 and ScreenBPP set to 8 (e.g., the user changed from Polymer to Classic), don't tick Polymer and don't select an OpenGL mode.