| Age | Commit message (Collapse) | Author |
|---|
|
|
|
-Fix: make sure this glitch can never happen again (ai.c)
|
|
-Fix: the AI speed control is done by the AI-core, individual AIs don't have to do it (so, AIs were delayed twice ;)
-Add: Support for AI-network-clients (an AI, connecting to a remote server)
-Fix: minor AI-core problems
|
|
in the network code
|
|
serverside
-This means that one company can only have one setting for renew and replacing
more clients will not fight due to different settings anymore
-This is a needed step in the line to fix autoreplacing dualheaded locomotives
NOTE: savegame revision bump (peter1138 + me in coop)
|
|
|
|
|
|
|
|
|
|
|
|
Added support for Mersenne Twister random number generator (not implemented in network yet)
Wrap player randoms around #ifdef
|
|
|
|
|
|
of abusing _decode_parameters as text buffer. This should prevent several possible buffer overruns and is a bit cleaner to use. As bonus it reduces the size of most command packets by 79 bytes.
|
|
- CodeChange: add comments to a very ugly part of the code (network_client.c); that enforces that the server gives an ID to the client.
|
|
- Fix: [ 1153926 ] All my settings in vain... IGNORED!
- Change: I hope I got it all right. Pressing 'New Game' (either choosing random or a preset scenario) and 'Create Scenario' will start a new game with the settings and difficulty in the intro menu. Using 'Load Game' and 'Play Scenario' will take the values from the savegame/scenario itself.
|
|
introduce stre{cpy,cat}, see string.h for their semantics
|
|
new header (though i think some of these function don't belong into strings.c)
|
|
|
|
'set rcon_pw <password>'
Which enables rcon. A client can now do:
'rcon <password> "<command>"'
The command will be executed on the server. (guru3)
-Fix: 'kick 1' did crash dedicated servers
-Fix: server password is now correctly saved
!!Warning!!: do not give your rcon password to people you do not thrust!
|
|
-Fix: MSVC6 project file update (bociusz)
-Fix: MSVC6 default language is now English (UK), no longer Swedish (bociusz)
|
|
- Added green dot if company income is positive (else red dot)
- Added lock icon if company is password protected
|
|
|
|
server
reads beyond the size of the packet
-Fix: [Network] A server no longer crashes when a client sends an
invalid DoCommand, but drops the client instead.
|
|
A server can ban people via ClientList
Both server and dedicated can do it via console:
'ban', 'unban', 'banlist'.
|
|
company-index was sent, which should have been the client-index
|
|
-Fix: removed some warnings
-Enabled ENABLE_NETWORK on windows again which I accidentally commented out.
|
|
|
|
company
went bankrupt, without joining the wrong company (or even failing to do so)
|
|
hardcoded strings)
|
|
now
stay a spectator even if someone else joins.
|
|
and 'void' where needed, prefixed all functions, typedefs and global
vars with 'Network' and organized all externals nicely.
|
|
|
|
They joined, they got the map, but it took some time before the new
player was created. In this time a player could do stuff, causing a
player-id-mismatch, and the player was kicked out. Now a player get's a
nice GUI which says: registering.. When that is gone, the player is
joined and can play safely. Tnx to Moriarty for bugging me with this bug ;)
|
|
-Codewise: [Network] Cleaned some code, removed some comment, changed
some wrong comment
|
|
Removed globalness of _openttd_revision and put all such ifdefs into one place. If server has a revision only the same revisions can join; if the server has no revision everyone can join. I reckon this should be a server-side option to allow people to join or not to join.
|
|
hopefully killed the windows-revision problem. If WITH_REV is defined,
for both Windows as *nix system _openttd_revision is filled with normal
info, else _openttd_revision is set to 'norev000'
-Fix: Small possible server-crash
|
|
(somehow
windows uses WITH_REV_HACK instead of WITH_REV :s)
|
|
unique id (generated via md5)
|
|
endian-problems. To fix this, we are sending dparams byte by byte
(instead of an uint32). Because of this dparam is sent
not-uint32-endian-safe, but char-endian-safe. Too bad dparam can no
longer be used for normal stuff (which is currently not the case) (tnx
to Tron and Bjarni)
|
|
|
