Server message format

At its basic level, a server message consists of the ID of the last client command that it received along with a list of server operations. There are several different types of server operations. To identify which type of operation it is, each operation in the list starts with an 8-bit identifier code. The last operation in the list is identified with a special EOF code.

Structures

nop

8 bits

0x01 nop indicator

gamestate

8 bits	0x02 gamestate indicator
32 bits	server's last command ID
(<configstring> | <baseline>)*
8 bits	0x08 EOF indicator
32 bits	client ID
32 bits	randomly generated checksum seed value

configstring

8 bits	0x03 configstring indicator
16 bits	config ID
string	config value

baseline

8 bits	0x04 baseline indicator
<entity>

server-command

8 bits	0x05 server-command indicator
32 bits	command ID
string	command value

download

8 bits	0x06 download indicator
16 bits	block number
if {block number} == 0x0000: // first block
32 bits file size if {file size} == 0xFFFFFFFF: // error indicator string error message else 16 bits block size data data stream, {block size} bytes long
else
16 bits block size data data stream, {block size} bytes long

snapshot

8 bits	0x07 snapshot indicator
32 bits	server time
8 bits	last frame num
8 bits	snapshot flag bits
8 bits	area mask size
data	area mask, {area mask size} bytes long
<player-state>
<entity>*
10 bits	0x3FF end of entities indicator

entity

10 bits	entity ID
1 bit	removed-entity indicator
if removed-entity != 0x1:
1 bit changed-entity indicator if changed-entity == 0x1: 8 bits max changed field index loop {max changed field index} times: 1 bit changed-field indicator if changed-field == 0x1: 1 bit is-not-zero indicator if is-not-zero == 0x1: <int-value> | <float-value> // choice depends on field index else: // value = 0

int-value

N bits

value, where N depends on field index

float-value

1 bit	full-float indicator
if full-float == 0x1:
32 bits value, as C float
else:
13 bits value, as unsigned int; subtract (1<<12) to return signed int

If a field has floating point values, but its current value is a small integer (< ±4096) without a fractional part, then it is transmitted as a 13-bit unsigned integer instead of transmitting it as a full 32-bit float. To convert a small signed integer that is between -4096 and 4095 to a 13-bit unsigned integer, we simply add 4096 (1<<12) to the number before we send it and then subtract 4096 after it's received.

player-state

8 bits	max changed field index
loop {max changed field index} times:
1 bit changed-field indicator if changed-field == 0x1: <int-value> | <float-value> // choice depends on field index
1 bit	changed-state indicator
if changed-state == 0x1:
<values> // player stats <values> // persistent player stats <values> // weapon ammo <values> // power-up timeouts

values

1 bit	changed-values indicator
if changed-values == 0x1:
16 bits changed value bitmask loop for each bit set in {changed value bitmask}: 16 bits value

See also

• Netchan message
• Client message format