vilyaem/Xrasher
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

engine: common: net_encode: fix inaccuracy in DT_TIMEWINDOW_* encoding

Browse source 
By avoiding double rounding, we get more accurate time on client when timebase
gets rounded down and target time gets rounded up:

```
>>> round(123.1) - round(124.51)
-2
>>> round(123.1 - 124.51)
-1
```
This commit is contained in:
Alibek Omarov 2024-04-17 06:09:49 +03:00
parent 1677835b45
commit 35783bcec2
1 changed files with 15 additions and 32 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

47
engine/common/net_encode.c
View file

@ -1034,33 +1034,17 @@ static qboolean Delta_CompareField( delta_t *pField, void *from, void *to, doubl
}
else if( pField->flags & DT_TIMEWINDOW_8 )
{
val_a = Q_rint((*(float *)((byte *)from + pField->offset )) * 100.0f );
val_b = Q_rint((*(float *)((byte *)to + pField->offset )) * 100.0f );
val_a -= Q_rint(timebase * 100.0);
val_b -= Q_rint(timebase * 100.0);
fromF = FloatAsInt( val_a );
toF = FloatAsInt( val_b );
val_a = *(float *)((byte *)from + pField->offset );
val_b = *(float *)((byte *)to + pField->offset );
fromF = Q_rint(( timebase - val_a ) * 100.0 );
toF = Q_rint(( timebase - val_b ) * 100.0 );
}
else if( pField->flags & DT_TIMEWINDOW_BIG )
{
val_a = (*(float *)((byte *)from + pField->offset ));
val_b = (*(float *)((byte *)to + pField->offset ));
if( !Q_equal( pField->multiplier, 1.0f ))
{
val_a *= pField->multiplier;
val_b *= pField->multiplier;
val_a = (timebase * pField->multiplier) - val_a;
val_b = (timebase * pField->multiplier) - val_b;
}
else
{
val_a = timebase - val_a;
val_b = timebase - val_b;
}
fromF = FloatAsInt( val_a );
toF = FloatAsInt( val_b );
val_a = *(float *)((byte *)from + pField->offset );
val_b = *(float *)((byte *)to + pField->offset );
fromF = Q_rint(( timebase - val_a ) * pField->multiplier );
toF = Q_rint(( timebase - val_b ) * pField->multiplier );
}
else if( pField->flags & DT_STRING )
{
@ -1144,6 +1128,7 @@ static qboolean Delta_WriteField( sizebuf_t *msg, delta_t *pField, void *from, v
int signbit = FBitSet( pField->flags, DT_SIGNED ) ? 1 : 0;
float flValue, flAngle;
uint iValue;
int dt;
const char *pStr;
if( Delta_CompareField( pField, from, to, timebase ))
@ -1210,19 +1195,17 @@ static qboolean Delta_WriteField( sizebuf_t *msg, delta_t *pField, void *from, v
}
else if( pField->flags & DT_TIMEWINDOW_8 )
{
signbit = 1; // timewindow is always signed
flValue = *(float *)((byte *)to + pField->offset );
iValue = (int)Q_rint( timebase * 100.0 ) - (int)Q_rint( flValue * 100.0 );
iValue = Delta_ClampIntegerField( pField, iValue, signbit, pField->bits );
MSG_WriteBitLong( msg, iValue, pField->bits, signbit );
dt = Q_rint(( timebase - flValue ) * 100.0 );
dt = Delta_ClampIntegerField( pField, dt, 1, pField->bits );
MSG_WriteSBitLong( msg, dt, pField->bits );
}
else if( pField->flags & DT_TIMEWINDOW_BIG )
{
signbit = 1; // timewindow is always signed
flValue = *(float *)((byte *)to + pField->offset );
iValue = (int)Q_rint( timebase * pField->multiplier ) - (int)Q_rint( flValue * pField->multiplier );
iValue = Delta_ClampIntegerField( pField, iValue, signbit, pField->bits );
MSG_WriteBitLong( msg, iValue, pField->bits, signbit );
dt = Q_rint(( timebase - flValue ) * pField->multiplier );
dt = Delta_ClampIntegerField( pField, dt, 1, pField->bits );
MSG_WriteSBitLong( msg, dt, pField->bits );
}
else if( pField->flags & DT_STRING )
{