[Arkshine]

Looking at the engine dll, the array size for the sounds are not 255. There are 3 arrays.

- PF_precache_sound_I -> 512
- PF_precache_model_I -> 1024 but the check is set to 512 (Something which could be changed)
- PF_precache_generic_I -> 527

[ot_207]

Originally Posted by Arkshine Looking at the engine dll, the array size for the sounds are not 255. There are 3 arrays. - PF_precache_sound_I -> 512 - PF_precache_model_I -> 1024 but the check is set to 512 (Something which could be changed) - PF_precache_generic_I -> 527

You forgot that the sound index that can be transmitted via SVC_SOUND does not go further than 255. That is one problem.

[Arkshine]

I can't agree with you. :p

To find where SVC_SOUND is used : PF_sound_I -> SV_StartSound -> SV_BuildSoundMsg

Based on Quak1SDK, SV_BuildSoundMsg() looks like :

PHP Code:

int SV_BuildSoundMsg (edict_t *entity, int channel, char *sample, int volume, float attenuation, int flags, int pitch, origin?, datagram? )
{       
    int sound_num;
    int field_mask;
    int ent;
    
    if ( volume < 0 || volume > 255 )
        Sys_Error( "SV_StartSound: volume = %i", volume );

    if ( attenuation < 0 || attenuation > 4 )
        Sys_Error( "SV_StartSound: attenuation = %f", attenuation);

    if ( channel < 0 || channel > 7 )
        Sys_Error( "SV_StartSound: channel = %i", channel );

    if ( pitch < 0 || pitch > 255 )
        Sys_Error( "SV_StartSound: pitch = %i", pitch );
       
    if ( sample )
    {
        if ( sample[0] == '!' || sample[0] == '#' )
        {
            field_mask = flags | SND_SENTENCE(?); // 1 << 4
            
            if ( Q_atoi( sample + 1 ) > 1536 )
            {
                Con_Printf( "invalid sentence number: %s", sample + 1 );
                return 0;
            }
        }   
    }

    sound_num = SV_LookupSoundIndex( sample )
    
    if ( !sound_num || !sv.sound_precache[sound_num] )
    {
        Con_Printf( "SV_StartSound: %s not precacheed\n", sample );
        return 0;
    }

    ent = NUM_FOR_EDICT( entity );

    if ( volume != 255 )
        field_mask |= SND_VOLUME; // 1 << 0
        
    if ( attenuation != 1.0 )
        field_mask |= SND_ATTENUATION; // 1 << 1
        
    if ( pitch != 100 )
        field_mask |= SND_PITCH; // 1 << 3
        
    if ( sound_num > 255 )
        field_mask |= SND_?; // 1 << 2

    MSG_WriteByte( datagram, SVC_SOUND );
    MSG_StartBitWriting( datagram );
    MSG_WriteBits( field_mask, 9 );
    if ( field_mask & SND_VOLUME )
        MSG_WriteBits( volume );
    if ( field_mask & SND_ATTENUATION )
        MSG_WriteBits( attenuation * 64, 8 );
    MSG_WriteBits( channel, 3 );
    MSG_WriteBits( ent, 11 );
    MSG_WriteBits( sound_num, sound_num <= 255 ? 8 : 16 );
    MSG_WriteBitVec3Coord( origin );
    if ( field_mask & SND_PITCH )
        MSG_WriteBits( pitch, 8 );
    MSG_EndBitWriting( &sv.datagram );
    
    reutrn 1;
} 


And looking at SV_LookupSoundIndex(), basically check the "sample" in the array and returns its index if found.

So, there is not a problem like you say, there is probably some bits manipulation like you can see in the code. If I have well understood. :p

[ot_207]

Oh, I thought it was the same as quake.

[Arkshine]

Just based on it with a lot of changes. But It can help to understand some things.

[Drak]

What I was refering to, is what "ot_207" said. But, not just the precache list/limit/whatever. Like Arkshine was talking about, open the quake source, and look through "zone.c" it's all the memory allocating stuff. This file, is exactly the same ( or super close - as a decompiled showed) to the HL engine.

When you use precache_sound/model() - it add's to all kinds of internal limits (that have been discussed millions of times). But behind all the HL related stuff, memory wise it loads it into. When you use precache_generic() it loads into a "dynamic" hunk. This data will be whiped/re-loaded upon, a map change. When it's anything else, it permanently keeps that data (just the string to the file name) within the engine. Now, this data does get re-loaded, but it doesn't get "pushed" around. You can set model onto and ent, display it in-game, etc. When it's precached. This also can cause problems. (Cache_TryAlloc() and Hunk_Alloc() errors)

OVERALL. DON'T PRECACHE MODEL/SOUNDS - IF YOU JUST WANT A PLAYER TO DOWNLOAD THEM.

[ConnorMcLeod]

I was thinking that this tut would be more about a snippet like this :

- Make sure the file exists before trying to precache it (prevent servers crashes).
- Ability to format the file path.

PHP Code:

PrecacheFile(fmt[], any:...)
{
    new szFile[256], n
    n = vformat(szFile, charsmax(szFile), fmt, 2)

    if( n > 4 )
    {
        if( equali(szFile[n-4], ".mdl") || equali(szFile[n-4], ".spr") )
        {
            if( !file_exists(szFile) )
            {
                new szError[512]
                formatex(szError, charsmax(szError), "File %s doesn't exists", szFile)
                set_fail_state(szError)
                return 0
            }
            return precache_model(szFile)
        }
        else if( equali(szFile[n-4], ".wav") )
        {
            new szFullPathFile[262] = "sound/"
            add(szFullPathFile, charsmax(szFullPathFile), szFile)
            if( !file_exists(szFullPathFile) )
            {
                new szError[512]
                formatex(szError, charsmax(szError), "File %s doesn't exists", szFullPathFile)
                set_fail_state(szError)
                return 0
            }
            return precache_sound(szFile)
        }
        else // .mp3 .tga
        {
            if( !file_exists(szFile) )
            {
                new szError[512]
                formatex(szError, charsmax(szError), "File %s doesn't exists", szFile)
                set_fail_state(szError)
                return 0
            }
            return precache_generic(szFile)
        }
    }
    return 0
} 


[Lawer]

PHP Code:

client_cmd( id, "spk subfolder/sound.wav" ) 


You can use EmitSound or no?

[GXLZPGX]

Originally Posted by Lawer PHP Code: client_cmd( id, "spk subfolder/sound.wav" )  You can use EmitSound or no?

Yes