[Powerlord]

Originally Posted by FlaminSarge My sig tells all. I'll look it up some time tomorrow.

Unless Valve throws an additional update in there, it appears that the Linux versions retain their same signaturess. I'm not sure how to check the Windows versions, though.

[Ravu al Hemio]

I checked with the gdc tools. Linux/OS X locations are given as symbol names, which gdc tries to find in the symbol table. Windows locations are given as byte patterns, which gdc tries to find in the code section of the DLL, making sure it finds each only once.

Both seem to be unchanged for the methods used by TF2Attributes.

[FlaminSarge]

Nothing broke with this with SteamPipe on Windows, so we're good to go.

[realwx]

Sort of still stuck in a circle of figuring out why item attributes are applying themselves to the "Action" slot despite filtering out item ID's numbers by checking against the string trie. It's currently taking Melee (or, if available, PDA/PDA2 onto Action.) Even if I remove the melee slot before TF2Attributes apply, then it will just apply the secondary slot's attributes onto the Action slot. (I did try replacing "decl" with "new" in several places, Sarge, but it only made a difference on the first round, not any subsequent ones)

Tested by viewing stats in the MVM maps. Any clues on whether I'm doing this all wrong?

public TF2Items_OnGiveNamedItem_Post(client, String:classname[], itemDefinitionIndex, itemLevel, itemQuality, entityIndex)
{
    new applyOnSpawn = GetConVarInt(cvarApplyOnSpawn);
    
    if(enabled && ( (applyOnSpawn == 1 && !IsFakeClient(client)) || applyOnSpawn == 2) )
    {
        decl String:tmpID[16];
        decl String:strAttrib[64];
        decl String:fullAttr[256];
        
        Format(tmpID, sizeof(tmpID), "%d__%s", itemDefinitionIndex, selectedMod);

        if (!GetTrieString(g_hItemInfoTrie, tmpID, fullAttr, sizeof(fullAttr)))
        {
            Format(tmpID, sizeof(tmpID), "%d__default", itemDefinitionIndex);
            if(!GetTrieString(g_hItemInfoTrie, tmpID, fullAttr, sizeof(fullAttr)))
            {
                PrintToConsole(client, "Could not load value from %s.", tmpID); //debug
                return;
            }
        }
        
        TF2Attrib_RemoveAll(entityIndex);

        decl String:attrArray[MAX_ATTRIBUTES][32];
        new num_attribs = ExplodeString(fullAttr, " ; ", attrArray, MAX_ATTRIBUTES, 32);

        for (new i=0; i < num_attribs+1; i += 2)
        {
            TF2II_GetAttribName(StringToInt(attrArray[i]), strAttrib, sizeof(strAttrib));

            if (TF2Attrib_IsIntegerValue(StringToInt(attrArray[i+1])))
            {
                new value = StringToInt(attrArray[i+1]);
                TF2Attrib_SetByName(entityIndex, strAttrib, Float:value);
            }
            else
            {
                TF2Attrib_SetByName(entityIndex, strAttrib, StringToFloat(attrArray[i+1]));
            }
        }
        TF2_AddCondition(client, TFCond_SpeedBuffAlly, 0.01);
    }
}

[FlaminSarge]

Have you tried flat-out rejecting certain indices or classnames instead of relying on GetTrieString to reject them?

[realwx]

Oh man. I just... wow. I'm an idiot. I did try that days ago, but when I revisited it just now, I had a && where a || should've been. (Plus, I fixed something odd with ExplodeString returning ghost attributes, so the other problem is fixed.)

Thank you so much Sarge, and really sorry I've been bugging you

[Oshizu]

Hello i've got a small problem with plugin. Halloween Attributes doesn't seem to work even with 'tf_forced_holiday'

[MasterOfTheXP]

Valve intentionally broke those, I believe. (What with Halloween being over, and all...still though )

[CoolJosh3k]

Ya, Valve broke both Halloween and MvM stuff just for the creative community

Should be a work around in the future, somehow.

[thesupremecommander]

Can a native to get all of the attributes on an entity be implemented (as opposed to having to blindly guess whether an entity even has an attribute)?