[WildCard65]

It could be that SMObject uses the STL where StringMap uses sm_stringhashmap which uses ke::HashMap

[Neuro Toxin]

Benchmarking object creation, GetInt, SetInt into one benchmark time doesnt accuratly compair this extension to Dynamic.

Dynamic is a dymically typed collection with automatic type conversion and would spend most of its extra time creating objects.

Also note Dynamic has offsets which speedup the getters and setters dramaticly.

Using the class builder gives an example of this.

Edit: There is a dynamic-test plugin in the Dynamic package. Run this to get a proper benchmark with a high percision timer broken down into internal functions.

[headline]

Originally Posted by WildCard65 It could be that SMObject uses the STL where StringMap uses sm_stringhashmap which uses ke::HashMap

I think it was just that SMObjects was doing too many lookups in the older version. There were always at least too full lookup operations taking place for every object write.
Now I've limited it to just one lookup

This looks much better. We're getting close, but StringMap is still slightly faster.

View both profilers here

[SMOBJECTS] Object creation took 0.002454 seconds
[SMOBJECTS] Object writing took 0.013774 seconds
[SMOBJECTS] Object reading took 0.004659 seconds
[SMOBJECTS] Object deletion took 0.007978 seconds
[SM] Plugin SMObjects-Profiler.smx reloaded successfully.

[StringMap] Object creation took 0.004738 seconds
[StringMap] Object writing took 0.011586 seconds
[StringMap] Object reading took 0.003434 seconds
[StringMap] Object deletion took 0.005322 seconds
[SM] Loaded plugin StringMap-Profiler.smx successfully.

[DYNAMIC] Object creation took 0.022605 seconds
[DYNAMIC] Object writing took 0.189613 seconds
[DYNAMIC] Object reading took 0.031048 seconds
[DYNAMIC] Object deletion took 0.038578 seconds
[SM] Plugin Dynamic-Profiler.smx reloaded successfully.

[KeyValues] Object creation took 0.006072 seconds
[KeyValues] Object writing took 0.016962 seconds
[KeyValues] Object reading took 0.006804 seconds
[KeyValues] Object deletion took 0.005861 seconds
[SM] Loaded plugin KeyValues-Profiler.smx successfully.

Originally Posted by Neuro Toxin Benchmarking object creation, GetInt, SetInt into one benchmark time doesnt accuratly compair this extension to Dynamic. Dynamic is a dymically typed collection with automatic type conversion and would spend most of its extra time creating objects. Also note Dynamic has offsets which speedup the getters and setters dramaticly.

If you really want to you're more than welcome to write a dynamic profiler based off of these operations, but I'm more concerned with StringMaps tbh

EDIT: Fuck it I did it anyway. Scroll up.

[shavit]

Is it possible to iterate an object for its members with this extension?
Otherwise, this is very neat and I'm glad someone has done it!

Do you have any plans to add de/serialization of keyvalue files from/into objects?

[headline]

Originally Posted by shavit Is it possible to iterate an object for its members with this extension? Otherwise, this is very neat and I'm glad someone has done it! Do you have any plans to add de/serialization of keyvalue files from/into objects?

These ideas are possible. I toyed with the idea of allowing iteration, but I am currently unsure exactly how I'd like for it to be implemented. Do you have any ideas?

[Neuro Toxin]

Originally Posted by Headline These ideas are possible. I toyed with the idea of allowing iteration, but I am currently unsure exactly how I'd like for it to be implemented. Do you have any ideas?

This might help.

https://github.com/ntoxin66/Dynamic/...rating-Members

[headline]

Originally Posted by Neuro Toxin This might help. https://github.com/ntoxin66/Dynamic/...rating-Members

This wraps std::map, and requires an actual iterator. Can't just index members like that.

Idk if I'd want to implement an iterator type, either. :/

[shavit]

Originally Posted by Headline These ideas are possible. I toyed with the idea of allowing iteration, but I am currently unsure exactly how I'd like for it to be implemented. Do you have any ideas?

Not really.. I'm not very experienced with C++.

But.. I guess you could assign an ID for every object member, add a .MemberCount (or a similar name) member and then allow members to be accessed by member IDs. That'll also need .MemberType that'll return things like Type_Integer, Type_Float etc.

[headline]

Originally Posted by shavit Not really.. I'm not very experienced with C++. But.. I guess you could assign an ID for every object member, add a .MemberCount (or a similar name) member and then allow members to be accessed by member IDs. That'll also need .MemberType that'll return things like Type_Integer, Type_Float etc.

I just finished it (yay)

Here's what it'll look like.

Make sure you're using 1.0.1-git87 from the downloads page and don't forget to update your includes. I'll write up more docs later, but usage is pretty self explanatory. Let me know if you have issues

PHP Code:

// this code prints every single key and it's value, actual usage depends on what you wanna do
for (int i = 0; i < obj.MemberCount; i++)
{
    char key[32];
    obj.GetMemberKey(i, key, sizeof(key));
    
    ObjectType type = obj.GetMemberType(i);
    
    switch (type)
    {
        case Type_Integer:
            PrintToServer("\"%s\": %i", key, obj.GetMemberInt(i));
        case Type_Boolean:
            PrintToServer("\"%s\": %i", key, obj.GetMemberBool(i));
        case Type_Float:
            PrintToServer("\"%s\": %.2f", key, obj.GetMemberFloat(i));
        case Type_String:
        {
            char value[128];
            obj.GetMemberString(i, value, sizeof(value));
            PrintToServer("\"%s\": %s", key, value);
        }
        case Type_CellArray:
        {
            int size = obj.GetMemberArraySize(i);
            any[] array = new any[size];
            obj.GetMemberArray(i, array, size);
            PrintToServer("\"%s\": %s", key, array);
        }

    }
} 


[Neuro Toxin]

Now that you can iterate, you have the bones for a write key values function.