Arrays and table.insert with named keys

[Sideshow]

Hello

I'm making my own dps addon, just for fun.

I'm trying to add all combatants to an array like this:

players = {
    ["Name1"] = {
        ["class"] = "WARRIOR",
        ["values"] = { 0,4,1 }
    },
    ["Name2"] = {
        ["class"] = "ROGUE",
        ["values"] = { 0,4,1 }
    }
}

The problem is, that table.insert does not let me insert named keys.... it just uses numbers.... so when I do table.insert(players,somedata), the result is:

players = {
    ["Name1"] = {
        ["class"] = "WARRIOR",
        ["values"] = { 0,4,1 }
    },
    ["Name2"] = {
        ["class"] = "ROGUE",
        ["values"] = { 0,4,1 }
    },
    [1] = somedata...
}

I would like something like table.insert(players,mykeyname,somedata), but this obviously does not work ... :<

[Slakah]

players[mykeyname] = somedata

[nightcracker]

A tip mate, try to avoid table.insert AS MUCH AS POSSIBLE. It's really inefficiënt.

If it's a number indexed table this is always faster then table.insert:

local function tinsert(tab, val)
    tab[#tab+1] = val
end

[Slakah]

Originally Posted by nightcracker A tip mate, try to avoid table.insert AS MUCH AS POSSIBLE. It's really inefficiënt.

Meh, it's still pretty insignificant difference, and by converting it to a function as you do, pretty much nulls any performance gain.

http://www.pastey.net/135124-3xsy

results:

table.insert took 	0.02700 sec
local table.insert took 	0.02300 sec
new tableinsert func took 	0.02500 sec
arr[#arr + 1] = took 	0.00800 sec

Another important thing to note is that table.insert and table.remove are required if you want to properly resize the table.

http://www.lua.org/pil/19.2.html

An example:

Lua 5.1.4  Copyright (C) 1994-2008 Lua.org, PUC-Rio
> T = { 1, 2, 3, 4 }
> T[2] = nil
> for Index = 1, #T do print(Index, T[Index]) end
1       1
2       nil
3       3
4       4
> for Index, Value in ipairs(T) do print(Index, Value) end
1       1
> for Key, Value in pairs(T) do print(Key, Value) end
1       1
3       3
4       4
> T = { 1, 2, 3, 4 }
> table.remove(T, 2)
> for Index = 1, #T do print(Index, T[Index]) end
1       1
2       3
3       4
> for Index, Value in ipairs(T) do print(Index, Value) end
1       1
2       3
3       4
> for Key, Value in pairs(T) do print(Key, Value) end
1       1
2       3
3       4

If proper table resizing is not a concern (it's not for most projects), then inserting or removing entries manually is the most efficient solution - as pointed out in Slakah's example.

[Shadowed]

Originally Posted by Slakah Meh, it's still pretty insignificant difference, and by converting it to a function as you do, pretty much nulls any performance gain. http://www.pastey.net/135124-3xsy results: table.insert took 0.02700 sec local table.insert took 0.02300 sec new tableinsert func took 0.02500 sec arr[#arr + 1] = took 0.00800 sec

And odds are, you aren't inserting entries into a table often enough that it's noticeable.

[nightcracker]

Originally Posted by Shadowed And odds are, you aren't inserting entries into a table often enough that it's noticeable.

Not with my library keeping track of GUID's to transform GUID's to UnitID's

[Shadowed]

Originally Posted by nightcracker Not with my library keeping track of GUID's to transform GUID's to UnitID's

No, it's still not worth it. Slakah's test was 100,000 iterations, the difference between a local table.insert and a arr[#arr + 1] is 0.015 seconds. You "save" 0.00000015 seconds per call, nobody is going to notice that kind of optimization and it's pointless to do. Even if you're (for some reason) inserting 100,000 rows into a table, the extra 0.015 seconds really will not be noticed.

[Seerah]

Here's the thread/post I was looking for... http://forums.wowace.com/showthread.php?t=10301

(This is the book: http://www.amazon.com/Preventative-P.../dp/1584502576 )

Originally Posted by Donald Knuth, "Structured Programming with go to Statements" We should forget about small efficiencies, say about 97% of the time: premature optimization is the root of all evil. Yet we should not pass up our opportunities in that critical 3%.

Discussion: http://stackoverflow.com/questions/2...ot-of-all-evil

[Slakah]

Originally Posted by Seerah Here's the thread/post I was looking for... http://forums.wowace.com/showthread.php?t=10301

Really? It doesn't seem to have much to do with whats going on here (unless of course it's going completely over my head).

I think rule #1 of http://www.lua.org/gems/sample.pdf suits more.

[Seerah]

Sorry, I just meant for nightcracker to read all of it. Premature Optimization is mentioned in post #18 of that thread (I forgot to mention that).

[nightcracker]

Originally Posted by Seerah Sorry, I just meant for nightcracker to read all of it. Premature Optimization is mentioned in post #18 of that thread (I forgot to mention that).

Sorry, I just remembered this post(took me a while to find it):
http://www.wowwiki.com/API_tinsert

I only just realized(after reading it again) that they use 5e6, which is 5000000.