cy6erbr4in/arrets

A silly non-performant array implementation on top of ets

arrets

A silly non-performant array implementation on top of ets.

For "benchmarks" scroll all the way down.

Why?

I "needed" an array/stack with Javascript-like semantics: ability to push, pop, randomly access and update elements, slice and splice.

What?

A 2-D dynamically growing array implemented on top of a set table in ets.

Caveats?

Unless a Row parameter is specified, all functions by default operate on Row 0. In other words: if you do not specify a Row, your array will behave as an array with one row only.

If something crashes, your data is gone, as nothing is done to protect the ets table (yet?).

If several processes access the dictionary, strange things will happen (once again: ets).

API

Note: all API calls that accept From, Count also accept a negative From. If From < 0, starting element will be counted from the end of the array.

Lifecycle

• new(): get a random handle to a new array
• new(Handle::atom()): get a new array with your own handle
• teardown(Handle): delete the array

Info

• length(Handle): length
• length(Handle, Row): length of row

Data manipulation. Destructive

• push(Handle, Item): push. Equivalent to insert_at(Handle, 0, Item)
• push(Handle, Row, Item): push to row. Equivalent to insert_at(Handle, Row, 0, Item)
• pop(Handle): pop. Equivalent to at(Handle, 0)
• pop(Handle, Row): pop from row. Equivalent to at(Handle, Row, 0)
• pop_n(Handle, N): pop N items. Equivalent to splice(Handle, 0, N)
• pop_n(Handle, Row, N): pop N items from Row. Equivalent to splice(Handle, Row, 0, N)
• slice(Handle, From, Count): Only leave the items between From and From + Count in the stack
• slice(Handle, Row, From, Count): Same for a row
• splice(Handle, From, Count): Remove and return items between From and From + Count in the stack
• splice(Handle, Row, From, Count): Same for a row
• insert_at(Handle, Index, Item): Insert Item at Index
• insert_at(Handle, Row, Index, Item): Insert Item at Index in Row
• update_at(Handle, Index, Item): Update Item at Index
• update_at(Handle, Row, Index, Item): Update Item at Index in Row
• empty(Handle): Empty the entire array (remove all items from all rows)

Data manipulation. Non-destructive

• range(Handle, From, Count): Return items between From and From + Count. Leave array intact
• range(Handle, Row, From, Count): Same for a row
• at(Handle, Index): Alias for nth
• at(Handle, Row, Index): Alias for nth
• nth(Handle, Index): Return Item at Index. Leave array intact.
• nth(Handle, Row, Index): Return Item at Index at Row. Leave array intact.

"Benchmarks"

Non-scientific ballpark figures using a Macbook Pro and statistics/1.

> F = fun(Fun) ->
        statistics(runtime),
        statistics(wall_clock),
        Fun(),
        {_, Time1} = statistics(runtime),
        {_, Time2} = statistics(wall_clock),
        U1 = Time1,
        U2 = Time2,
        io:format("Code time=~p (~p) milliseconds~n", [U1,U2])
     end.
> L = lists:seq(0, 10000).

> %% Create new
> F(fun() -> arrets:new(test) end).
Code time=0 (0) milliseconds

> %% Push 10 000 items
> F(fun() -> lists:foreach(fun(E) -> arrets:push(test, E) end, L) end).
Code time=2530 (2564) milliseconds

> %% Pop those back. It's slower due rather inneficient code governing this
> %% But don't fret, it's not too bad. See times for splice/range/at below
> F(fun() -> lists:foreach(fun(_) -> arrets:pop(test) end, L) end).
Code time=8430 (8511) milliseconds

> %% Repopulate
> lists:foreach(fun(E) -> arrets:push(test, E) end, L).

> %% Get back a random range
> F(fun() -> arrets:range(test, 1337, 1337) end).
Code time=0 (3) milliseconds

> %% Remove a random range
> F(fun() -> arrets:splice(test, 1337, 1337) end).
Code time=10 (10) milliseconds

> %% Leave a random range
> F(fun() -> arrets:slice(test, 1337, 1337) end).
Code time=0 (5) milliseconds

> %% Empty and repopulate
> arrets:empty(test), lists:foreach(fun(E) -> arrets:push(test, E) end, L).

> %% Access a random element
> F(fun() -> arrets:at(test, 1337) end).
Code time=0 (1) milliseconds

> %% Pop n elements
> F(fun() -> arrets:pop_n(test, 1337) end).
Code time=10 (8) milliseconds

> %% Empty array
> F(fun() -> arrets:empty(test) end).
Code time=0 (1) milliseconds

> %% Destroy array
> F(fun() -> arrets:teardown(test) end).
Code time=0 (0) milliseconds