[erlang-questions] hmac sha256 erlang

[Steve Davis]

I had RTFM before making that suggestion! However, with my interest 
mildly piqued by this, I checked a little further, adding a "manual 
append" comparison just for good measure:

test(Num) ->
	T0 = funtime(fun lcopy/0, Num),
	T1 = funtime(fun bcopy/0, Num),
	T2 = funtime(fun ccopy/0, Num),
	{T0, T1, T2}.
lcopy() ->
	list_to_binary(lists:duplicate(128, 16#36)).
bcopy() ->
	binary:copy(<<16#36>>, 128).
ccopy() ->
	ccopy(<<16#36>>, 128).
ccopy(Bin, Count) when Count > 0 ->
	ccopy(<<Bin/binary, 16#36>>, Count - 1);
ccopy(Bin, 0) ->
	Bin.

funtime(F, Repeats) ->
	{A1, B1, C1} = erlang:now(),
     exec_fun(F, Repeats),
	{A2, B2, C2} = erlang:now(),
	((A2-A1) * 1000000 + (B2-B1)) * 1000 + (C2-C1) / 1000.

exec_fun(_, 0) ->
	ok;
exec_fun(F, Count) ->
	F(),
	exec_fun(F, Count - 1).

1> copier:test(10000).
{265.0,62.999,280.999}

Not really any kind of formal test but it does seem to confirm the 
choice/advice.

/s

On 5/8/2011 4:49 PM, Bob Ippolito wrote:
 > binary:copy/2 is more useful for this purpose.
 >
 > On Sunday, May 8, 2011, Robert Virding
 > <robert.virding@REDACTED>  wrote:
 >> In the docs for binary:copy it says:
 >>
 >> "This function will always create a new binary, even if N = 1. By 
using copy/1 on a binary referencing a larger binary, one might free up 
the larger binary for garbage collection."
 >>
 >> Although there is a caveat saying that it might be actually useful 
in special cases.
 >>
 >> Robert
 >>
 >> ----- "Steve Davis"<steven.charles.davis@REDACTED>  wrote:
 >>
 >>> Maybe not but in this case it seems marginally faster and I can't see
 >>> any other impact?
 >>>
 >>> I'm definitely open to correction/enlightenment, but I don't see why
 >>> not here!
 >>>
 >>> /s
 >>>
 >>> On May 6, 4:43 pm, Robert Virding<robert.vird...@REDACTED
 >>> solutions.com>  wrote:
 >>>> Why would you want to do binary:copy here? It is not what it is
 >>> intended for.
 >>>>
 >>>> Robert
 >>>>
 >>>> ----- "Steve Davis"<steven.charles.da...@REDACTED>  wrote:
 >>>>
 >>>>> ...though now I would favor binary:copy instead of
 >>>>> list_to_binary(lists:duplicate()), but that depends on what OTP
 >>>>> version you have...
 >>>>
 >>>>> /s
 >>
 >