From: Anthony Martinez <pi@...>
Date: 2007-06-11T14:41:02+09:00
Subject: Array#inject to create a hash versus Hash[*array.collect{}.flatten] -- Speed, segfault

--TRYliJ5NKNqkz5bu
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

So, I was tinkering with ways to build a hash out of transforming an
array, knowing the standard/idiomatic

id_list =3D [:test1, :test2, :test3]
id_list.inject({}) { |a,e| a[e]=3De.object_id ; a }

I also decided to try something like this:

Hash[ *id_list.collect  { |e| [e,e.object_id]}.flatten]

and further (attempt to) optimize it via
Hash[ *id_list.collect  { |e| [e,e.object_id]}.flatten!]
and=20
Hash[ *id_list.collect! { |e| [e,e.object_id]}.flatten!]

Running this via Benchmark#bmbm gives pretty interesting, and to me,
unexpected, results (on a 3.2 GHz P4, 1GB of RAM,  FC5 with ruby 1.8.4)

require 'benchmark'
id_list =3D (1..1_000_000).to_a
Benchmark::bmbm do |x|
    x.report("inject")   { id_list.inject({})      { |a,e| a[e] =3D e.objec=
t_id ; a} }
    x.report("non-bang") { Hash[ *id_list.collect  { |e| [e,e.object_id]}.f=
latten] }
    x.report("bang")     { Hash[ *id_list.collect  { |e| [e,e.object_id]}.f=
latten!] }
    x.report("two-bang") { Hash[ *id_list.collect! { |e| [e,e.object_id]}.f=
latten!] }
end

Rehearsal --------------------------------------------
inject    16.083333   0.033333  16.116667 (  9.670747)
non-bang 1657.050000   1.800000 1658.850000 (995.425642)
bang     1593.716667   0.016667 1593.733333 (956.334565)
two-bang 1604.816667   1.350000 1606.166667 (963.803356)
-------------------------------- total: 4874.866667sec

               user     system      total        real
inject     5.183333   0.000000   5.183333 (  3.102379)
non-bang zsh: segmentation fault  ruby

Ow?

Also, I just thought of a similar way to accomplish the same thing:

x.report("zip")      { Hash[ *id_list.zip(id_list.collect {|e| e.object_id}=
)] }

Array#collect! won't work right with this, of course, but it seems to
have equally-bad performance. Is Array#inject just optimized for this,
or something?

-- Pi


--=20
panic ("No CPUs found.  System halted.\n");
        2.4.3 linux/arch/parisc/kernel/setup.c

--TRYliJ5NKNqkz5bu
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQFGbOBsKaiGM/xGKzQRAvjjAJ9FYZabPf88j0GdKq/Y4+Vu1/UfYwCgwPBL
xng4PjGyzyz00YSTL3jhm6o=
=qOdB
-----END PGP SIGNATURE-----

--TRYliJ5NKNqkz5bu--