From: "Shot (Piotr Szotkowski)" <shot@...>
Date: 2008-02-22T01:01:44+09:00
Subject: Re: Object#freeze as a basis for caching of method results?

--jUyrq1hnz4WvHzxT
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

// Thanks a lot for that delegate tutorial in the other post =E2=80=93
// I finally understood what=E2=80=99s delegation for and how to do it! :)

Robert Klemme:

> 2008/2/21, Shot (Piotr Szotkowski) <shot@hot.pl>:

>> +  @@cache =3D Hash.new { |hash, key| hash[key] =3D {} }
>>    def beta_f
>> -    outputs.to_blanket
>> +    @@cache[Marshal.dump self][:beta_f] ||=3D outputs.to_blanket
>>    end

> Funny that you mention it: I had thought of copying the key via
> Marshal#dump and #load. Using the marshaled string is a nice idea!

> You can make this a tad more efficient by doing

> @@cache[Marshal.dump(self).freeze][:beta_f] ||=3D outputs.to_blanket

> because there is an optimization in Hash that copies unfrozen Strings
> that are used as Hash keys in order to avoid aliasing effects.

I ran some very rough and unscientific tests
with ruby-prof and the results are as follows:

uncached:            44.42s
Marshal.dump:         0.60s
Marshal.dump.freeze:  0.59s
freezing the object:  0.43s

This is for 96 calls, though, so I assume freezing the dump will have
a bigger impact on more-often-called methods (but then the gain from
simply caching the value =E2=80=98as is=E2=80=99 and freezing the object in=
 question
will be bigger as well, as marshalling takes its toll on every call).

Running the whole application in Ruby 1.9 is roughtly
three times faster than in 1.8, but I can=E2=80=99t=C2=B9 profile it. :(

=C2=B9 http://groups.google.com/group/ruby-talk-google/browse_thread/thread=
/b3d26892c642779c

> Now, whether you use the String or demarshal probably mainly depends
> on memory usage. If the String is short enough that approach is
> certainly preferable because it incurs less processing overhead
> (demarshaling).

Ok, I=E2=80=99m totally lost here. I don=E2=80=99t see any demarshalling ha=
ppening,
just marshalling (and using that as a key)=E2=80=A6 What am I missing?

If keeping the marshalled objects in the memory turns out to be
a problem, I can try trading it for speed (and a tiny bit of confidence)
by MD5-hashing them, but I assume MD5 is relatively slow, so it=E2=80=99d b=
e an
exchange at a rather bad rate.

>> So the Ether Bunny goes hippety hopity down the garden path, waylaying
>> innocent fieldmice and anesthetising them, so he can sell their teeth
>> to the Tooth Fairy to support his milk-and-cookies habit.

> What kind of dope are *you* smoking? :-)

I=E2=80=99m a PHP programmer=C2=B2 by day.

=C2=B2 http://civicrm.org/

-- Shot (seriously, though, it=E2=80=99s a sig of unknown origin from Stewa=
rt
Stremler=E2=80=99s collection: http://www-rohan.sdsu.edu/~stremler/sigs/sig=
s.html)
--=20
The most secure computer in the world is one not connected to
the Internet. That's why I recommend Telstra ADSL.   -- FreeFrag

--jUyrq1hnz4WvHzxT
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2.2 (GNU/Linux)

iD8DBQFHvZuZi/mCfdEo8UoRArfTAJ4x10p7/XbIdK/iOovNToaY5DZkrgCdHWDd
idElaEIR3n2z1GpcJ5q34pU=
=SxbB
-----END PGP SIGNATURE-----

--jUyrq1hnz4WvHzxT--