From: "Shot (Piotr Szotkowski)" <shot@...>
Date: 2008-02-20T16:01:35+09:00
Subject: Re: Object#freeze as a basis for caching of method results?

--dycUMCmLTP6kl/eW
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

Ezra Zygmuntowicz:

>	The usual idiom for caching heavy results is something like this:

> def foo
>    @foo ||=3D  some_long_calculation
> end

Thanks. I=E2=80=99m aware of the ||=3D construct, but it doesn=E2=80=99t re=
ally answer the
bigger issue of how to handle cache freshness (see my original email).

Basically, I was wondering whether freezing can be (sanely) used to make
sure we don=E2=80=99t have =E2=80=98stale=E2=80=99 cache (in a bit of a rev=
erse/radical way =E2=80=93 by
freezing the object when we do *any* caching).

Alternatively, does the approach of invalidating a cache on
object-changing method calls make sense? If so, is there a way
to get all the methods that change an object (say, a Set) =E2=80=93 i.e.,
the ones that would trigger the =E2=80=98TypeError: can't modify frozen X=
=E2=80=99
exception?

-- Shot
--=20
Transitiveness works a bit differently in love. "I love A,
and A loves B, which means I want to make a light snack
out of B's internal organs."             -- Alistair Young

--dycUMCmLTP6kl/eW
Content-Type: application/pgp-signature
Content-Disposition: inline

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

iD8DBQFHu8dji/mCfdEo8UoRAh91AJ0X7R6q2sq1xGpNE4gqvyD5TevcrACgrrll
0dR12l2tkBdxWc7C/6WmIHc=
=0CjU
-----END PGP SIGNATURE-----

--dycUMCmLTP6kl/eW--