Re: [Webmacro-devel] Caching issue - "how to"?

[Lane S. <la...@op...>]

endre:

Your point is not relevant or correct. Obviously, you have not bothered=20
to notice that the open doors cache mechanism searches all generations=20
not just the current generation to retrieve an object. The caching=20
system provided is not something I pulled out of my wazoo. Having=20
designed and implemented main frame code for Virtual Memory system=20
kernels in both tightly and loosely coupled bus architectures and over=20
the wire, I would not implement a memory model that did not provide a=20
consistent image of the "truth".

Look, endre, if you think you can "prove" that the generational caching=20
system does not publish a consistent image over time, please do so=20
offline so I can look at your proof. Make sure you examine the code. The=20
WM discussion list is not the place for that kind of a discussion.

-Lane



Endre St=F8lsvik wrote:

>On Thu, 3 Apr 2003, Lane Sharman wrote:
>
>| You should definitely take a look at org.opendoors.cache.* and
>| org.opendoors.cache.immutable.* which is a generational cache system.
>| Lots of good features.
>
>.. and only works on thightly coupled multiprocessor systems (which
>excludes a lot of the "big iron" machines out there). This because you
>need cache-coherency for the "generational" thing to work (it doesn't
>synchronize properly).
>
>I'd vote for using subsystems that actually _work as intended_ on every
>system, and then you might configure your particular setup to use
>architecture-dependent code.
>
>Isn't there some cache-wrapping code in WebMacro that enables you to
>select caching via config?
>
> =20
>

--=20
Lane Sharman
Learn About Conga, All Java GUI Builder: http://opendoors.com/conga
Java Software Portal: http://opendoors.com