in my office please.

so you bitch at me all day long and then I realize that the reason there is
the extra invocation code in the entity container is because you duplicate
the code to add 2 features "getCacheSize" and "flushCache".

dude, you're gonna get a spanking.

1- the feature is weak.  We want to enable high end cache policies (beyond
the brain damaged option A and B of the spec) by exposing a
"invalidateEntry(Object primaryKey)" on the cache.  For generic usage the
primary key must be the native one not the cachekey since outside tools or
apps have no real way to know it. This way you can invalidate some of the
cache.  Here you just flush the full cache, much less useful.

2- why do you expose it at the container invocation level, meaning only
threads going in there can have access to that, why not have a pointer to
the cache (make it an mbean and have a direct pointer to it in Registry) so
you can go about invalidating guys one by one. In other words this should be
at the cache level directly

3- say you really want to keep the container stuff, then we would do it in
the interceptor that deals with the cache so we don't have to see the cache
invalidation logic in the container invoker. that is what interceptors are
for.

4- because I truly love you I will also say that I understand why you coded
it there in the fashion you did... it was the simplest hack to get it
working and it makes me giggle like a little girl it is so efficient.
Anyway... it sucks :)

marcf


_______________________________________________
Jboss-development mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/jboss-development

Reply via email to