ContinuousQueryCache and non-idempotent Filters

Question

From the documentation I read on the ContinuousQueryCache, I assume that it works in two stages. First, it runs the distributed query based on the given Filter and caches the results. Second, it registers MapListeners that update the query result whenever an underlying cache INSERT, DELETE or UPDATE occurs. This will work fine with idempotent Filter implementations. But this may fail if underlying Filter is allowed to produce different results over time.
To give a concrete example. I would like my cached values to have a Publish Time. My NamedCache would have all values in it. And my Filter would check whether System.currentTimeMillis()>value.getPublishTimeMillis(). I would expect my ContinuousQueryCache to show the values whose Publish Time has arrived reasonably soon after the Publish Time.
Naturally, there may be no UPDATE events that would trigger the Filter to be re-applied as soon as the Publish Time arrives. So, the values would not appear in my ContinuousQueryCache.
One workaround I can think of is to have a refresh() method in my ContinuousQueryCache that I could run on a scheduled basis from a separate thread. Such refresh() method would rerun the underlying distributed query.
Is there such capability? Or is there a better way people typically handle Publish Time? (I realize that I could put a dummy value in cache which would expire at Publish Time and then expect the cache to retrieve the real value via a CacheLoader. But for many reasons, I do not like this approach.)

Answer 1

We are JDK 1.4.2, so no AtomicReferences here :o) But
I get your point.
It is available for 1.4.2 as a separate JAR file, AFAIK.
Look at http://g.oswego.edu/dl/classes/EDU/oswego/cs/dl/util/concurrent/intro.html
In my opinion, CQC would still be preferable because
it uses events to attempt to keep the query result up
to date. This will allow me to run query *less
frequently* than I would have to in your solution.I don't see why. The events will in no case will mirror the passing of time, therefore you will need to rerun the query to get entries which became of age since the last running of the query.
This is keeping in mind that my Filter may be more
complex than just by Publish Time. There may be many
other additional complex conditions.If they can be indexed, that might even help if it reduces the result set size significantly.
>
So, do you think there's a way to rerun the
underlying query ad hoc?
I don't know of one, if it is not in the Javadoc, as the intentional purpose of the CQC is exactly not having to rerun that query. However this does not mean that there is not one.
Best regards,
Robert