not necessary, we could still use the current (in memory) caching
mechanism.
We should then keep delta values there, and after interval expiry,
read the
pagecount value from the JCR node for each entry from the im-mem
cache, add
the delta values and write back to JCR.
That would take a lot of writes still... Perhaps it's better to have a
single binary blob.
This does assume that we have locking capabilities on the JCR page.
This way we would also support multiple wiki instances.
Most JCR implementations do support locking (aka "mix:lockable").
Including Priha.
/Janne
