Re: estimating proper object cache size

From: Marcin Skladaniec (marci..sh.com.au)
Date: Sat Jul 05 2008 - 06:54:04 EDT

  • Next message: Chad Smith: "Re: help: Why is there validation w/ a delete action"

    On 05/07/2008, at 8:51 PM, Andrus Adamchik wrote:

    >
    > On Jul 5, 2008, at 8:20 AM, Marcin Skladaniec wrote:
    >
    >> During my tests I was fetching 1k, 10k or 100k records using ROP.
    >> Select queries with no paging, no query caching. It was always that
    >> the first query took a while longer (about 20%),
    >
    > Probably this was due to Cayenne warmup (class loading delays and
    > such). Usually in performance tests the results of the first pass
    > should be thrown away as they are not representative.
    >
    >> but as mentioned before the object cache size setting did not
    >> influence it.
    >
    > Yes it won't. SelectQuery always refreshes objects in the cache,
    > unless you call 'setRefreshingObjects(true)' (I am actually thinking
    > of deprecating this method).
    >
    > Shared object cache (actually this is DataRow cache) as opposed to
    > query list cache, is intended for relationship resolving, and
    > creating updates. So when you run a straight query, generally it has
    > no effect.

    ah, this makes sense...
    Thank you
    Marcin





    This archive was generated by hypermail 2.0.0 : Sat Jul 05 2008 - 07:12:02 EDT