Is there anywhere we can make a note of this so when we do go to 1.5 it gets put in the code?

and possibly on the SolrPerformance wiki page so that 1.5 users can get the performance boost I'm expecting this kind of thing would give.

are there any disadvantages of using ConcurrentHashMap that a java n00b like me should be aware of? anyone else using this in production?

and also any other things like this in the code that could be "instant wins" ?

Fuad Efendi wrote:
:)
Have you tried using a ConcurrentHashMap?
- Of course.
And having code like
Object get(Object key) {
        synchronized (map) {
                ... <atomic>.incrementAndGet()
                ...
        }
forces me to do more "sanity checks"... Now I understand why SOLR uses
single overloaded CPU in 4-CPU system when faceting is enabled.

Even get() method is synchronized...

I need to learn how to use plugin jars, don't want to alter source...
Thanks!


On 1-Apr-08, at 6:58 PM, Fuad Efendi wrote:
Can we have anything better? I can't use 4-CPUs :(
Thanks!
You can have anything your heart desires... Solr is open-source :)

Have you tried using a ConcurrentHashMap?

-Mike





Reply via email to