[13:00:44 CDT(-0500)] <EricDalquist> The String.split method annoys me
[13:00:56 CDT(-0500)] <EricDalquist> every time you call it you're compiling a new Pattern
[13:01:04 CDT(-0500)] <EricDalquist> so it is useful
[13:01:07 CDT(-0500)] <EricDalquist> it saves a little code
[13:01:10 CDT(-0500)] <EricDalquist> but used in hot spots
[13:01:17 CDT(-0500)] <EricDalquist> it really chews CPU
[13:07:52 CDT(-0500)] <athena> yeah
[13:07:57 CDT(-0500)] <athena> that's a good point
[13:08:08 CDT(-0500)] <athena> can you feed it a pre-compiled pattern?
[13:08:14 CDT(-0500)] <EricDalquist> no but you can do:
[13:08:27 CDT(-0500)] <EricDalquist> p = Pattern.compile(...);
[13:08:33 CDT(-0500)] <EricDalquist> p.split(String)
[13:08:40 CDT(-0500)] <EricDalquist> and store p as a static final
[13:08:47 CDT(-0500)] <athena> that makes sense
[13:09:05 CDT(-0500)] <EricDalquist> maybe someday uPortal's code will be clean enough that we can add checkstyle 
[13:11:25 CDT(-0500)] <athena>
[13:42:31 CDT(-0500)] <JoeMoore> Im trying to configure portlet parameters. It wont let me edit them at the beginning, but when I try to edit them with the rich editor I get http://pastebin.com/Q5Y9j5Cw
[14:50:19 CDT(-0500)] <JoeMoore> I guess my pastebin timed out (one hour). Let me know if someone wants to take a look and I'll paste again with a longer timeout on the problem of configuring portlet parameters.
[15:40:06 CDT(-0500)] <drewwills> EricDalquist I'd like to pick your brain for a sec about UP-3161: Misbehaving portlets can cause channel rendering threads to leak and bring the portal to its knees
[15:40:14 CDT(-0500)] <EricDalquist> ok
[15:40:20 CDT(-0500)] <EricDalquist> I have a few spare moments
[15:40:30 CDT(-0500)] <EricDalquist> working franticly toward a uP4.0.4 upgrade in 7 days 
[15:40:46 CDT(-0500)] <drewwills> the way I remember it, things lie as follows: there is logic to detect and log the problem, which is "on" be default...
[15:41:06 CDT(-0500)] <EricDalquist> are we talking 3.2 or 4.0
[15:41:12 CDT(-0500)] <EricDalquist> because they have different mechanisms
[15:41:16 CDT(-0500)] <drewwills> there is logic to prevent the offending portlets from taking more threads, which is "off" by default
[15:41:20 CDT(-0500)] <drewwills> all of this is 3.2
[15:41:23 CDT(-0500)] <EricDalquist> ok
[15:41:47 CDT(-0500)] <drewwills> and (now) for 4.x, there's different realities, and I'm not sure how much is there
[15:42:50 CDT(-0500)] <drewwills> question 1: can we turn on the thread-capping tech? (currently based on positive-integer threadshold for not allocating more)
[15:43:17 CDT(-0500)] <drewwills> question #2: do we need any of that translated and added to 4?
[15:43:19 CDT(-0500)] <EricDalquist> for 3.2, if you're very confident it wont' break existing 3.2 uPortal installs: yes
[15:43:22 CDT(-0500)] <EricDalquist> for 4.0
[15:43:32 CDT(-0500)] <EricDalquist> the portlet execution is done by the org.jasig.portal.utils.threading.QoSThreadPoolExecutorFactoryBean
[15:43:41 CDT(-0500)] <drewwills> pretty sure... OHIO has been running it for 2 quarters or more
[15:43:43 CDT(-0500)] <EricDalquist> sorry that creates the ExecutorService that is used
[15:43:56 CDT(-0500)] <EricDalquist> and the #s will scale well to schools of various sizes?
[15:44:13 CDT(-0500)] <EricDalquist> the Queue for that executor service is org.jasig.portal.portlet.rendering.worker.PortletWorkerExecutionQueue
[15:44:19 CDT(-0500)] <drewwills> i should think... we can pick a high number if we're worried
[15:44:25 CDT(-0500)] <EricDalquist> that would be good
[15:44:42 CDT(-0500)] <EricDalquist> the idea in 4.0 is that the queue that feeds the thread pool is actually a pool of queues
[15:44:46 CDT(-0500)] <EricDalquist> with one queue per fname
[15:44:55 CDT(-0500)] <drewwills> maybe change default thread pool size to 100, max errant to 10?