[07:50:42 EST(-0500)] * jayshao (n=jayshao@jayshao.oirt.rutgers.edu) has joined ##uportal
[09:14:43 EST(-0500)] * ac_chan (n=alex@tempoutsidepix.pratt.edu) has joined ##uportal
[09:44:22 EST(-0500)] * jayshao_ (n=jayshao@wg-195-hill012.rutgers.edu) has joined ##uportal
[10:12:04 EST(-0500)] * athena7 (n=athena7@tp-wireless.its.yale.edu) has joined ##uportal
[10:15:03 EST(-0500)] * EricDalquist (n=dalquist@bohemia.doit.wisc.edu) has joined ##uportal
[10:15:39 EST(-0500)] <EricDalquist> hey athena7, any luck this morning with the portal problem?
[10:15:47 EST(-0500)] <athena7> Yes! thank you
[10:16:03 EST(-0500)] <EricDalquist> was it a proxy/timeout issue?
[10:16:05 EST(-0500)] <athena7> we had a cwebproxy channel that was trying to connect to a url that hung indefinitely
[10:16:20 EST(-0500)] <athena7> changing the timeout didn't make a difference though
[10:16:31 EST(-0500)] <EricDalquist> yeah, we've had that problem with Sockets before too
[10:16:57 EST(-0500)] <EricDalquist> when things are stuck waiting on socket.read the threads can't be killed and things seem to wait forever
[10:17:07 EST(-0500)] <EricDalquist> that was actually a big motivation for us writing the proxy portlet
[10:17:26 EST(-0500)] <athena7> ah
[10:17:34 EST(-0500)] <EricDalquist> it uses commons http client and we have socket connect and read timeouts set
[10:17:38 EST(-0500)] <athena7> right
[10:17:56 EST(-0500)] <athena7> it seems like it'd be worth changing cwebproxy to use httpclient
[10:18:00 EST(-0500)] <EricDalquist> probably
[10:18:18 EST(-0500)] <EricDalquist> the control over the HTTP connect/write/read process you get with that library is huge
[10:18:37 EST(-0500)] <EricDalquist> we actually ran into similar problems with our oracle JDBC driver hanging indefinitely in certain db outage instances
[10:18:53 EST(-0500)] <athena7> it's worth mentioning that this also exists in ResourceLoader.getResourceAsURL, since that's what cwebproxy uses
[10:19:02 EST(-0500)] <athena7> yeah, there's a bug in the oracle jdbc driver
[10:19:05 EST(-0500)] <EricDalquist> yup
[10:19:35 EST(-0500)] <EricDalquist> the more I and others I know have worked with the JDK HTTP/URL support the more we realize it is not very well written
[10:19:51 EST(-0500)] <EricDalquist> like did you know that the URL object does DNS lookups when you create it?
[10:20:15 EST(-0500)] <EricDalquist> and new URL("http://foo.com"
; can have different hash codes based on the IP that foo.com resolves to?
[10:20:23 EST(-0500)] <EricDalquist> that is all sorts of fun
[10:21:32 EST(-0500)] <athena7> aieee
[10:21:38 EST(-0500)] <athena7> but yes, httpclient is way better
[10:21:58 EST(-0500)] <EricDalquist> or you could always use the proxy portlet
[10:22:58 EST(-0500)] <athena7> i think eventually we will
[10:23:04 EST(-0500)] <athena7> moving in that direction 
[10:28:57 EST(-0500)] * athena7_ (n=athena7@lumina.its.yale.edu) has joined ##uportal
[10:45:09 EST(-0500)] * SusanBramhall (i=Susan@dhcp128036196209.central.yale.edu) has joined ##uportal
[10:46:19 EST(-0500)] <EricDalquist> well back to rewriting UserInstance
General
Content
Integrations