uPortal IRC Logs-2012-07-17

[11:12:56 CDT(-0500)] <rlewis> Hi, I am setting up uPortal 4.0.5 trying to use external cas server. put change in local.properties ran "ant deploy-ear" and "ant start" , no luck, login button still says localhost. Help!

[11:15:04 CDT(-0500)] <EricDalquist> so you followed all the steps here? https://wiki.jasig.org/display/UPM40/CAS

[11:17:58 CDT(-0500)] <EricDalquist> athena: https://issues.jasig.org/browse/UP-3517

[11:19:13 CDT(-0500)] <athena> yay thanks!

[11:19:36 CDT(-0500)] <athena> rlewis: set the Cas login url in security.properties to a blank string, redeploy, and restart

[11:36:18 CDT(-0500)] <rlewis> athena: do I make that change to the security.props file under uPortal or the file under tomcat? I am using the quick start pkg.

[11:38:00 CDT(-0500)] <athena> depends on how you want to work with that codebase - if you anticipate rebuilding it at some point you should probably change the source file and redeploy

[11:38:04 CDT(-0500)] <athena> otherwise you can just change the one in tomcat

[11:38:36 CDT(-0500)] <rlewis> ok, thanks

[13:21:40 CDT(-0500)] <drewwills> you there EricDalquist?

[13:21:44 CDT(-0500)] <EricDalquist> YES

[13:21:49 CDT(-0500)] <EricDalquist> oops ... capslock

[13:22:00 CDT(-0500)] <drewwills> I've gotten myself in a bit of a bind...https://gist.github.com/3131072

[13:22:16 CDT(-0500)] <drewwills> remember that restapi invoker thingee?

[13:22:47 CDT(-0500)] <drewwills> at first I didn't think i would need the ability to invoke from a portlet, just from a servlet

[13:22:56 CDT(-0500)] <drewwills> but I was wrong, i do

[13:23:11 CDT(-0500)] <EricDalquist>

[13:23:27 CDT(-0500)] <EricDalquist> that is quite the stack trace

[13:23:37 CDT(-0500)] <drewwills> so i'm going back and enhancing it... adding the extra layer of portlet -> proxyservlet -> xContext include

[13:23:50 CDT(-0500)] <drewwills> yes... a bit ahrd to follow

[13:24:21 CDT(-0500)] <drewwills> i think i may not be wrapping so object sufficiently

[13:24:41 CDT(-0500)] <EricDalquist> heh ... I'm not sure this is even possible

[13:24:46 CDT(-0500)] <EricDalquist> let me read through the stack for a minute

[13:24:54 CDT(-0500)] <drewwills> https://gist.github.com/3131108

[13:25:41 CDT(-0500)] <drewwills> (sorry for system.out noise on that one)

[13:26:31 CDT(-0500)] <drewwills> (proxying servlet https://gist.github.com/3131119)

[13:26:36 CDT(-0500)] <EricDalquist> so this is going from

[13:27:51 CDT(-0500)] <EricDalquist> portal -> pluto container -> crossCTX -> pluto servlet -> your custom portlet filter -> local servlet dispatcher -> crossCTX -> portal rest api

[13:29:36 CDT(-0500)] <drewwills> sounds about right

[13:30:19 CDT(-0500)] <drewwills> trying to get at /api/permissions/etc

[13:30:33 CDT(-0500)] <EricDalquist> uhg

[13:30:38 CDT(-0500)] <EricDalquist> this is really really really gross

[13:31:02 CDT(-0500)] <drewwills> (the more I mess with this, the more curios I get about maybe putting an API jar in shared and registering a request attribute)

[13:31:23 CDT(-0500)] <EricDalquist> heh

[13:31:24 CDT(-0500)] <EricDalquist> yeah

[13:31:29 CDT(-0500)] <EricDalquist> that is probably more sane ...

[13:32:23 CDT(-0500)] <EricDalquist> really the best would be a API jar in shared and providing it as a PortletContext attribute

[13:32:33 CDT(-0500)] <drewwills> bb in a bit...

[13:32:35 CDT(-0500)] <EricDalquist> though that may require a bit more work up front in uPortal to expose that

[13:32:39 CDT(-0500)] <EricDalquist> oh ...

[13:32:47 CDT(-0500)] <EricDalquist> well let me know when you're back

[13:32:49 CDT(-0500)] <EricDalquist> and we can look at this more

[13:33:31 CDT(-0500)] <drewwills> yeah sorry mtg crept up on me

[13:39:08 CDT(-0500)] <drewwills> ok, false alarm... meeting moved

[13:39:21 CDT(-0500)] <drewwills> i have another at noon ()20 min though

[13:39:41 CDT(-0500)] <EricDalquist> so in that stack trace

[13:39:51 CDT(-0500)] <EricDalquist> the last caused by

[13:40:55 CDT(-0500)] <drewwills> yep

[13:41:37 CDT(-0500)] <EricDalquist> it looks like spring is doing something around request attributes and includes

[13:42:08 CDT(-0500)] <drewwills> yeah... messing with them when wrapping, restoring when unwrapping

[13:42:12 CDT(-0500)] <EricDalquist> yeah

[13:42:25 CDT(-0500)] <EricDalquist> I'm assuming it is doing this to deal with spring dispatcher including another spring dispatcher

[13:42:29 CDT(-0500)] <EricDalquist> which is good since that is happening here

[13:42:57 CDT(-0500)] <EricDalquist> the problem is we have logic in uPortal that is supposed to prevent portlet code from messing with certain classes of attributes

[13:43:15 CDT(-0500)] <drewwills> yeah, looks that way

[13:43:27 CDT(-0500)] <EricDalquist> so when spring is doing it's restore it triggers that protection

[13:43:36 CDT(-0500)] <EricDalquist> one thought is to change PortletRequestContextImpl

[13:43:43 CDT(-0500)] <EricDalquist> and have it do a silent ignore

[13:43:50 CDT(-0500)] <EricDalquist> or a logger.debug ignore

[13:43:53 CDT(-0500)] <EricDalquist> instead of throwing

[13:44:09 CDT(-0500)] <EricDalquist> which would probably be reasonable

[13:44:16 CDT(-0500)] <EricDalquist> so go into org.jasig.portal.portlet.container.PortletRequestContextImpl.setAttribute(PortletRequestContextImpl.java:169)

[13:44:23 CDT(-0500)] <EricDalquist> and change the throw

[13:44:26 CDT(-0500)] <EricDalquist> to a logger.debug

[13:44:30 CDT(-0500)] <EricDalquist> and a return

[13:44:47 CDT(-0500)] <EricDalquist> that will have the desired effect of not letting portlets touch those attributes

[13:45:01 CDT(-0500)] <EricDalquist> and let spring's restore process work

[13:47:33 CDT(-0500)] <drewwills> trying that

[13:49:27 CDT(-0500)] <drewwills> i bet a java API (as a PortletContext attr) would cover most use cases if it just had user attributes, groups, permissions

[14:04:47 CDT(-0500)] <athena> that seems plausible

[14:05:00 CDT(-0500)] <EricDalquist> yeah and will be easier if I can get time to finish this spring-portlet contrib work

[14:05:08 CDT(-0500)] <EricDalquist> that would fit well into the new portlet-application level context config

[14:05:15 CDT(-0500)] <EricDalquist> since we would have the PortletContext availble there

[14:09:38 CDT(-0500)] <mmoayyed> Hi everyone. I am experiencing a strange issue. Working on an instance of the email-preview portlet, for some reason the USER_INFO map is completely empty. Any ideas?

[14:10:26 CDT(-0500)] <EricDalquist> you have user attributes configured in portlet.xml?

[14:11:30 CDT(-0500)] <mmoayyed> I have the user attribute entries defined for both password and user.login.id. I have configured clearpass and I do see the log entry that indicates psw is being retrieved. I have also added the 'cachedPasswordUserInfoService' bean into the context

[14:11:46 CDT(-0500)] <mmoayyed> still, isEmpty() returns true

[14:11:51 CDT(-0500)] <EricDalquist> weird

[14:16:17 CDT(-0500)] <mmoayyed> I had no values for attributes, that would have been slightly better. But a completely empty map makes very suspicious. javadocs say the attribute needs to be supported by the current runtime system. Is there any other configuration that I should add?

[14:16:27 CDT(-0500)] <mmoayyed> *if I had...

[14:16:55 CDT(-0500)] <athena> that does seem odd - maybe set a breakpoint in CachedPasswordUserInfoService?

[14:17:07 CDT(-0500)] <athena> make sure it' sgetting called, retrieving successfully, etc.

[14:17:35 CDT(-0500)] <EricDalquist> so the parts that need to match are:

[14:17:35 CDT(-0500)] <EricDalquist> the attribute(s) are needed in portlet.xml

[14:17:35 CDT(-0500)] <EricDalquist> the same attribute(s) need to exist in personDirectory

[14:17:35 CDT(-0500)] <EricDalquist> the exception to the previous point are "special" attribute names for things like the cached password or CAS PGT

[14:27:22 CDT(-0500)] <mmoayyed> I definitely have uid, username and user.login.id defined in person-dir. This particular user does exist inside UP_USER. my person directory config for uPortalJdbcUserSource and uPortalAccountUserSource is identical to the copy of the file on github

[14:29:08 CDT(-0500)] <mmoayyed> just one thing though. I dont have to do anything special for CachedPasswordUserInfoService to kick in right? user info beans are all loaded by type?

[14:29:22 CDT(-0500)] <EricDalquist> I believe so

[14:29:29 CDT(-0500)] <mmoayyed> 'cause the documentation suggested that it needed to be added to somewhere else, but I suspect that's old

[14:29:33 CDT(-0500)] <EricDalquist> @Autowired

[14:29:33 CDT(-0500)] <EricDalquist> public void setUserInfoServices(Set<UserInfoService> userInfoServices) {

[14:29:37 CDT(-0500)] <EricDalquist> looks like it

[14:41:55 CDT(-0500)] <mmoayyed> does UP has a logging aspect, such that if I turn up the TRACE level, I'd be able to review method arguments on entering and exiting methods?

[14:42:56 CDT(-0500)] <EricDalquist> no

[14:43:03 CDT(-0500)] <EricDalquist> would be a nice thing to add

[14:43:05 CDT(-0500)] <mmoayyed> I am working on a remote linux server, and it's hooked up in a way that makes the build very difficult locally. If it's possibly, I'd rather debug the issue through logs

[14:43:24 CDT(-0500)] <mmoayyed> This is what CAS does:

[14:43:25 CDT(-0500)] <mmoayyed> http://grepcode.com/file/repo1.maven.org/maven2/org.jasig.cas/cas-server-core/3.4.11/org/jasig/cas/aspect/LogAspect.java#LogAspect

[14:44:01 CDT(-0500)] <mmoayyed> should be able to use the same exact code, with very little changes

[14:44:42 CDT(-0500)] <EricDalquist> yeah

[14:50:09 CDT(-0500)] <EricDalquist> athena: what was that JS templating library you were looking at?

[15:00:56 CDT(-0500)] <athena> backbone.js

[15:01:00 CDT(-0500)] <athena> builds on jQuery and underscore

[15:01:20 CDT(-0500)] <athena> been getting a lot of use on other projects at unicon - seems to have a pretty good reputation

[15:01:26 CDT(-0500)] <athena> playing with it right now, actually

[15:03:42 CDT(-0500)] <athena> has templating, events, some nice stuff

[15:04:09 CDT(-0500)] <EricDalquist> cool

[15:04:14 CDT(-0500)] <EricDalquist> just passed it on to nick

[15:04:29 CDT(-0500)] <EricDalquist> he was complaining about mockup up a week view of a calendar in html

[15:04:33 CDT(-0500)] <athena> i'm certainly not an expert, but people like matt have been doing a lot of work with it

[15:04:33 CDT(-0500)] <EricDalquist>

[15:04:58 CDT(-0500)] <athena> trying it out on the news portlet right now

[15:05:20 CDT(-0500)] <EricDalquist> cool