uPortal IRC Logs-2012-06-29

Owned by Apereo Infrastructure
Last updated: Jun 29, 2012
7 min read

[09:59:33 CDT(-0500)] <athena> EricDalquist: do you know what the best way to check this new codebase into a new branch of the web proxy portlet is?

[09:59:46 CDT(-0500)] <EricDalquist> so you just have a bunch of code?

[09:59:51 CDT(-0500)] <athena> yeah

[10:00:04 CDT(-0500)] <athena> can i just run git init, add the existing repo as a remote and check it into a branch that way?

[10:00:14 CDT(-0500)] <athena> or do i need to clone the project, delete everything, and copy this stuff over?

[10:00:17 CDT(-0500)] <EricDalquist> git probably has a fancy way of doing it ...

[10:00:25 CDT(-0500)] <athena> seems like it should

[10:00:26 CDT(-0500)] <athena> (tongue)

[10:00:36 CDT(-0500)] <athena> it has fanciness for everything

[10:01:09 CDT(-0500)] <EricDalquist> so a non cool git way would be:

[10:01:09 CDT(-0500)] <EricDalquist> create branch in existing project

[10:01:09 CDT(-0500)] <EricDalquist> git rm -r *

[10:01:09 CDT(-0500)] <EricDalquist> cp -r your new code into the branch

[10:01:09 CDT(-0500)] <EricDalquist> git add .

[10:01:10 CDT(-0500)] <EricDalquist> git commit

[10:01:20 CDT(-0500)] <EricDalquist> if you want I can look into a fancy git way

[10:02:06 CDT(-0500)] <athena> yeah i'd considered doing that

[10:02:14 CDT(-0500)] <athena> are we fine with just doing that?

[10:02:22 CDT(-0500)] <athena> i'm not sure if it really matters since this new codebase is so different

[10:02:28 CDT(-0500)] <EricDalquist> yup

[10:02:30 CDT(-0500)] <athena> but didn't want to just do it in case we had concerns

[10:02:32 CDT(-0500)] <EricDalquist> I'm fine with doing that

[10:02:36 CDT(-0500)] <athena> ok, lets just keep things easy then

[10:02:45 CDT(-0500)] <athena> i am ok with being uncool.

[10:02:56 CDT(-0500)] <athena> (tongue)

[10:03:03 CDT(-0500)] <EricDalquist> oh wait

[10:03:21 CDT(-0500)] <EricDalquist> git checkout --orphan NEWBRANCH

[10:03:28 CDT(-0500)] <athena> magic!

[10:03:40 CDT(-0500)] <EricDalquist> then do

[10:03:40 CDT(-0500)] <EricDalquist> git rm -rf .

[10:03:47 CDT(-0500)] <EricDalquist> and copy in the new files

[10:03:59 CDT(-0500)] <athena> cool

[10:04:00 CDT(-0500)] <EricDalquist> you won't get the big delete commit then

[10:04:04 CDT(-0500)] <athena> oh, nice

[10:04:07 CDT(-0500)] <EricDalquist> it is just a branch with a different root parent

[10:04:16 CDT(-0500)] <EricDalquist> since git repos can have multiple "first commits"

[10:04:17 CDT(-0500)] <athena> nice

[10:04:21 CDT(-0500)] <EricDalquist> yay for directed graphs!

[10:04:22 CDT(-0500)] <athena> that sounds perfect

[10:04:22 CDT(-0500)] <EricDalquist> (tongue)

[10:04:47 CDT(-0500)] <athena> so i've got working XSLT and JSON portlets in addition to the web proxy

[10:05:11 CDT(-0500)] <athena> those aren't very sophisticated - no parameters, no header replay, no authentication

[10:05:14 CDT(-0500)] <EricDalquist> oh nice

[10:05:18 CDT(-0500)] <athena> and the web proxy still needs a whole lot of work

[10:05:21 CDT(-0500)] <EricDalquist> using jsoup?

[10:05:24 CDT(-0500)] <EricDalquist> right

[10:05:24 CDT(-0500)] <athena> yep

[10:05:33 CDT(-0500)] <EricDalquist> awesome!

[10:05:45 CDT(-0500)] <athena> but hopefully that'll be a good starting point for us to collaborate

[10:05:52 CDT(-0500)] <EricDalquist> yeah

[10:07:00 CDT(-0500)] <athena> the XSLT and JSON pieces are probably usable for trivial cases now

[10:07:04 CDT(-0500)] <athena> web proxy is definitely not usable yet

[10:07:04 CDT(-0500)] <EricDalquist> cool

[10:07:07 CDT(-0500)] <EricDalquist> yeah

[10:07:23 CDT(-0500)] <athena> think that'll take some work before we've got all the critical pieces together

[10:07:33 CDT(-0500)] <athena> but there's clipping and some half-done URL rewriting

[10:07:41 CDT(-0500)] <EricDalquist> so cool

[10:08:00 CDT(-0500)] <athena> oh, also made sure all three support pulling content from either a URL or a classpath location

[10:08:11 CDT(-0500)] <EricDalquist> great

[10:08:26 CDT(-0500)] <EricDalquist> I'm assuming jsoup lets us use httpclient under the hood

[10:08:26 CDT(-0500)] <athena> i know that's one thing CWebProxy had but the portlet didn't

[10:08:30 CDT(-0500)] <athena> yep

[10:08:32 CDT(-0500)] <EricDalquist> yeah

[10:08:37 CDT(-0500)] <EricDalquist> I kinda have a love/hate with that feature

[10:08:43 CDT(-0500)] <athena> and i split that out entirely - have jsoup using the call that just takes an inpu stream

[10:08:46 CDT(-0500)] <athena> yeah, agreed

[10:08:47 CDT(-0500)] <EricDalquist> we need a way to turn that off at the portlet.xml or deploy time level

[10:09:08 CDT(-0500)] <athena> fair enough

[10:09:27 CDT(-0500)] <EricDalquist> like otherwise we couldn't use it on the live build site (whenever I get that working again)

[10:09:29 CDT(-0500)] <athena> at least this version is just classpath and not a file path

[10:09:38 CDT(-0500)] <EricDalquist> ah true

[10:09:43 CDT(-0500)] <EricDalquist> so that is much better

[10:09:47 CDT(-0500)] <athena> still don't want to have it available in all installs, but seems like that limits things somewhat

[10:09:50 CDT(-0500)] <EricDalquist> yup

[10:10:03 CDT(-0500)] <athena> i just don't feel comfortable having something that can read the entire filesystem, even if it's admin-only

[10:10:07 CDT(-0500)] <athena> too much potential for oopses

[10:10:11 CDT(-0500)] <EricDalquist> yup

[10:10:16 CDT(-0500)] <EricDalquist> steal all the config on the machine (tongue)

[10:10:19 CDT(-0500)] <athena> yes

[10:10:24 CDT(-0500)] <athena> ooh db passwords!

[10:10:47 CDT(-0500)] <athena> master passwords to banner and exchange!

[10:10:50 CDT(-0500)] <EricDalquist> yup

[10:10:51 CDT(-0500)] <athena> bad.

[10:10:56 CDT(-0500)] <EricDalquist> well

[10:11:08 CDT(-0500)] <EricDalquist> it would be just the proxy portlet's classpaTH

[10:11:10 CDT(-0500)] <EricDalquist> and shared/lib

[10:11:13 CDT(-0500)] <EricDalquist> and tomcat's classpath

[10:11:15 CDT(-0500)] <EricDalquist> still

[10:11:18 CDT(-0500)] <EricDalquist> not good

[10:13:10 CDT(-0500)] <athena> i mean if it could read the whole file system

[10:13:18 CDT(-0500)] <EricDalquist> right

[10:13:21 CDT(-0500)] <EricDalquist> but even so

[10:13:32 CDT(-0500)] <EricDalquist> in the classpath it can read the web proxy portlet's db config

[10:13:39 CDT(-0500)] <athena> yeah

[10:13:50 CDT(-0500)] <EricDalquist> well

[10:13:51 CDT(-0500)] <athena> true

[10:13:56 CDT(-0500)] <EricDalquist> we could put that in WEB-INF

[10:14:01 CDT(-0500)] <EricDalquist> instead of resources

[10:14:06 CDT(-0500)] <athena> good plan

[10:14:11 CDT(-0500)] <athena> at least there isn't as much interesting stuff in the portal db as there is other places

[10:14:26 CDT(-0500)] <EricDalquist> yeah

[10:14:26 CDT(-0500)] <athena> one advantage of uportal not copying in stuff locally

[10:24:55 CDT(-0500)] <athena> https://github.com/Jasig/WebproxyPortlet/tree/v2

[10:24:58 CDT(-0500)] <athena> https://github.com/Jasig/WebproxyPortlet/tree/v2

[10:25:02 CDT(-0500)] <athena> oops.

[10:25:51 CDT(-0500)] <EricDalquist> yay!

[10:29:13 CDT(-0500)] <EricDalquist> this is an actual method api in the oracle jdbc driver:

[10:29:14 CDT(-0500)] <EricDalquist> doOALL(boolean, boolean, boolean, boolean, boolean, byte, int, byte[], int, Accessor[], int, Accessor[], int, byte[], char[], short[], int, DBConversion, byte[], InputStream[][], byte[][][], OracleTypeADT[][], OracleStatement, byte[], char[], short[], T4CTTIoac[], int[], int[], int[], NTFDCNRegistration)

[10:29:20 CDT(-0500)] <athena> one of the current problems is that the filter interface i wrote can't figure out what the requested URL was (https://github.com/Jasig/WebproxyPortlet/blob/v2/src/main/java/org/jasig/portlet/proxy/service/web/URLRewritingFilter.java)

[10:29:28 CDT(-0500)] <athena> so a base url is hard-coded at the moment

[10:29:30 CDT(-0500)] <athena> need to fix that

[10:29:38 CDT(-0500)] <athena> aaaaaaaaah my eyes, they burn! wow.

[10:29:42 CDT(-0500)] <EricDalquist> ah

[10:29:43 CDT(-0500)] <EricDalquist> right

[10:29:52 CDT(-0500)] <EricDalquist> so you need to pass around the final url for the page

[10:30:06 CDT(-0500)] <athena> yeah

[10:30:06 CDT(-0500)] <EricDalquist> which when you start dealing with redirects isn't just the url httpclient requests

[10:30:19 CDT(-0500)] <athena> that's a good point

[10:30:37 CDT(-0500)] <athena> so do we want to just add the URL to the interface's method?

[10:30:51 CDT(-0500)] <EricDalquist> well

[10:30:53 CDT(-0500)] <athena> or do we need broader access to something else (like the http response entity)

[10:30:56 CDT(-0500)] <EricDalquist> might want a ProxyRequest object

[10:31:04 CDT(-0500)] <EricDalquist> that you can add stuff to

[10:31:08 CDT(-0500)] <athena> yeah

[10:31:14 CDT(-0500)] <EricDalquist> since I'm guessing url isn't the last thing you'll find you need

[10:31:16 CDT(-0500)] <athena> yeah

[10:31:19 CDT(-0500)] <athena> that was my thought as well

[10:31:28 CDT(-0500)] <athena> was wondering if you already had other things in mind

[10:31:35 CDT(-0500)] <EricDalquist> not really (smile)

[10:31:38 CDT(-0500)] <athena> though having a ProxyRequest object at least gives us one place to start adding stuff

[10:31:39 CDT(-0500)] <athena> fair enough (smile)

[10:31:59 CDT(-0500)] <athena> maybe we could even subclass that into HttpProxyRequest and ClasspathProxyRequest

[10:32:11 CDT(-0500)] <EricDalquist> yeah

[10:32:22 CDT(-0500)] <athena> that way we could easily know that a given request just doesn't have any of that base url nonsense

[10:32:40 CDT(-0500)] <EricDalquist> yup

[10:33:14 CDT(-0500)] <athena> sounds good

[10:33:25 CDT(-0500)] <athena> i also need to subclass the content retrieval service, i think

[10:33:35 CDT(-0500)] <athena> or extend the interface, rather

[10:33:54 CDT(-0500)] <athena> since the http version probably needs to be able to return something beyond just an input stream

[10:33:57 CDT(-0500)] <athena> headers, etc.

[10:34:06 CDT(-0500)] <EricDalquist> yup

[10:34:12 CDT(-0500)] <EricDalquist> and we need to be able to set headers

[10:34:26 CDT(-0500)] <EricDalquist> also need some way to coordinate a HttpState object with the current user

[10:34:34 CDT(-0500)] <athena> yes

[10:34:41 CDT(-0500)] <athena> and that's certainly not there yet

[10:34:51 CDT(-0500)] <athena> also whatever magic is needed to make forms work

[10:35:05 CDT(-0500)] <EricDalquist> (smile)

[10:35:17 CDT(-0500)] <athena> do we currently just replay parameters or something?

[10:35:49 CDT(-0500)] <EricDalquist> oh like a proxied form?

[10:35:52 CDT(-0500)] <athena> yeah

[10:36:18 CDT(-0500)] <EricDalquist> well it isn't that much different than normal request proxies

[10:36:28 CDT(-0500)] <EricDalquist> one thing is that the rewrite filter needs to change all forms to POST

[10:36:38 CDT(-0500)] <athena> yeah

[10:36:40 CDT(-0500)] <athena> got that part

[10:36:47 CDT(-0500)] <EricDalquist> then you rewrite the form action url to something so that you know it is a form

[10:37:04 CDT(-0500)] <EricDalquist> and also all proxy portlet parameters need a prefix of some sort

[10:37:13 CDT(-0500)] <EricDalquist> then all non-prefixed parameters get passed through

[10:39:46 CDT(-0500)] <athena> makes sense

[10:40:08 CDT(-0500)] <EricDalquist> you need to encode what type of form was originally used though

[10:40:14 CDT(-0500)] <EricDalquist> so if you rewrite a get form to post

[10:40:32 CDT(-0500)] <EricDalquist> you still need to do a get in the portlet

[10:40:54 CDT(-0500)] <EricDalquist> and you need to make sure that for POSTd forms your handling url vs multipart form params correctly

[10:41:00 CDT(-0500)] <EricDalquist> 90% of webapps don't really care

[10:41:01 CDT(-0500)] <EricDalquist> but some do

[10:42:29 CDT(-0500)] <athena> ok, that makes sense

[10:42:41 CDT(-0500)] <athena> i wasn't quite sure what all the logic in the current portlet was doing - thanks!

[10:42:46 CDT(-0500)] <EricDalquist> yeah

[10:42:54 CDT(-0500)] <EricDalquist> it really isn't very clear

[10:43:02 CDT(-0500)] <EricDalquist> and I'm sure we'll miss parts as we go

[10:43:08 CDT(-0500)] <EricDalquist> but hopefully the result will be easier to grok

[10:43:25 CDT(-0500)] <athena> yeah

[10:43:31 CDT(-0500)] <athena> there's not a lot of code right now (smile)

[11:06:59 CDT(-0500)] <athena> does spring or commons have a convenient URL parsing method by now?

[11:07:40 CDT(-0500)] <EricDalquist> not that I know of

[11:07:48 CDT(-0500)] <EricDalquist> would be worth a look though

[11:07:56 CDT(-0500)] <EricDalquist> might be something in httpclienty you can use

[11:09:08 CDT(-0500)] <athena> yeah

[11:09:19 CDT(-0500)] <athena> just seems stupid to construct base urls manually from a URL

[11:16:04 CDT(-0500)] <athena> ooh i think this works

[11:16:05 CDT(-0500)] <athena> http://hc.apache.org/httpcomponents-client-ga/httpclient/apidocs/org/apache/http/client/utils/URIBuilder.html

[11:16:22 CDT(-0500)] <athena> built from the original string url, then removed the query and set the path to a blank string

[11:16:27 CDT(-0500)] <EricDalquist> nice

[11:17:45 CDT(-0500)] <EricDalquist> oo! new laptop shipped!

[11:22:50 CDT(-0500)] <athena> ooooo

[11:22:51 CDT(-0500)] <athena> yay!

[11:49:05 CDT(-0500)] <athena> EricDalquist: want to talk about the courses portlet at all?

[11:49:17 CDT(-0500)] <EricDalquist> yes but not sure I'll have time today

[11:49:23 CDT(-0500)] <EricDalquist> I'll check back in around 2

[11:49:27 CDT(-0500)] <athena> sure thing (smile)

[11:49:30 CDT(-0500)] <athena> i keep getting sidetracked too

[22:07:45 CDT(-0500)] * Topic is 'http://uportal.pastebin.com/ - https://wiki.jasig.org/display/UPC/uportal+IRC+Logs' set by EricDalquist!~apollo@76.210.64.142 on 17:44:12 CDT(-0500)