As I was saying a few days ago, choosing a portal is a difficult task : many specifications and many implementations. We've decided to broad our choices to any kind of portal (open source and commercial) and ended up with an impressive list (in alphabetic order): Apache Pluto, ByLine, Exo Portal, GridSphere, IPoint Portal, Jahia, Jetspeed 2, JA-SIG uPortal, JBoss Portal, jPorta, jPortlet, Liferay, Light Portal, Open Portal, Oracle Portal, SAP Netweaver, Stringbeans, Vignette Portal, Weblogic Portal, Websphere Portal Server.

Because we can‘t study them all, we had a quick look at there websites, checked if each project was still alive or not, read some feedback from the mailing lists and other source of information. After one day we ended up with a first short-list :

Exo Portal Jahia Jetspeed 2 JBoss Portal Liferay Open Portal Oracle Portal Vignette Portal Weblogic Portal Websphere Portal Server

What we are planning to do now is to take these 10 portals and go through a check list. This list has different axis :

• Specification : Which spec does the portal follow (JSR 168, 286, 301, 188, 170, WSRP V1/V2) ?
• Support : This is a long term choice for my customer. Dozens of developers will be developing portal applications and hundreds of user will use it on a day to day basis. It is really important to have good documentation, mailing list, support, training if needed and so on.
• Technical : Because the portal will be used by hundreds of users, it has to scale well, clusterable, have a cache mechanism… The developers use JSF, so a portlet to JSF bridge will be welcome. Some teams develop in .Net, and Groovy/Grails is used more and more. Interoperability is needed (not for Grails, it‘s just Java ;o) Does the portal integrate with the authentication platform (SSO) ?
• Business : Users will have different applications and navigation policies depending on their profile. The applications will mainly use Ajax. Is the portal easily skinnable ?

Once this checks are done, we will end up with 3 or 4 candidates (I‘ll let you know next week) and then we‘ll start the proof of concept bit. We might even involve the companies/open source groups in this phase. Hopefully we‘ll make our definite choice in a month or so.