How could development in Liferay be improved?

Could development in Liferay be improved? The answer is allways yes. And, how could development be improved? This may have several answers, and one of them is Romulus.

Romulus is an R&D project funded by the European Union. The overall purpose of the project is to improve the productivity of developers of the Java platform (which has a huge installed base in Europe) by applying a variety of innovative ideas and techniques. The project involves a lot of research including evaluation of development technologies available (for Java and other platforms), determining the main pain points for developers, identifying areas where more productivity can be achieved, etc. It also includes the improvement of the two Open Source projects involved in Romulus, Roma and Liferay, to make it easier to develop for and with them.

Eurpean projects are carried out by consortiums of research labs and companies. Together with Liferay there are other 6 partners: Germinus (a Spanish company and project leader), AssetData, Imola (both Italian companies), Deri (an Irish reseach lab), UPM (the largest Spanish technical university) and ICI (a Romanian research institution).

What are Liferay's tasks? We are responsible for all the research related to portals as well as for all the development oriented to achieve maximum productivity when working with Liferay Portal in particular. The project developments will be made to allow faster portal app development through the use of scripting languages and frameworks, allowing the creation of client side applications (aka lifelets) and portal mashups (reusal of functionalities provided by the portal by portal applications), etc.

That's about it for now. For those interested please feel free to ask questions and offer ideas. There is a website of the project (developed with Liferay) available at http://www.ict-romulus.eu/

Blogs
I can see some areas to improve :

- script : a portlet plugin to grails can be very usefull. Or another way to develop potlets on a script language.
- to be able to extend liferay without the need of use the ext enviroment.
- better error messages when something goes wrong.
- make everything hot deployable.
More areas to improve:

-Full i18n support (Polls portlet, Web Form portlet, Calendar portlet, welcome message, terms of use agreement page).
-Upgrade to Spring 2.5. Including AOP for cross cutting concerns (transaction, log, audit), creating Liferay NamespaceHandler for easy spring config, dependency injection for core struts actions.
-Support for multiple environment, DTAP model (Development, Test, Acceptance and Production).
Maven support would nice for those of use who use maven. Full on maven adaptation isn't required. A good starting point could be just providing maven artifacts for all the jars and war. Next step could be creating some maven plugins that would help creating Liferay plugins such as having Liferay plugins sdk as a maven plugin. Also Service Builder plugin would be great.
I have to strongly agree with this - increasingly I can't imagine living w/o Maven - and the hoops required to jump through to use Maven in Liferay have been painful.

Some items I wish for:
* Easier testing - better hooks for mock/unit testing & setups
* Much better developer documentation (e.g. much of the ServiceBuilder produced classes can be very difficult for a new developer to decipher, since they tend to lack any reasonable javadoc/others
* simpler application of cross-cutting concerns & capabilities (comments, blogs, auditing, backups, i18n, etc - e.g. ideally you should be able to easily implement some standard java i18n and have Liferay be able to dynamically inject langauges/bundles

On the tools side, have you looked at Spring Roo? I'm not sold yet, but it looks interesting