3-9-07 Interface Cell Minutes

Attending:

Stan, Kirsten, John N., Matt K, Andrea, Matt B., Mike P., Andrew, Carlos

Keeners calling in early: Matt K, John, Stan and Kirsten.

Agenda And Discussion:

• There will be a 'shame maintenance' call at this time every second Friday.

Assignment of Roles:

Documentation:

Mike P

• Keep track of phone calls
• Document user interface as we go in various directions
• Act as Liaison with documentation people from other cells

Design:

Stan, in consultation with really everyone else in the cell.

Implementation:

Matt B.

• Shepherding the process
• Coordinating between the sketching and programming sides

Evaluation:

Catherine and Matt K.

Responsibilities of the Cell:

• Responsible for user-interface design, implementation, evaluation. This includes data visualization.

Functionalities:

• Review the use-case templates and choose function based on those (including use-cases from WordHord and NORA).

Programming Language:

What programming platform shall we use?

Goals:

• We all want something slick, sexy, cross platform, that works fast.
• The development cell is committed to creating small applications that sit on top of the data, rather than are built into it.
• John N. reports (from the data cell): The object model will be as neutral as possible for talking to an end-user interface, e.g. SOAP, XML.
• Small programs then can pass things back and forth with each other. We'll need some communication protocols between these pieces e.g. the Wordhoard Reader by John and the Wordhoard calculator by Pib.

Language Choices For Interface:

OL:

• Performance / missing advanced text functionality using OL / Flash with large text collections.
• Matt B suggests: Maybe we can break up text into smaller, easier to manage chunks to help reduce performance issues?

Java / Swing:

• Swing is still missing advanced text functionality.
• It is agreed that it can be difficult to get custom look and feel beyond what Swing provides (John N reports having trouble doing the text reader with Swing. They ended up writing their own for Wordhoard inside Swing).
• Some of the things that should be easy are really not.

Coding in Swing can sometimes be like swinging chickens over your head, one after another, until one works. - John N.

AJAX:

• Still a possibility.
• More difficult than OL to get the proper look and feel, but not as difficult as Swing.
• Similar amount of development time to OL.

Homework:

• NORA members will familiarize themselves with Wordhoard (if not already).
• Matt K will look into NINES to see what they used to program their interfaces.
• Consider other programming languages (e.g. AJAX, Swing, ?) and report back during the next conference call in two weeks (3-23-07 10:00 AM MST).

I think we have a good sense that OL would work fairly well for the "Goals" stated on this page. However, I'm not sure we've kicked the tires enough on more AJAX/DHTML alternatives like Google Web Toolkit, Echo2, Flex and ZK (see this comparison of GWT and Echo2; it's also possible to roll our own client-side framework, especially with libraries like Dojo). Once the Nora interface is burried, I think it would be worth seriously considering another hackfest that would be focused on experimenting further with one of these options. Trouble is (maybe) that most of the viable alternatives are Java-based... Posted by sgs@mcmaster.ca at Mar 12, 2007 19:40