list at joreybump.com
Thu Sep 2 13:08:23 EDT 2004
Graham Dumpleton wrote: > What I believe is more important, whatever the candidate project may be, is that > a single initial set of mock up web pages be developed by someone who is good at > web design and that everyone use these as a starting point. In other words, the > final design of each implementation should look more or less the same to the user. > The only thing that should really change is the underlying implementation which > supports it. If the user facing interface is always the same, it would make it easier > for someone evaluating the different implementations to see how each differs. > If everyone were free to come up with totally different page designs and features, > would be much harder to compare. > > Thus, start with raw HTML mock up pages which could be viewed statically with > some dummy data. PSP people could take that and convert them into PSP. The > mpservlet people could convert them to their way of doing things, similarly for > Albatross etc etc. > > Someone evalulating it can then look at how a particular static HTML page gets > translated into a dynamic page in a particular implementation. > > BTW, there may need to be two reference HTML designs. One for old style HTML > table driven approaches and one for CSS driven designs where HTML tables aren't > used for layout. This may be appropriate as different implementation methods > may be biased towards one or the other. Speaking as both a web developer and a programmer, I wouldn't find this restriction particularly interesting in a bakeoff. I'm rarely happy with the designs I'm handed, but I'm always interested in learning new techniques. An unrestricted approach provides an opportunity to learn from others, even if they don't emerge as the overall "winner". There's no question that a poorly programmed webapp that looks fantastic on the surface might steal thunder from another with a superior code base, so it's obvious that the entries would need to be judged by multiple criteria. A battery of tests needs to be devised to test performance and scalability.