Healthy Skepticism
In a recent column, Larry O’Brien displays a healthy skepticism of rich Internet applications:
Personally, I like native clients on the one hand and hyperlinked text on the other. I understand that this is as philistine as insisting there is no finer food than pizza, but I like process lists and taskbar icons that map directly to specific applications. I like hypermedia as the engine of application state (the fundamental philosophy of REST) and feel confident that if an application is built with such an architecture, it can be tarted up with whatever effects are in vogue.
Part of the reason I don’t like DHTML-based applications is, undoubtedly, too much knowledge of what goes into the sausage. With Internet Explorer’s loss of total market dominance, every release of a browser application has a constant pitter-patter of browser incompatibility flaws: the popups don’t work with Firefox, the text wraps in Chrome, the sliced images don’t look right in IE. I find those defects demoralizing; when I see a burndown chart flattened out because we can’t make rounded corners look right when the page scrolls or somesuch, I feel like we’ve barely progressed since the days of making windows on 80×25 screens with ASCII “line drawing” characters.
A bigger part of why I don’t like browser-based applications, though, is that for all the surprising capability of DHTML and CSS, as well as the recent arms-race in JavaScript performance, there are still huge performance benefits to be had from a local application running at full speed on the Common Language Runtime and with reasonable access to local resources. And, darn it, there are some times when you just need to draw an actual graph.
Just so.
It’s the knowledge of those gaping flaws that makes testing the damn things so frustrating. It’s the ignorance, often willful, of those gaping flaws that causes organizations to produce so many of them, many of which are crap because they don’t work whether tested or not.
