robjsoftware.info

Sigh, another two weeks gone by without a post. Still, that last post was a doozy, so cut me some slack, willya? Incidentally, my readership has increased by 50% in the last two weeks according to Feedburner. I am honored to have all of you paying attention to me, and I will try to be worthy of your eyeballs and neurons.

This post is going to be a bit of a catch-all: some about me, some about my last big-ass posting, and some about the GWT conference.

First, sorry I missed last week — I got another cold (endemic in the wintertime with two young kids in the house), and my wife kicked into moving high gear. Yes, we’re selling our house, for a wide variety of reasons I will discuss in personal email if you’re interested enough to send me some. (Not in the comments, sorry!) We’re shooting for having it on the market March 1, which means there is no time at all to waste. I’ll be doing no hacking for the next two and a half months, that’s an absolute given. I’ll keep up the blogging, but weekly may be tough; bimonthly WILL happen if it kills me!

Which is all too bad, in a way, because there’s plenty to do in hackland. My manifesto was rather well received. The comments on the post itself have a number of intriguing links. I also posted about it on Lambda the Ultimate, where it got a number of other fascinating responses. Thanks to all who engaged with it in either location. I am sorry I won’t be able to touch it until March first at the earliest, but that’s the nice thing about research — interesting ideas don’t go away!

I also wanted to give an update on the GWT conference. My talk was quite well received if the consensus speaker’s feedback was anything to go by. I appreciate the appreciation, to say the least. I did hear that there was a significant fraction (over 50%?) of the audience that felt it was too technical. Evidently a substantial quorum of GWT2007 attendees were relatively unfamiliar with GWT at any level. I apologize to anyone who wasn’t experienced enough to get the most out of my talk; I hope that you can find helpful material on the GWT site and/or any of the various GWT books. I’ve put my slides online for anyone interested. Evidently it’ll be Youtubed sooner or later, but I’ve yet to get a date on that.

The last morning of the conference, we had a participatory breakfast session where Joel Webber and I led a discussion of declarative UI and declarative data binding in GWT. One thing that was clear from my talk is that there are a lot of people doing form-like web applications, and GWT underserves that population right now. Joel spent some time going over the declarative UI work he’s doing; he specifically mentioned that various internal Google customers want him to ensure that you can write a GWT-XML document that lays out a GWT widget tree with HTML fragments mixed in almost arbitrarily. This seems like a win insofar as Joel’s talk on performance specifically mentioned that innerHTML is your friend; the DOM is faster at inserting HTML fragments than it is at handling individual DOM element insertions. So everyone keep an eye on Joel’s stuff as it hits the incubator.

I asked the table what kinds of form-data-binding approaches would be most useful to implement. The consensus seemed to be split. A number of people were familiar with the JSF-like structure of binding to backing beans and using expression language to specify the coupling from bean fields to the UI. I mentioned JSR 295 and various people expressed interst in that. There is a recent thread on the GWT contributors forum about some community work in this very area, which I look forward to investigating in depth (in early March! 😛 ). Check that thread out if you want GWT to do this. I talk at more length there about a potential GWT EL parser.

There was one gentleman (whose card I got, and subsequently lost; please contact me again, sir!) who was a forceful advocate of an XForms-like approach; evidently XForms lets you declaratively specify data dependencies within your model, and allows you to lay out interactions witihn your UI purely in your interface specification. This was pretty much news to everyone at the table. I will definitely have to look into XForms now. However, it seems to me that a JSF-like, JSR-295-like bean-binding approach covers many of the basic use cases people want, and should certainly be a near-term project.

OK, that’s it for this week. Happy, happy holidays to you all!