Tapestry 2009 Summer Tour: England, Estonia

Just dropping a note from the road ... I'm currently in York, England for three days of Tapestry training here for a private client. Next week, five days of training and mentoring in Tartu, Estonia. I was a bit trepidatious about heading to Estonia, but now that I've read up on its history, I can't wait to get there.

After that, a bit of vacation in London and Paris ... then back to Portland (hopefully the temperatures will have dropped) to concentrate on Tapestry 5.1 documentation ... and learning Cappuccino for my upcoming talk at The Pacific Northwest Software Symposium.

Infrequent commands

An odd usability thought just hit me as I'm making some simple updates to my slide decks in Keynote. Adding and positioning the page number is not something you can do without using the mouse (to click the button in the inspector and drag it into place). There's no menu item for this.

That's normally OK with me ... it's a very infrequent operation so why take up valueable menu space (as well as valuable user comprehension space) with it?

But the odd thing is that I think quite often, when you need to use an infrequent operation you need to use it a lot at once. That's probably why Microsoft applications have byzantine menu structures ... they never want anything to require the mouse and from a usability perspective they throw the baby out with the bath water over those infrequent commands.

All I'm missing from Keynote is a simple macro-recording feature: turn on page numbers, select it, drag it to position, update its style. I'd love to be able to run through my master slides (10 per presentation times 7 presentations!) and just it cmd-m or something.

Of course, if Keynote had master-masters, I might be able to make the change in one place per presentation (which would apply to the ten masters and the 60 or 70 slides per presentation).

Caught between Two IDEs

I seem to be caught between two IDEs: Eclipse and IntelliJ. I abandoned Eclipse a couple of years back, partly based on wide spread recommendations from many different people, and partly because Eclipse just stopped working for me (it crashed out).

After I got started with IntelliJ I started to appreciate its merits, despite a generally clunky interface (with lots of modal windows), truly awful documentation. Many things are streamlined and only a ctrl-alt-shift-coke-bottle-touch-your-nose away.

However, over time, using IntelliJ got slower and slower and slower. It also started running the Tapestry test suite horrifically slowly: 40 minutes and up (it should be about five). It would often go away, even when memory wasn't tight. Indexing? Checking Repositories? Computing primes? No way to tell.

Meanwhile, Eclipse has been moving forward, with Eclipse Galileo being a Cocoa (not a Carbon) application. Critical plugins such as M2Eclipse have gotten nice, and the Clojure plugin is mostly better than the IntelliJ one (though both are very early).

For a while I was using IntelliJ when teaching Tapestry (as part of the VMWare image I use when training) ... and I got a lot of resistance. People were much happier with Eclipse on the last couple of go-rounds, and I'm sticking with it.

Overall, I'm feeling that most of what I've grown used to in IntelliJ is present in Eclipse, just handled a bit differently. The Clojure plugins are a wash; IntelliJ has the edge on the Git plugin. I think Subversion inside Eclipse is actually better.

I've even cranked up NetBeans but didn't find anything there compelling enough to switch.

It seems like all my major tools (Firefox, Firebug, Eclipse, IntelliJ) are in the habit of growing too complex, and doing too much stuff in the background that I don't care about. All those intentions in IntelliJ that you have to turn off (for performance reasons), and all those extra plugins for Eclipse that you need to not download in the first place ... they're all getting in my way.

I think a lot of this falls into the general category of accidental complexity ... to address the limitations of the Java programming language, all this extra stuff is coming into play: tools and wizards and plugins and indexes and whatnot. I find it pretty pleasant to work with Clojure instead, where the accidental complexity of Java is managed and isolated and the IDE doesn't feel the need to be overly ambitious. That's the Clojure concept right there ... grow the language to your needs, rather than building up tools. I think that's the Tapestry ethic as well.