Coty's Weblog

New gig

This has never been much of a work blog, but I thought I'd mention that last month I joined Vocalocity as VP of Engineering. It has only been a few weeks, but I'm having a really good time. I've got a lot to learn, but I've got great people all around me to learn from.

Certainly not the least of those is our CTO, Jeff Haynie, who got his blog going again late last year. His latest post addresses the role of of VoiceXML and CCXML in an IMS architecture and includes a link to his recent presentation on the topic at SpeechTek West.

I don't know how much the new gig will influence my blogging. I'll certainly leave the deep voice tech. to Haynie, but you may notice more posting on Java, software development, etc. here than there has been recently.

It was going to be a really cool post, too

I was just about done with a relatively long post. I was using Performancing's Firefox blog posting extension and switched from the editing view to the markup view to see whether an HTML entity I'd typed in would carry over. It didn't. Nothing did. Crap.

Oh, well. It was a post about Mark Liberman's Old School post on Language Log commenting on Joel Spolsky's The Peril of Java Schools post and how it was cool for two otherwise unrelated blogs to come together like that. Maybe I'll try to resurrect my thoughts, but it probably isn't worth it.

KMaps evolving

Ivan is on fire. KMaps is just getting better and better. Ivan has moved the latest version from the blog where he introduced it to a new site. Ivan has made a few fixes, and the latest KMaps is working just fine on my Treo 600. More importantly, though, Ivan has really started showing the location-based services that KMaps is enabling. IMO, this is bigger than the maps component. The maps are an enabler, but ultimately it is going to be the traffic reports, the restaurant reviews, the parking assistance, etc.

I need to look around to see what is happening on other mobile platforms in this space. I know about dodgeball, but that is pretty different. I'm sure there must be similar things out there for Symbian, etc.

KMaps - Google Maps on the go on Treo

Ivan Mitrovic (with whom I had the pleasure of working at my last job) has made my day.

I love Google Maps. Who doesn't? I love my Treo. (I wish it were a 650, but I still love it.) Alas, the Treo browser is far from being able to handle the rather ingenious JavaScript used to make Google Maps work. (Hopefully mobile device makers realize they are going to need to address this JavaScript weakness soon.) To get around this limitation, Ivan has created KMaps, a Java application that uses the Google Maps API to give you a great mapping experience on your Treo.

If you have a Treo, get it. (Ivan tested on his 650, but it is working on my 600, as well.) If you have another mobile device (Nokia Series 60 and many others), you might try Mobile GMaps. It works on a Treo, too, but it requires access to a image conversion server that isn't required by KMaps.

Thanks, Ivan!

Strange(berry) News

A while back I wondered aloud what Arthur van Hoff and others were doing over at Strangeberry. I ran across them while looking for a Lisp implementation of ZeroConf/Rendezvous. They had a Java implementation, but it wasn't clear what else they were up to. Well, it still isn't obvious, but whatever it is must at least have something to do with television as they were acquired by TiVo in January according to TiVo's latest 8-K filing.

MultiJava [Curiouser and curiouser!]

MultiJava is an open project to add open classes and symmetric multiple dispatch to Java.

Open classes allow one to add to the set of methods that an existing class supports without creating distinct subclasses or editing existing code. Multiple dispatch, found in languages such as allows the method invoked by a message send to depend on the run-time classes of any subset of the argument objects.

In concert with AspectJ this would appear to add some very powerful capabilities to Java.

Of course Greenspun might argue that you should just use Lisp in the first place... ;-)

[Curiouser and curiouser!]

I'm going back through a queue of drafts in NetNewsWire, so this is a bit out of date. Nevertheless, I think it is interesting enough to post. I saw the poster for this project at OOPSLA in 2001, and I was really excited. The funny thing is that I wasn't excited by the multiple dispatch that gives MultiJava its name. Rather, it was the open classes that I find exciting. This ability to add new methods to existing classes was one of my favorite features of Objective-C. C# partial types sound a bit like this, but the implementation seems all wrong to me. What is great about Categories in Objective-C is that they allow you to extend binary code after the fact. Partial types in C# require that you use the partial keyword up front. LtU has a couple of threads (here and here) discussing this in more detail.

Start Ant 5 times faster! [crazybob.org - web log]

While at the TSS conference last weekend, Drew Sliwkowski, a clustering guru for BEA, pointed me to an awesome Ant task called Server. Basically, you start an ant server in the background and use a native client to make it execute targets. By doing this, you avoid both the JRE and Ant startup times. Whereas running an empty target 5 times took 13s with the classic method, it only took 2.3s with the server method (or 1/5th of the time). ...

[crazybob.org - web log]

While I haven't been doing anything with Java recently, this looks like a good tip. It never hurts to speed up builds. In fact, in my experience, it can really help a project.

Detecting Duplicate Code

Tom Copeland has written a nice piece on the open source CPD (copy/paste detector). Having tools like this can really help out, and seeing the amount of copy/paste in the JDK source itself is scary. However, what do I *really* want these types of tools to find out... [ONJava.com]

I haven't tried this CPD thing, yet, but it sounds intriguing enough to throw right into the blog. The worst part about copy/paste in my experience is that it is always old code that got written when you didn't know better that gets promulgated.

Checked and unchecked exceptions in Java

This is yet another concrete language design decision: Should exception consistency be checked?

Checked exceptions, declared as part of subroutine signatures, are appealing. They make the abstraction interface more readable, and ensure error handling consistency.

However, they introduce coupling into the system, can be seen to break encapsualtion, and can cause a ripple effect of changes during maintenance.

This article weighs the pros and cons.

[Lambda Web Log]

Cites the Eckel article I mentioned a few days ago. This article has a good discussion of layering and exceptions that is worth giving some consideration.

Bruce Eckel: Does Java Need Checked Exceptions?

Bruce Eckel is the author of Thinking in Java (download) and Thinking in C++ (download 1/download 2) before that. This article gives some background on the possible evolution of checked exceptions in Java and questions their necessity/usefulness. I particularly like the ExceptionAdapter that Bruce and Heinz Kabutz (which reminds me that I need to re-subscribe to The Java Specialists' Newsletter) put together to standardize the use of RunTimeException to bypass exception checking while preserving the original exception type, as well as the comment from one of the C# language designers regarding the utility of exception specifications in large vs. small projects. I think it is dead-on in noting the attractiveness of checked exceptions in small examples vs. their many drawbacks when developing in the large.