The power of dynamic
I blogged about this Javacast the other day, and I finally got around to listening to it today. Tune into it at minute 1:22. The guys describe a small web app that they built using a "lite" Java stack - Spring and Hibernate instead of the full J2EE stack. Time to build - 4 months. Ok, they decided, on a whim, to try Ruby on Rails. Time to build?
4 Days
That's a pretty amazing productivity jump, and they were pretty shocked. They went on to state that dynamic languages just have it all over Java in this sphere, mentioning Ruby, Lisp, Python, and Smalltalk (stating that Ruby has momentum).
Then there's an interesting riff starting at minute 1:26, where they go into Continuation based web development, and they point out how much simpler things like Seaside are - they went so far as to say that they figure that in 3-5 years, we'll all be doing development like this. They mentioned Seaside and Squeak specifically (note - Seaside works in VW as well). They point out that you can't get there in Java, because you can't do Continuations. Instead, there are frameworks that build huge state machines (complexity alert!).
So, if you want the power and productivity now, you can grab VW or Squeak and start using Seaside. Or you can get the all too typical pale reflection in Java...
Comments
Ways and Means
[Patrick Logan] September 7, 2005 0:34:55.898
I would choose Smalltalk over Java any day of the week. If I had to use the Java platform but not necessarily the Java language per se, then I would choose SISC (Scheme with continuations) or Javascript (the Rhino? implementation in Java also with continuations), or Jython (no continuations but at least it is Python).
At least there are ways to use the Java platform without using Java. Same with dotnet. The tide is slowly turning.
Hang on...
[murphee ( http://www.jroller.com/page/murphee )] September 7, 2005 4:41:29.743
Don't you think that having written the app once already has made it easier to implement the app *again*. Most of the design (domain model,...) and GUI design etc. would have already been done in the 4 months, so the re-implementation could just re-use all that knowledge. So, that should put the "4 days instead of 4 months" into perspective.
Having said that: Ruby on Rails is giving many Java web framework developers quite a reality check by showing that it's not necessary to modify 5 files just to add a single action. I'm a Java developer but hearing the name "Struts" causes me nightmares of being stuck in one of the more kafkaesque Kafka stories...
[Vincent Foley] September 7, 2005 9:04:15.835
murphee: Yes, of course they benifitted from having a better understanding of the problem, possibly having HTML and CSS code already prepared that they could just paste and modify, etc. However 122 days versus 4 days to develop the same application? I think there's also a lot of things that Ruby and Rails do right that help achieve this sort of productivity.
so was losing alan kay another disaster for HP?
[James Governor] September 7, 2005 9:12:42.410
I had a somewhat contrarian view of the Alan Kay question. i just couldnt see an obvious HP revenue stream in the work we was apparently doing at HP, so it seemed to make business sense to me that they would decide to let him go. All this Squeak and what not. But if you're right then losing him was disastrous from HP standpoint. a spurned opportunity again. i also figured Alan might be able to do more with Squeak in the right contexts - education, for example, than at HP.