David Vallner
11/11/2006 9:00:00 PM
Charles Oliver Nutter wrote:
> David Vallner wrote:
>> The apparent popularity of Groovy disturbs me, for fear of overhyping
>> and the fact I wish their JSR a speedy death. Then again, I don't give
>> much credit to blog polls, especially ones that are instant cues for a
>> programming language flamewar. Last time I remember, it was a certain
>> Mr. Spolsky that was being accused of trolling for hits by doing more or
>> less that.
>
> I expected Groovy to do well; in the past it's been the most "official"
> language (JSR process and all), it's been hyped quite a bit for the past
> couple years, and, well, it has a pretty good name. It's certainly not a
> language I'd ever want to use though, so I think the JRuby thing has
> more potential (of course).
>
The "officiality" is a marketing stunt, nothing more. At the time the
JSR was submitted, it was pending several syntax changes, parser
rewrites, and significant feature additions (read: creep). The shameless
audacity of proposing that an unstable language without a track record
become the "official" scripting language of the JVM I found disgusting.
YMMV, but I'd much rather see either BeanShell (for the ease of adoption
and similarity), or JRuby (for subjective reasons, being proven in its
applications, and having an object model reasonably compatible with
Java, just more manipulable). The fact Java 7 will with some probability
introduce closures might make me withdraw support for the latter unless
I see a way to make the JRuby ones compatible - this might end up to be
not a problem, of course.
>> Wishing no ill fortune to the JRuby folks, I voted for BeanShell on that
>> one for rational reasons. The role I see for scripting the JVM is in
>> things usually more people on the team and later maintainers will change
>> (involved configuration, rules in business rule engines), and making
>> those easily understood and modified by people you hire to work on a
>> Java app (read: people who know mostly Java) is paramount. (If you
>> wouldn't use a scripting language on a part of the app you expect to
>> change often, where then?).
>
> Fair enough...and I agree, if Ruby didn't exist I'd probably be using
> Beanshell or Jython or SISC instead. In the end though I hope we can
> really bring all languages up at the same time by better supporting them
> at the VM level. That will help everyone, including those poor pure Java
> devs out there.
>
True, though considering that scripting Java was an option for quite a
while and remained more or less a niche activity makes me skeptical.
Then again, I've already seen valid use-cases myself (Spring bean
configuration, something similar in DWR) - bringing scripting back to
its original purpose of glue between components.
>> (Disclaimer: Anyone proposing anything but Java be used to actually
>> implement some library / framework to be consumed by other people from
>> Java deserves being handled by the nearest rusty fork. I'm a
>> multilanguage purist, keep to each its advantages.)
>
> I don't necessarily agree that Ruby can't be used for such things...but
> I agree it's best targeted at tying such things together.
>
The problem would be (apart from the usual performance considerations,
those might be negligible depending on the role of the component) just
how well that library would integrate with Java, with regards to being
type-safe, securable, etc. Then again, my surrogate brain tells me JRuby
can implement Java interfaces, so that might be an irrational concern
that can be handled well instead.
David Vallner