[PJUG Javamail] great presentation last night

[Howard Abrams]

I agree with that assessment, Howard did a great job, and it was all  
very entertaining and illuminating. I for one, love learning new  
languages. However, as was pointed out last night, I seldom risk the  
maintenance associated with introducing something vastly different.

One perspective: During college, I spent a couple of years deeply  
involved with Lisp after many years working in C, and I found Lisp  
baffling and unreadable. However, it didn't take long to get the hang  
of it. So, any time you encounter something new, don't sell yourself  
short. Of course, I can't make heads or tails out of some Perl scripts  
I wrote last year, so there you go...

Concerning supporting someone's code written in Clojure vs. Java: you  
still need to mentally parse through the code. And if something can be  
done in 2 lines of dense Clojure vs. 20 lines of Java, sometimes it  
may be quicker to parse the 2 line even if you aren't as familiar with  
the syntax. I run across this all the time using Groovy (which is more  
succinct than Java).

Last thought... I first ran across the transactional memory in a  
Haskell presentation, and was intrigued. Of course, most of us depend  
on transactional support in our databases (which also gives us  
persistence). However, I wondered if we couldn't just write a library  
to implement a flexible type that could hold data and handle changes  
to it transactionally. Something like:

	STM stm = STM.getInstance();
	stm.startTransaction();
	float checking = stm.get('checking');
	float savings  = stm.get('savings');
	stm.put ('checking', checking - amount );
	stm.put ('savings', savings + amount );
	stm.endTransaction();

However, that is quite a bit of extra code that something like Clojure  
could with less verbosity.
Hrm... just looking at that Java code makes me thinking how much less  
code (and more clarifying) I could make it if I used a Groovy builder  
and closures...

Anyway, Sun has been working on putting transactional memory into the  
JVM:
	http://video.google.com/videoplay?docid=4575567975642725150

Also, it looks like someone is making transactional memory as a  
library for Java:
	http://web.ist.utl.pt/~joao.cachopo/jvstm/


On Jan 21, 2009, at 9:11 AM, Benjamin Berman wrote:

> Just wanted to thank Howard for the great presentation on Clojure last
> night. While I disagree that Clojure is easy on the eyes compared to
> Java, I like the idea of software transactional memory-- much more
> elegant than Java concurrency support.
>
> Regards,
> Benjamin Berman
> _______________________________________________
> Web Site - http://www.pjug.org/
> Javamail mailing list
> Javamail at pjug.org
> http://www.pjug.org/mailman/listinfo/javamail