commons-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Spies, Brennan" <>
Subject RE: javaflow & aranea continuations
Date Fri, 29 Aug 2008 16:56:46 GMT

You sound a little discouraged. 

As someone who is trying to get my own project off the ground, I can
sympathize a bit with you. It's a lot of work. I can't speak for you, but my
own motivation is the challenge of the task itself. I'm scratching an itch,
if you will, to build a little cooler mouse trap.

As for javaflow, I suspect that continuations on the JVM are going to become
very important very soon. John Rose has added an initial implementation
( of it in the
OpenJDK (of course, there is still no language support in Java, but other
languages...), but this still leaves a gaping JSE 5.0 and 6.0-sized hole to
be filled. With the advent of Comet/Reverse Ajax and the planned asynchronous
API in Servlet 3.0, continuations become the "right way" of handling this. I
haven't delved too much into the current asynchronous implementations in
Tomcat or Jetty, but my understanding is that they are based on
exceptions--not really continuations at all, but a pretty weak imitation.

Honestly, I am surprised that the Tomcat or Jetty team (or any other Java app
server) has not shown interest in javaflow. It seems to me that it would be
in their best interest. The only sticking point I can think of is the ability
to control the depth of the call stack that is saved (perhaps an idea for a
feature here?).

Anyway, keep up the good work.



-----Original Message-----
From: Torsten Curdt [] 
Sent: Thursday, August 28, 2008 3:05 PM
To: Commons Users List
Subject: Re: javaflow & aranea continuations

Hey Stefano

> I've discovered this cool stuff only few days ago and I browsed a  
> lot the web searching for info and now I'm here ;-)
> I didn't find aranea-continuations mentioned in javaflow, but I  
> think it is a very interesting abstraction layer on top of javaflow,  
> it is pretty generic (core is generic, then they have specific stuff  
> for swing and their mvc framework) and it is Apache "ALv2" Licensed.

I didn't hear about that before.

> I read this:
> but it seems a bit outdated (asm support marked as "maybe" seems to  
> be there, the BCEL issues seems fixed following rhe links)...  
> Running the tests seems that ASM and BCEL enhancer have the same  
> quality level, but I may be missing something as I simply looked at  
> test results.

Well, the ASM enhancer should still be considered experimental. And  
someone reported it to be much slower than the BCEL one.
(Which is a bit of a surprise) Still I would love to see ASM being  
used as the main engine. Just because the library has a much better  

> I see there is no action around javaflow in the last year: is this  
> because of critical issues with its continuation approach or simply  
> lack of interest?

Well, I don't really use it at work anymore. And rarely people even  
just gave feedback. Not talking about any contributions. Especially  
when in the end I hear about people using it in their PhD thesis or  
some cool projects I am a little in between. "Cool!" and "WTF didn't  
you show up on the mailing lists". It's too much work if you don't get  
paid for it and don't even use it. Plus I am not a big fan of one man  

> About "talk to the RIFE guys"/"talk to the aspectwerkz and aspectj  
> folks", did anything happen?

I only have talked briefly with Geert. He was considering changing the  
license and working together on it. While it is not as generic as  
javaflow he is happy with his implementation. So I guess I just did  
not have enough energy following up on this.

> I didn't find too much activity (docs/webpages) around RIFE  
> continuations or any of the above and even about the JSR proposal.

Geert did file the JSR proposal though. Unfortunately it got rejected.  
Even from the ASF. Which really annoyed me a little. But anyway.

> I checked out the sources with eclipse and enabled m2eclipse, the  
> only issue I found is that the main pom references commons-jci-core: 
> 1.0-SNAPSHOT. I see commons-jci-core:1.0 has been released so I  
> removed "-SNAPSHOT" and it succesfully built.


> I found at least a couple of projects around using self compiled  
> javaflow libraries so it seems that even if it is incomplete it is  
> useful to the world. Is there any motivation against making a 0.1  
> release in commons or is it simply lack of time?

A few people asked for a release. But I didn't know about so many  
projects using it. (Guess I know about 3-4 projects plus a couple of  
PhD thesis). I would be happy to follow up and do a release. But there  
are a couple of things that I would like to see fixed before you  
finalize and commit to an API. And that's still quite a bit of work.

> Indeed JavaFlow is cool stuff!

Thanks :)


To unsubscribe, e-mail:
For additional commands, e-mail:

To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message