Traits are a good feature but they are localized to the class that does the extending. AspectJ is still useful for targeting a whole set of classes, say in the same package or marked with an annotation. What would be great is if the aspect could then add the mixin scala-style using the trait api - but that would require "AspectS" as mentioned by Miles.
Besides AspectJ can do a lot more than simulate mixin behavior. Personally I think there will always be room for aspectj in any of the jvm languages because of its close and honest proximity to the bytecode and because it provides a way of encapsulating otherwise scattered host language constructs whatever they might be.
Maybe we need AspectJVM, AspectJ, AspectS, AspectG and AspectC - the last one being for clojure - although It would be interesting to see a jvm language absorb an aspectj flavor directly into its compiler.
On 23 Jul 2009, at 08:31, Tahir Akhtar wrote: When I was reading "scala as long term replacement", following para caught my eyes: "Scala also has proper mixins (traits) so you don't have to muck about with AOP wackiness to get nice modular code." Being totally new to Scala at first I took it as "Scala doesn't need AOP at all". This thread made me go back to the blog post, now I understand that he was just referring to mixins and other features of AOP like advices were not subject of this comment. Am I right? Regards Tahir Akhtar Dave Whittaker wrote: I'm not saying I'd advocate switching focus from Groovy to Scala, but for what it's worth the thing that got me interested was that Groovy's creator, James Strachan, blogged about Scala recently and said: "I can honestly say if someone had shown me the Programming in Scala book by by Martin Odersky, Lex Spoon & Bill Venners back in 2003 I'd probably have never created Groovy."
I looked through the tour last week and it does seem awfully cool...... On Jul 22, 2009, at 4:48 PM, Ashley Williams wrote: So the scala team would have to get up to speed with the pointcut parser and coordinate with the aspectj team. That's assuming they are already familiar enough with aspectj already. And then put some sort of implementation together and go through many test and release iterations. Whilst all the time attending to main scala issues.
Given that I don't even see much interest in pursuing this, it looks scala users will need to be familiar with the source transformations for a long time to come in order to use aspectj. Unless you suddenly switch from Groovy to Scala, Andy ;-)
On 22 Jul 2009, at 17:59, Andy Clement wrote: The AspectJ pointcut parser is a reusable component (used in Spring Framework, for example). However, it does have a well defined set of join points which may or may not make sense in the context of another language (most likely more would be needed too).
The AspectJ pointcut matcher is a reusable component. If you can conjure up something that looks like a join point (using whatever means you like), it will tell you whether a pointcut matches that and what any required runtime test would be. The AspectJ weaver is an example of something that consumes bytecode and presents pieces of it as candidates to the pointcut matcher. We are progressing with also implementing a Eclipse JDT source level system that would allow AspectJ pointcuts to be matched against Java source (so matching without the need for compilation).
Personally I'm spending more time with Groovy right now than Scala. However, in both cases AspectJ will have issues with the compiled code. We already have an open bug about being unable to weave Scala code and I know that groovy code does a huge amount of transformation to produce its class files so when AspectJ would be looking for join points, they would be very hard to pick out. I would estimate it is a large amount of work to get weaving and the semantics of that weaving consistent across these languages.
Andy
2009/7/20 Ashley Williams <ashpublic@xxxxxxx> I second that question. Am I right though in thinking that AspectJ is really about targeting join points at the byte code level? If this is remains true - and I can't see how it would be practical to change this - then you'd always have to be aware of how the scala source code for example gets de-sugared.
I'd be very interested in tighter integration between scala and AspectJ.
- Ashley
On 20 Jul 2009, at 14:02, Michael McCray wrote: Hi All, I was wondering if it is required that AspectJ is tied to the Java Language? I see that is what the J is for, but there are some new languages like JavaFX and Scala that I don't think have supporting aspect extension languages. There is http://functionaljava.org/, a variant of Java that supports closures and other functional language concepts. What if AspectJ could become AspectJVM and support other JVM languages? I understand LTW can do much of what I ask, how far off is LTW from a real integration? I mean it would be nice to write my aspect code using other language features as well at least this would not be possible until the aspectj language specification changed. Why not add closures to AspectJ to support it's usage in aspects? How is it decided to evolve the language? Mike
_______________________________________________
aspectj-users mailing list
aspectj-users@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/aspectj-users
_______________________________________________
aspectj-users mailing list
aspectj-users@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/aspectj-users
_______________________________________________
aspectj-users mailing list
aspectj-users@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/aspectj-users
_______________________________________________
aspectj-users mailing list
aspectj-users@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/aspectj-users
aspectj-users mailing list
aspectj-users@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/aspectj-users
|
