Hi,
Unfortunately, I can't mail the heap dump. It is way to big (1.6G) and
contains information I cannot share. However, all the memory consumption is in
the same area: the world. This is what the AjState contains for the largest
project:
AjState: 340M
AjState.world: 220M (excluding model)
AjSt
Hi Stephen,
your best options for now to understand what is being captured by pointcuts are
the Eclipse AJDT tools. You can e.g. get a list of all source code locations
affected by a pointut (i.e. join point shadows). There also is a basic
visualization: the Aspect Visualization perspective. Su
Ramnivas, sorry to have to contradict you (and thanks for triggering my memory)
but there has been some work on a tool that does just that: pointcutdoctor (
http://pointcutdoctor.sourceforge.net/ ) by Ye and De Volder. I used it in its
early incarnations and it proved to be quite helpful. Sadly
Hey Andy,
Sorry about the last confusing email. I was able to figure out what was causing
me to have problems with using aspects inside the bundle where the aspect was
defined. The ant build just has to have the inpath set correctly. I was having
trouble because of some stale copy compilation b
Ramnivas is aware of the pointcut doctor (I believe) - we (AJDT) are
looking to try and revive the doctor as we are seeing an increasing
number of users that would benefit from what it can do.
Andy
On 14 July 2010 06:19, Johan Fabry wrote:
>
> Ramnivas, sorry to have to contradict you (and thank
Your code works for me. Here it is in a more complete example:
@Aspect
class Azpect {
@Pointcut(value = "execution(public * *.getManagedClass())")
public void notInMethods() {
}
@Around(value = "!notInMethods() && execution(public * *.get* (..))
&& this(d
Yes, proxying could be getting in the way and you may need to adjust
your pointcut to cope with an invocation via a proxy (particularly if
the proxy is actually invoking the code via reflection - using AspectJ
with reflective invocation needs special consideration).
Andy
On 14 July 2010 08:02, Ka
That would be cool, I hope it works out ! :-)
On 14 Jul 2010, at 11:28, Andy Clement wrote:
> Ramnivas is aware of the pointcut doctor (I believe) - we (AJDT) are
> looking to try and revive the doctor as we are seeing an increasing
> number of users that would benefit from what it can do.
>
>
Hey,
Yes, that is where I expected memory to be going.
> I tried adding -Xset:typeDemotion=true, but that doesn't help (I thought that
> option was enabled by default since 1.6.7). I also disabled the spring-aspect
> library, reducing the number of aspects to 1, which only contains declare
> erro
Indeed, PointcutDoctor could be of help--especially to flag errors based on
heuristics.
-Ramnivas
On Wed, Jul 14, 2010 at 9:18 AM, Johan Fabry wrote:
>
> That would be cool, I hope it works out ! :-)
>
> On 14 Jul 2010, at 11:28, Andy Clement wrote:
>
> > Ramnivas is aware of the pointcut docto
Thanks Andy, I did not attach the code beyond the pointcuts due to the
pointcuts not working correctly. I assumed there were a glaring problem in
the pointcuts that anyone beyond a novice (like myself) would just catch at
a glance.
But apparently the pointcuts not working are due to my toolchai
Stephen,
I don't believe PointcutDoctor works with current AJDT (as Andy mentions, it
may be revived soon).
Just to clarify: The kind of snippet Andy included is what I was looking for
(not just aspect, but also a representative target Java class). That way it
is easy to understand your intention
It seems to work just fine by using "execution". Is it okay to use execution?
From the documentation, like call vs. execution, it seems like call is the
preferred way of doing things? Is it something I shouldn't worry about or are
there negative things I should consider about just using executio
On Wednesday 14 July 2010 18:19:50 Andy Clement wrote:
> Yes, that is where I expected memory to be going.
>
> > I tried adding -Xset:typeDemotion=true, but that doesn't help (I thought
> > that option was enabled by default since 1.6.7). I also disabled the
> > spring-aspect library, reducing the
I prefer execution personally. It will cause AspectJ to intercept one
place, the execution place, unlike call which will intercept
potentially hundreds of call sites. execution cannot be used if you
aren't able to pass the code where the execution join point exists to
the AspectJ compiler for wea
Yes, any solution I put together will work with incremental
compilation. The structure model changes are much more
straightforward than using typedemotion in a compile time weaving
environment, so I'll get that part of the change out first I think.
Andy
On 14 July 2010 11:29, Emond Papegaaij wr
It sounds like I am good to go then :)
Thanks for responding to all my Emails Andy (there were quite a few!). I have a
way better understanding of what is going on now, and everything works :D
From: aspectj-users-boun...@eclipse.org [aspectj-users-boun...@
Did not see this response before my other comments.
Thanks for your responses. I'll try your suggestions. Perhaps it is AJDT.
I had to do the first approach as the class is a very common base class that
I need to enhance but true to your comments, I just found another class that
needs roughly t
On 14 July 2010 13:24, aappddeevv wrote:
> I should have added that my aspects library (aspects and classes) processes
> 2 other libraries on its inpath to add methods/attributes to those classes.
> In fact, the base class I am enhancing comes from that
> external-to-the-project library and the cl
Does annotation based @AspectJ support precedence?
If I have
@Aspect
public class LowerPriorityAspect {
..
}
@Aspect
public class HigherPriorityAspect {
..
}
Then
public class SomeAdvisedClass {
public void someMethodAdvisedByBothAspects() {
}
}
So, how would I perform the equivalent of
Hi Stephen,
Did you see @DeclarePrecedence ?
http://eclipse.org/aspectj/doc/next/adk15notebook/ataspectj-declare.html
cheers,
Andy
On 14 July 2010 15:49, Stephen Boesch wrote:
> Does annotation based @AspectJ support precedence?
> If I have
> @Aspect
> public class LowerPriorityAspect {
> ..
>
oh nice!
I did not even know about the Developer's Handbook. That's goldmine stuff.
I just perused the whole thing, and the "if()" may provide a workaround for
some of the issues i'm facing..
2010/7/14 Andy Clement
> Hi Stephen,
>
> Did you see @DeclarePrecedence ?
>
> http://eclipse.org/aspec
22 matches
Mail list logo
