[aspectwerkz-user] Re: Hot deployment questions

Jonas Bonér <[email protected]>
Newsgroups gmane.comp.java.aspectwerkz.user
Message-ID <[email protected]>

On Mon, 11 Apr 2005 09:44:13 +0200, Alexandre Vasseur <[email protected]> wrote:
> Jonas: yes / no
> 
> Given the AW weaving model that is using this jit JoinPoint concept,
> we have to open classes in some cases f.e. call of private members
> thus we have to do schema change as well.
> 

Sure, I made a simplification to make a point (after all call pointcuts are not so often needed on private members): that we require the use of deployment scope in many cases where it is actually not needed

> Aside, on a conceptual level I like this scope idea since it gives you
> a strong way to ensure that ONLY classes (or pointcut) that you
> actually define will be exposed to the hot deployment API ie you can
> make sure to exclude some critical parts of your app etc.
> 
> What do you think ? There has been a long discussion about aspect and
> security on AOSD-discuss recently, and LTW and hot deployment were
> seen as a risk by some. Do you see any area in AW that we should
> refine to allow more exclusions / borders.
> 

Sure, we have been over this before and that is why the implementation is as it is. 

I am in favor of using deployments scope.  It will among other things, make the team you're working with less scared of what you are doing since you're working your magic within predefined borders ;-)



> Alex
> 
> 
> On Apr 11, 2005 2:22 AM, Jonas Bonér <[email protected]> wrote:
> > this preparation step made by the deployment scope construct is only
> > necessary for execution pointcuts, since they are the only ones that
> > triggers a schema change (since we need to add a method to implement
> > this construct).
> > 
> > However, this distinction has not been made in the current
> > implementation (although it would not be hard to do so).
> > 
> > /Jonas
> > 
> > On Apr 10, 2005 11:57 PM, Vincent Massol <[email protected]> wrote:
> > > Hi Michael,
> > >
> > > I believe the reason it works is simply because the modified classes are
> not
> > > yet loaded by the classloader when the aspects are weaved using hot
> > > deployment. Otherwise, I believe it won't work and a deployment scope
> would
> > > be needed.
> > >
> > > In other words, it's only by chance that it works and in general you
> should
> > > always use a deployment scope when doing hot deployment.
> > >
> > > Thanks
> > > -Vincent
> > >
> > > > -----Original Message-----
> > > > From: Michael Haupt [mailto:[email protected]]
> > > > Sent: dimanche 10 avril 2005 18:19
> > > > To: [email protected]
> > > > Subject: RE: [aspectwerkz-user] Hot deployment questions
> > > >
> > > > Hi Vincent,
> > > >
> > > >
> > > > Vincent Massol said:
> > > > > Thanks you very much Mihir, that's very helpful. I have found that
> it
> > > > > works without any aop.xml and without Deployment Scope.
> > > >
> > > >
> > > > could you please post your source code (or a very small example)? I'm
> > > > interested in knowing how hot deployment works without aop.xml and
> > > > deployment scopes.
> > > >
> > > > Best,
> > > >
> > > >
> > > > Michael Haupt
> > > >
> > >
> > >
> > 
> > --
> > /Jonas
> >
> 

-- 
/Jonas
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.