See inline [Andreas] On Fri, 10 May 2002 12:46:34 -0700, Peter Foreman <[EMAIL PROTECTED]> wrote:
>--- Andreas Håkansson <[EMAIL PROTECTED]> wrote: >> >*** BZZZZ - wrong. You ARE invasive if you ask for custom attributes. >> >You will never be able to serialize private attributes "non-invasive". >> >> [Andreas] I would be intressting to see how you could utilize >> aspect-orientated programming to get around having to derive >> all of objects that should be able to persist from a given >> class. Adding the layer of omnipresense to a class is sweet >> music to my *not-been-getting-my-hands-dirty-working-on-production >> -code-since-i'm-still-a-student* ears. > >I definitely think the generative programming route is the way to go to get performance and >flexibility. I feel it would be rare that you would need metadata at runtime for any technical >reason. I suspect it is just easier to implement via runtime reflection hence people look to go >down that route. I believe that some choose to go the runtime replication route because it's easier, but I also think that the majority start down that road since they don't know any other. I just recently came across the aspect-orientated programming model and thought that might be a better approach to do it. Please note that I made a misstake in my previous reply to this subject by claiming that aop would remove the need to derive from a base class, since there is a need to derive from ContextBoundObjects to intercept the messages. Doing so you add some overhead and gain some lose of performance. I do know however that it is possible to write AOP code without deriving from ContextBoundObject class, since thats how John Lam is doing it with his CLAW [1] project, I do not however know the details on doing it using that approach, other then it relies on the use of native (unsafe code) operations. [1] www.iunknown.com >Peter > > >__________________________________________________ >Do You Yahoo!? >Yahoo! Shopping - Mother's Day is May 12th! >http://shopping.yahoo.com > >You can read messages from the DOTNET archive, unsubscribe from DOTNET, or >subscribe to other DevelopMentor lists at http://discuss.develop.com. You can read messages from the DOTNET archive, unsubscribe from DOTNET, or subscribe to other DevelopMentor lists at http://discuss.develop.com.
