Re: [abc] Bytecode offset in SJP signature?

Ondrej Lhotak wrote:

>Hey gang...
>
>I've decided to use abc to collect dynamic program behaviour information
>(such as points-to sets). One of the things that I will need in all
>of my aspects will be a way to get the bytecode offset (and bytecode
>location in general) for each single-statement join point.
>
>Should I make this an abc extension, or is it generally useful enough
>that we should provide it natively in abc?
>
I'd be inclined to say it'd be useful enough to provide as a native
feature, as long as it wouldn't break any programs that used to work
without it. If it would have to break some (no matter how obscure)
programs, it's better off as an extension.

>In the former case, should it
>be part of the eaj extension, or a separate extension? Have we decided
>how we will handle extension composition?
>
That's a big question. Can we even do anything that's not a linear
sequence of extensions, one built on top of the other?

>For either case, does anyone
>have opinions on where the bytecode information should go? I could
>either add it to SourceLocation, or make a new BytecodeLocation and
>getter method in JoinPoint and JoinPoint.StaticPart.
>
>
Adding it to SourceLocation sounds reasonable, although the name will
then be somewhat misleading. Anyway, we need some more comments on this,
as I don't feel I know that code well enough to give an informed suggestion.

- P

>On a related note, I was looking at eaj a little bit. In EajFactory,
>there are several methods of the form:
>
>foo(parameters) {
> super.foo(parameters);
>}
>
>Why are they necessary? Shouldn't method dispatch just find them in the
>superclass?
>
>Ondrej
>
>
>
>
>
Received on Sun Sep 12 08:25:11 2004