Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [aspectj-dev] BCEL vs. ASM


ASM certainly makes some interesting claims about performance, that I have no reason to disbelieve. It also seems that they have changed licence recently to BSD (I'm sure of the exact t's & c's of that license, and whether it is compatible with the CPL, but I'm pretty sure it's better than their previous licence). License investigations aside, the weaver was originally designed to make replacing the weaving engine an option - but in practice we have a lot invested in the BCEL implementation and a lot of AspectJ programs out there that depend on its behaviour, so **safely** swapping bytecode libraries would be a big job. Also note that AspectJ uses a patched version of BCEL that is significantly quicker than the original - Erik replaced some core logic that used reflection with a straightforward lookup.  As a disclaimer, I haven't yet had a chance to look through the ASM apis and see whether or not they cover all the function that AspectJ would need either.

It's definitely something worthy of serious investigation, but would have to be prioritized alongside all of the Tiger support and exploitation work we need to do in the next release.

If there's a research group out there interested in building an ASM-based prototype of the weaver and making some comparisions I'd be very interested to see it...

-- Adrian
Adrian_Colyer@xxxxxxxxxx



"rbodkin" <rbodkin@xxxxxxxxxxxxxx>
Sent by: aspectj-dev-admin@xxxxxxxxxxx

08/04/2004 20:03

Please respond to
aspectj-dev

To
aspectj-dev@xxxxxxxxxxx
cc
Subject
Re: [aspectj-dev] BCEL vs. ASM





I've heard from several sources that ASM is significantly faster and uses
less memory. I'm also curious how much work it would be to switch to it.

On Thu, 08 Apr 2004 20:50:52 +0200, Martin Lippert wrote
> Hi,
>
> are there any thoughts on replacing the BCEL library with the ASM
> bytecode library for the weaver implementation (because of possible
> performance improvements)?
>
> And how tight do you think is the coupling between the weaver
> implementation and the BCEL library in the current version of AspectJ?
>
> Just asking... ;-)
> Martin



Ron Bodkin
Chief Technology Officer
New Aspects
m: (415) 509-2895
_______________________________________________
aspectj-dev mailing list
aspectj-dev@xxxxxxxxxxx
http://dev.eclipse.org/mailman/listinfo/aspectj-dev


Back to the top