[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
RE: [aspectj-users] Parameter construction pointcut?

I don't see why there couldn't be a precall join-point which preceded all argument evaluation.

f( g( h(), i( j() ) ), k() )

would match, in order, 

precall( * f( .. ) )
precall( * g( .. ) )
precall( * h( .. ) )
call( * h( .. ) )
	execution( * h( .. ) )
precall( * i( .. ) )
precall( * j( .. ) )
call( * j( .. ) )
	execution( * j( .. ) )
call( * i( .. ) )
	execution( * i( .. ) )
call( * g( .. ) )
	execution( * g( .. ) )
precall( * k( .. ) )
call( * k( .. ) )
	execution( * k( .. ) )
call( * f( .. ) )
	execution( * f( .. ) )

Josh

-----Original Message-----
From: aspectj-users-bounces@xxxxxxxxxxx [mailto:aspectj-users-bounces@xxxxxxxxxxx] On Behalf Of Eric Bodden
Sent: Friday, August 10, 2007 11:06 AM
To: aspectj-users@xxxxxxxxxxx
Subject: Re: [aspectj-users] Parameter construction pointcut?

On 10/08/07, Seva Popov <seva_popov@xxxxxxxxxxx> wrote:

This has come up before...

> Please, confirm my assumptions or suggest me any solutions besides
> fixing the code itself.

Your assumption is correct.

> P.S. I understand that AspectJ operates on the bytecode but not on the
> source code. But is it theoretically possible to augment the AspectJ
> with a parameter construction pointcut? Just curious.

That is impossible because it is undecidable in general. Where does
parameter construction start in general? Not even a human being could
answer that question.

W.r.t. your problem: I doubt that parameter construction is a
bottleneck in your code if all the construction is as easy as in the
case you showed. Maybe a profiler would give you more insight.

Eric

-- 
Eric Bodden
Sable Research Group
McGill University, Montréal, Canada
_______________________________________________
aspectj-users mailing list
aspectj-users@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/aspectj-users