Re: [abc-users] Obtaining jimplified advices in a ReweavingAnalysis

From: Eric Bodden <eric.bodden@mail.mcgill.ca>
Date: Fri Jun 15 2007 - 16:52:24 BST

On 15/06/07, Nathan Weston <npweston@gmail.com> wrote:
> Thanks for your (very) swift reply Eric!
>
> What I'm trying to do is to analyse advice essentially pre-weave-time
> - the idea is to do some pre-processing on the advice such that when I
> come to the analysis on the woven program I have some partial results
> which can be reused in the analysis. Ideally I'd like to analyse
> methods pre-weave-time as well - i.e. before invocations of advices
> are included.
>
> When you say getting the Jimple from an AdviceDecl, do you mean simply
> getting the Methods associated with the Application Classes and
> retrieving their bodies? That's what I'm doing presently, and as you
> say around advice is a little tricky :)

Yes, that's what I mean. If you want to analyze code prior to weaving,
this analysis should probably not even go into the RewavingAnalysis.
You should rather add a compiler pass to your ExtensionInfo, right
after the pass JIMPLIFY_DONE and execute the analysis/preprocessing
from there. At this point in time, you have Jimple for all advice
bodies but nothing has been woven yet.

Eric

-- 
Eric Bodden
Sable Research Group
McGill University, Montréal, Canada
Received on Fri Jun 15 16:52:30 2007

This archive was generated by hypermail 2.1.8 : Fri Jun 15 2007 - 17:00:10 BST