Fwd: [Soot-list] Paddle, JEDD & extending the points-to analysis

[Alan Ryan]

I forgot to post this to the community....

---------- Forwarded message ----------
From: Alan Ryan <alan.ryan at gmail.com>
Date: Sep 4, 2006 6:40 PM
Subject: Re: [Soot-list] Paddle, JEDD & extending the points-to analysis
To: Ondrej Lhotak <olhotak at uwaterloo.ca>

Hi Ondrej,

Thanks for getting back.  Sorry about the ambiguous explanation, I will try
and make myself clearer.

The extra functionality which comes with Paddle such as the different levels
of context sensitive  analysis, set representation etc.. are not of core
interest to me.

I want to use Jedd to write, for example,  a field & context sensitive
analysis but using different and additional domains that will suit my end
use - which is a sort of whole program reaching defs analysis. The default
analyses in  Paddle are not wholly suitable, but provide me with great
building blocks from where to start.


>I don't understand the connection with Eclipse. Do you want to write
>your extension in Eclipse, or do you want the extended Paddle/Soot
>to run within Eclipse?

Sorry again, I see where I misled you here.  I wish to develop in the
Eclipse environment using Jedd.

I want to use Jedd, whilst availing of the jimple provided by Soot in the
scene. Should I use the Paddle analysis as a template, write all my own Jedd
files and then run this within Eclipse so that I can debug etc.. essentially
creating a customised & scaled down version of Paddle?

I think you have already cleared up some misconceptions that I had.

Thanks for both your time and Jedd/Paddle

Alan





On 8/28/06, Ondrej Lhotak <olhotak at uwaterloo.ca> wrote:
>
> I'm not quite sure I fully understand your questions, but I'll try to
> explain.
>
> The pointsto example that comes with Jedd is a small example of a Jedd
> program. It does only points-to set propagation, using constraints read
> in from a file.
>
> Paddle/Soot is a much larger example of a Jedd program.
>
> When you say you need to extend the "default analysis", I don't
> understand which analysis you mean. It sounds like you want to extend
> Paddle, but then you ask whether you should extend Paddle or the
> small example. Do you need only points-to set propagation, or all the
> other things that Paddle/Soot provide?
>
> I don't understand the connection with Eclipse. Do you want to write
> your extension in Eclipse, or do you want the extended Paddle/Soot
> to run within Eclipse?
>
> Ondrej
>
> On Wed, Aug 23, 2006 at 06:52:03PM +0100, Alan Ryan wrote:
> >    Hi,
> >
> >    I am using Paddle to run a context-sensitive pointer analysis for my
> >    code.  I have also downloaded and run the pointsto example provided
> in the
> >    Jedd example folder.
> >
> >    I need to extend the default analysis to include some more
> information,
> >    for example  dealing with thread accesses, explicit program point
> >    information & perhaps augment handling of dynamic classes.  I have
> been
> >    looking at the points-to analysis which Paddle provides within the
> Soot
> >    framework, and I would like to be able to develop my analysis using
> this
> >    Paddle setup within Eclipse.  The reason I want to use the
> >    Paddle/Soot/Eclipse setup is because I can easily attain
> jimple/shimple
> >    from source files and then pass these to the analysis.  Whereas if I
> >    develop with  Jedd I will have to use a synopsis file similar to
> >    pag.small?
> >
> >    Do I have to develop my analysis using Jedd as a standalone tool and
> then
> >    incorporate this in to Soot/Paddle?
> >
> >    Please let me know if I am way off the mark here, I have read Ondo/ej
> MSc
> >    & PhD and the other papers detailing the pointer analysis framework
> but I
> >    am still relatively new to this field.
> >
> >    Thanks,
> >    Alan
>
> > _______________________________________________
> > Soot-list mailing list
> > Soot-list at sable.mcgill.ca
> > http://mailman.cs.mcgill.ca/mailman/listinfo/soot-list
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.CS.McGill.CA/pipermail/soot-list/attachments/20060904/c9a784c6/attachment.htm