[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Points-to analysis package



On Mon, Jul 29, 2002 at 10:27:08AM -0400, Pierre-Luc Brunelle wrote:
> Hi,
> 
> I see on Soot's home page that you are planning to release in the near
> future a points-to analysis package. I have a few questions regarding
> this package:
> (1) What are its features?

Briefly, the package supports flow-insensitive, context-insensitive
points-to analyses. Most of the emphasis is on subset-based
(Andersen-style) analyses, but the package has been designed to
also efficiently support unification-based (Steensgaard-style)
analyses, and to make it easy to implement analyses that are
a mixture of the two.

The best current description of the package and its features is
described in a poster that was presented at the PLDI Student Research
Forum, and is available at
http://www.sable.mcgill.ca/publications/#poster-2002-1
If you have specific questions after reading this, please ask them on
this mailing list.

> (2) When will it be available?

SPARK will be a part of Soot starting with the nest release. We plan a
major release (1.3.0) at "the end of the summer", meaning end of August
or beginning of September. The original intent was to include SPARK in
a version 1.2.4 that was to have been released earlier, but since SPARK
is still unstable, and nobody expressed an immediate need for it, that
did not happen. If there is enough interest, we may still make such a
release.

The most important thing missing at this point is documentation, which
will be in the form of a master's thesis, expected to be completed
in October. I may make a preliminary version of it available with the
1.3.0 release for the purpose of documentation.

> (3) Until then, is it possible to get an unfinished version?

Depending on how much immediate interest is expressed on this mailing
list by others this week, we will decide whether to make a 1.2.4 release
containing the current version of SPARK. If we do not, I will send you a
copy of the current state by the end of the week.

Thank you for your interest.

Ondrej

> We plan to use this package to compute def-uses for testing purposes.
> 
> Regards,
> 
> Pierre-Luc Brunelle
> Ecole Polytechnique de Montreal
> 
>