*********************************************************************
*                       Ecole Normale Supe'rieure                   *
*                                                                   *
*                              Se'minaire                           *
*                SEMANTIQUE ET INTERPRETATION ABSTRAITE             *
*                              P. Cousot                            *
*                                                                   *
*                        Vendredi, 14h00--15h30                     *
*                 Salle W, 4e`me e'tage, toits du DMI               *
*                 DMI  ENS  45 rue d'Ulm  75005 Paris               *
*********************************************************************


*** Vendredi 2 avril 1999 **** 14h00 ********************************

Rela^che

*** Vendredi 9 avril 1999 **** 14h00 ********************************

Rela^che

*** Vendredi 16 avril 1999 **** 14h00 *******************************

Relache (Dagstuhl Seminar on Program Analysis, 12.-16. April 1999)

*** Vendredi 23 avril 1999 **** 14h00 *******************************

Towards a domain theory for abstract interpretation

Roberto GIACOBAZZI (Universita' di Verona)

Re'sum'e:
In this seminar we will give account on recent results concerning the
systematic design of domains for abstract interpretation. The foundation of
a theory of domains for abstract interpretation was fixed by Cousot and
Cousot in their POPL'79 paper and, in more recent years, this has become a
specific research field in abstract interpretation and static program
analysis community. Our aim is to organize some of these results in a sound
theory of abstract domains: A kind of domain theory for abstract
interpretation. This makes available a number of high-level facilities to
tune program analysis in accuracy and costs, by refining, combining,
compressing, simplifying, and decomposing domains in abstract
interpretation.  All these operations are interpreted in a unifying
framework, which is still based on the notion of abstract interpretation,
but now lifted one level up: with objects of abstraction being domains. A
clean and simple theory of abstract domains can be designed in this way,
where all the well known operations on domains can be interpreted as
instances of a unique simple pattern.

*** Vendredi 30 avril 1999 **** 14h00 *******************************

Escape analysis for Object Oriented Languages. Application to Java.

Bruno BLANCHET (INRIA & X)

Re'sume' :
Escape analysis is an analysis that determines whether the lifetime of
data exceeds its static scope.

The main originality of our escape analysis is that it determines
precisely the effect of assignments, which is necessary to apply it to
object oriented languages with promising results, whereas previous
work applied it to functional languages and were very imprecise on
assignments. Our implementation analyses the full Java Language.

We have applied our analysis to stack allocation and synchronization
elimination. On average, we manage to stack allocate 23% of data,
eliminate 68% of synchronizations and get from 10 to 41% speedup (14%
on average). We give a detailed experimental study of the causes of
the speedups: the effect on the garbage collector, on data
locality. The improvement comes more from the decrease of the GC and
allocation times than from improvements on data locality, contrary to
what happened for ML.

*********************************************************************
Pour recevoir l'annonce par courrier electronique: cousot@dmi.ens.fr
WWW: http://www.dmi.ens.fr/~cousot/annonceseminaire.html
*********************************************************************