[english version below]
Bonjour à tous,
La semaine prochaine, jeudi 6 octobre, nous accueillerons Ryan Wisnesky de l'Université de Harvard qui nous donnera un exposé intitulé A Monadic Query Language (voir résumé ci-dessous). L'exposé débutera à 11H en salle 1D23 et sera précédé, comme c'est la tradition, par un pot informel pour nous retrouver autour de thé, café et croissant.
Mais avant l'exposé de Ryan, je vous rappelle que nous accueillons
demain jeudi 29 septembre, toujours à 11H en salle 1D23, Nikos Tzevelekos (qu'il me pardonne d'avoir écorché son nom dans l'annonce de la semaine dernière) de l'Université d'Oxford qui nous parlera de
Game Semantics for Good General References (voir le résumé rappelé ci-dessous). Vous êtes les bienvenus à partir de 10H45.
A Monadic Query Language
Ryan Wisnesky
(Harvard University)
Jeudi 6 octobre 2011, 11h salle 1D23
Abstract: Recent interest in general-purpose "cloud-scale" or "internet-scale" data processing has led to the development of a new class of declarative "cloud-query languages" such as MapReduce, DryadLINQ, and CloudHaskell which go beyond the traditional relational/SQL model in expressive power. Although these languages vary in the kinds of queries and collections they support, it is well-known that large fragments of these languages can be formalized in a uniform way using monads (to model collections), comprehensions (to model queries), setoids over polynomial datatypes (to model data), and folds (to model computation). Despite the theoretical attractiveness of this approach, significant open problems have hindered the development of a standard model of cloud-query languages.
In this talk I describe MQL, a new query language and compiler designed to be just such a standard model. MQL's principled foundation gives it capabilities beyond those found in current cloud-query languages, such as the ability to re-write queries into a normal form using data integrity constraints, fuse folds using monad-specific information, and emit verification conditions at compile time. At a theoretical level, MQL embeds Tannen's Calculus of Collections and Aggregates into a system of higher-rank, qualified types and provides an equational theory for comprehensions over idempotent monad algebras suitable for further study of topics at the intersection of relational database theory and programming language theory.
Game Semantics for Good General References
Nikos Tzevelekos
University of Oxford
Jeudi 29 septembre à 11H en 1D23
Abstract: We introduce a new fully abstract and effectively presentable denotational model for RefML, a paradigmatic higher-order programming language combining call-by-value evaluation and general references in the style of ML. Our model is built using game semantics. In contrast to the previous effectively presentable model by Abramsky, Honda and McCusker [AHM98], it provides a faithful account of reference types, and the full abstraction result does not rely on the availability of spurious constructs of reference type (bad variables). This is the first denotational model of this kind, preceded only by the trace model recently proposed by Laird [Lai07].
The talk presents our paper of the same title which appeared in LICS earlier this year.
**********
**********
English Version
**********
**********
Hi everyone,
In next week seminar Ryan Wisnesky from Harvard University who will tell us about a monadic query language (see abstract above) at 11h on October 6 in Room 1d23.
Before that, we shall welcome tomorrow the 29th, Nikos Tzevelekos (may Nikos forgive me for mispelling his name in last week's announcement) from the University of Oxford who shall speak about a Game Semantics for Good General References
As usual, we'll have an informal pre-seminar coffe-tea social meeting about 15 minutes prior to the seminar: you are welcome to come at 10H45!
=======
More informations on PPS seminar at http://www.pps.jussieu.fr/seminaire.
Best regards,
Alexis
--
Alexis Saurin
