From: Rob Economopoulos
To: Jurgen Vinju
Date: Wed, 27 Feb 2008 16:27:42 +0100
Subject: [Fwd: PEM Announcement: "Improving the performance of SGLR"]



-------- Original Message --------
Subject: 	PEM Announcement: "Improving the performance of SGLR"
Date: 	Wed, 27 Feb 2008 16:01:43 +0100
From: 	Rob Economopoulos
To: 	pem-list



Dear all,

I hereby cordially invite you to a Programming Environment Meeting 
(PEM). I will be presenting the research I have done to improve the 
performance of SGLR, the Scannerless Generalized LR parser used in the 
Meta-Environment.

The talk is on Thursday 28 (tomorrow) at 13:00, in room M280.

Abstract:

The RNGLR parsing algorithm provides an elegant correction to Tomita's 
GLR method of parsing. In comparison to Tomita and Farshi's approaches, 
the algorithmic improvements and optimizations of the RNGLR algorithm 
significantly reduce the number of operations performed during parsing. 
Unfortunately few tools take advantage of these improvements and still 
base their parsers on versions of Tomita or Farshi's algorithms.

In this presentation we review the runtime performance of SGLR, the 
Farshi based parser used in the Meta-Environment, and discuss the 
results of incorporating the RNGLR algorithm into SGLR. Early results 
indicate that although parsing time can be improved by 20% in some 
cases, the overall runtime of SGLR is usually only improved by 5%.

Cheers,
Rob


Paul Klint wrote:
> [Time and location details: 10:00@M376]
>
> Dear all,
>
> I hereby cordially invite you to a Programming Environment Meeting 
> (PEM). This time Jens Calame will be presenting work on
>     "BAiT: Behavior-adaptation in Testing".
> The talk will be on Thursday 22nd November, 10:00. Location details
> will be announced shortly.
>
> Title: "BAiT: Behavior-adaptation in Testing"
> Abstract:
>
> Conformance testing is one of the most rigorous and well-developed
> testing techniques. Model-based test generation is an essential part of
> the conformance testing approach. The main problem in this phase is the
> explosion of the number of test cases, often caused by large or infinite
> data domains for input and output data. In order to solve this problem,
> we proposed a test generation framework based on the use of data
> abstraction and constraint solving to suppress the number of test cases.
>
> In this presentation, we will give an overview on the test generation
> and execution process BAiT, which is based on this test generation
> framework. We will then show the practical use of the BAiT test
> execution unit with an example and describe the tool's internal
> structure and possibilities to extend it by plugging in custom
> algorithms for test trace and test data selection. Finally, we will give
> an insight into the ongoing work on a case study (test of the Mozilla
> Firefox CSS Box Model).
>
>
> Hope to see you all there!
>
> Cheers,
>
> Tijs van der Storm
>
>
>
>
>

-- 
Giorgios Robert Economopoulos
Department of Software Engineering
CWI
Kruislaan 413
P.O. Box 94079
1090 GB Amsterdam
The Netherlands

Phone: +31 (0)20 592 4102
Fax:   +31 (0)20 592 4199

http://homepages.cwi.nl/~economop