IRTAW 11
8 12 April 2002

Mount Tremblant, PQ

CANADA

Agenda

8 April 2002

16:00 Van from Dorval Airport to Club Tremblant
18:00 Van from Dorval Airport to Club Tremblant
20:00  Dinner at Mt. Tremblant

20:30 Late Van from Dorval Airport to Club Tremblant

Please contact Stephen Michell (stephen.michell@NRC.CA) to arrange for transportation to Club Tremblant

9 April 2002 

07:00 08:30 Breakfast

08:30 12:00 
Status of the Standard & Open AIs & User Experiences

Alan Burns Session Chair

This session will begin with a short report from Jim Moore, convener of WG 9 -- the ISO working group on Ada, on the status of the standard.The purpose of this presentation is to provide us with an understanding of what is happening with the language so that we have the appropriate context for use in the consideration ofthe language changes we may propose.

Following this presentation the session will review the other proposed changes to the language that are being consideredincluding:

AI-249 Ravenscar Profile for High-Integrity Systems

AI-265 - Partition Elaboration Policy for High-Integrity Systems

AI-250 Extensibleprotected types

AI-264 Exceptions as types

AI-266 - Task Termination procedure

The group will be asked to evaluate these changes, make recommendations for modifications, and make recommendations for approval of the changes.

As part of this session, the User Experiences related to the Ravenscar Profile will be considered.

Papers to be considered during this session include:

User Experiences with the Aonix ObjectAda RAVEN Ravenscar Profile Implementation

Experience Report on the Implementation of Extensible Protected Types for GNAT

Practical Implementations of Embedded Software Using the Ravenscar Profile

12:00 16:00 Lunch

16:00 - 2:00 
Status of the Standard & Open AIs & User Experiences -- continued

Alan Burns Session Chair

From a report given at the European Space Agency -- Mostly the reaction to the Ravenscar Profile definition was very favourable, but there were two important messages from the evaluators that I thought were important to share with our group:

(a) The lack of support for a timeout on wait (i.e. timed PO entry calls) was seen to be a major obstacle to fault tolerance.The "workaround" using some sort monitoring task that handles all timing requests was considered to be hard to program and carries high overhead. So there was quite a strong request for this feature to be supported directly by the runtime.

(b) There are several other standard real-time paradigms that violate the Ravenscar Profile restrictions in their "natural" coding, e.g. need multiple entries and/or entry queues, more complex barriers for things like multiple readers/writers for a bounded buffer, event queues, pulses etc.

The view was that it would be crazy for everyone to have to implement their own version of these RT primitives.Instead, at the very least, some standard design patterns should be included in the Rationale document to show how to circumvent the restrictions. This could then lead to implementers providing some standard libraries (outside of the Kernel) that implement these patterns to simplify use of the Ravenscar Profile.

20:00 Dinner

10 April 2002

07:00 08:30 Breakfast

08:30 12:00
Safety Improvements for Consideration

Juan de la Puente Session Chair

The purpose of this session is to open discussions on issues related to safety and what needs to be done to address these issues in the language.

As part of this session, the following topics and representatives are available for presentation and consideration:

Experiences with certifying VxWorks G. Romanski

Software Related Accidents K. Lundqvist

Usingpartitions for space-based security Bruce Lewis

Papers to be considered during this session include:

Software Portability Gains Realized With MetaH and Ada95

12:00 16:00 Lunch Break

16:00 - 20:00
Real-Time, Fault Tolerance & Distribution

Michael Gonzalez Harbour Session Chair

The purpose of this session is to open discussions on issues related to real-time, fault tolerance & distribution and what needs to be done to address these issues in the language.

Papers to be considered during this session include:

Using Ravenscar to Support Fault Tolerant Real-TimeApplications

Precise Response Time Analysis for Ravenscar Kernels

Modeling and Schedulability Analysis in the Development of Real-Time Distributed Ada Systems

20:00 -- Dinner

11 April 2002

07:00 08:30 Breakfast

08:30 12:00
Scheduling / Modeling & Analysis / Language Changes

Andy Wellings Session Chair

The purpose of this session is to review work that has been done in the area of scheduling, modeling and analysis, andhardware modifications to determine what may be recommended for changes & additions to the language.

Papers to be considered during this session include:

Modeling and Schedulability Analysis in the Development of Real-Time Distributed Ada Systems

Protected Ceiling Changes

Accessing Delay Queues

Application Defined Scheduling in Ada

Language Issues of Compiling Ada to Hardware

12:00 16:00 Lunch

16:00 18:30

Scheduling / Modeling & Analysis / Language Changes continued

18:30 Load Vans for Dinner Outing

19:30  Dinner at Sugarbush

12 April 2002

07:00 08:30 Breakfast

08:30 11:00 
Analysis of the Real-Time Specification for Java 

(from the Real-Time for Java Expert Group)

Ben Brosgol Session Chair
The aim of this session is to identify areas in which the experience with real-time Ada might influence the RT Java specification, review how well that specificationmeets the requirements of real-time applications on which we have Ada experience, and review how Ada needs to interact with the emerging Java technology. 

As a reference, please read relevant parts of the Java realtime specification . This is the HTML version which has links to

multiple parts. There is also a PDF version available from www.rtj.org

11:00 12:00  Meeting Summary Joyce Tokar

12:00 13:30 Lunch / vans depart for Dorval