From mboxrd@z Thu Jan  1 00:00:00 1970
X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on polar.synack.me
X-Spam-Level: 
X-Spam-Status: No, score=-1.9 required=5.0 tests=BAYES_00,XPRIO autolearn=ham
	autolearn_force=no version=3.4.4
X-Google-Language: ENGLISH,ASCII-7-bit
X-Google-Thread: 103376,a5886c32d2913c1,start
X-Google-Attributes: gid103376,public
X-Google-ArrivalTime: 2004-03-25 03:07:09 PST
Path: 
 archiver1.google.com!news1.google.com!news.glorb.com!news.cs.univ-paris8.fr!proxad.net!134.158.69.22.MISMATCH!in2p3.fr!irazu.switch.ch!switch.ch!newsfeed.sunrise.ch!news.sunrise.ch!not-for-mail
From: "Roger Blum" <roger_blum@swissonline.ch>
Newsgroups: comp.lang.ada
Subject: Gnat Glade newbie questions
Date: Thu, 25 Mar 2004 12:07:05 +0100
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2800.1158
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165
Message-ID: <4062bd5b$0$706$5402220f@news.sunrise.ch>
NNTP-Posting-Host: 62.167.215.170
X-Trace: 1080212827 news.sunrise.ch 706 62.167.215.170
Xref: archiver1.google.com comp.lang.ada:6502
Date: 2004-03-25T12:07:05+01:00
List-Id: <comp.lang.ada>

Hello

I'm new to gnat glade and im trying to build a framework for distributed
Monte-Carlo simulations.
Input (scenario and base data) comes from an Oracle database; output is
written into the same DB.
Statistical analysis of simulation results is then performed in the DB
itself.
I have a working system which is based on strictly sequential processing.

Now I intend to distribute the processing as follows:
-----------------------------------------------------
- Kind of 'shared memory' for base data (e.g. digital elevation model for
line of sight calculations)
- A worker process executing one independent 'run' of the whole simulation
- A master process spawning the desired amount of worker processes (runs)

My idea:
--------
- The base data:
  One process (an executable running as a service / an Ada task with
different entries accessing a
  protected object) acting as a server. Available to any worker process.
  Package categorized as REMOTE_CALL_INTERFACE (or SHARED_PASSIVE?).
- The worker process:
  Another executable
  Packages reading data from base data categorized as REMOTE_CALL_INTERFACE
or no categorization
  at all?
- The master:
  Starting worker processes with
GNAT.OS_LIB.NON_BLOCKING_SPAWN("worker.exe", args) as just creating
  Ada tasks wouldn't be enough to be migrated on openMosix (the target
platform).
  Probably no categorization necessary.

Now my questions:
-----------------
- All my tries to build and use GNAT Glade 3.15p from the source failed.
  Has anybody a working binary of GNAT Glade 3.15p for Windows and/or Linux
  (gnatdist and Garlic) or how can I build it from source without a lot of
error messages?

- Can gnatdist build all the necessary executables just on the base of
correct categorization
  of the packages or do I have to create 3 main procedures/packages?

- How can the worker get aware of the presence of the base data?

- Will the distributed processes generated that way migrate on openMosix?

- Has anybody experiences made with Ada / MPI or PVM?

Any help / information / suggestion is very much appreciated!

Thanks
Roger