Can any one shed some light on this problem.

Can any one shed some light on this problem.

[simon broadhead]

I am doing a research assessment at the university of Glamorgan on
Ada. I have been asked to re-examine the language and look at how
effective ADA would be in creating a microkernel, and to outline the
features which are lacking in Ada for such a project and also those
features which neatly map a microkernel.
I do not have much prograing skills so I am a bit lost on this. 
 

I have bean asked to examine the synchronization mechanism
(rendez-vous) which is promoted by ADA and contrast this to the
Dijkstra's (p/v) wait and signal mechanism.

 

If you have any thought on is mater I would be a grateful for your
help also if you can guide me to any other resources that could be
useful in my search it would be greatly appreciated. you can email me
at sabroadhead@yahoo.com

Re: Can any one shed some light on this problem.

[chris.danx]

"simon broadhead" <sabroadhead@yahoo.com> wrote in message
news:f076c667.0108030252.591a1812@posting.google.com...
> I am doing a research assessment at the university of Glamorgan on
> Ada. I have been asked to re-examine the language and look at how
> effective ADA would be in creating a microkernel, and to outline the
            ^
            this is the American Disabilities Association not Ada,
            I know it's being pernickity but ppl round here do
            tend to get annoyed about this.

> features which are lacking in Ada for such a project and also those
> features which neatly map a microkernel.
> I do not have much prograing skills so I am a bit lost on this.
>

Well I've not considered the method by which I'll be structuring my
micro-kernel in Ada, and it depends on the nature of scheduling;
preemptive, cooperative or fully preemptive.  An idea may be the use of
protected objects for kernel modules but this may only be appropriate for
cooperative/preemptive but not fully preemptive (a kernel mode operation
would complete with the use of protected objects and not be halted, upon a
given condition so kernel operations would not be fully preemtive.  I'm
still trying to get my head around the idea of full preemption so maybe I'm
wrong about this).  I know that for interrupts (they might not be in the
kernel proper but they are part of the OS), the likely choice is a
protected object.

> I have bean asked to examine the synchronization mechanism
> (rendez-vous) which is promoted by ADA and contrast this to the
> Dijkstra's (p/v) wait and signal mechanism.

I'm afraid I can't say much on this, someone else might be able to help.

Re: Can any one shed some light on this problem.

[MCL]

"chris.danx" <chris.danx@ntlworld.com> writes:


> > I have bean asked to examine the synchronization mechanism
> > (rendez-vous) which is promoted by ADA and contrast this to the
> > Dijkstra's (p/v) wait and signal mechanism.
> 
> I'm afraid I can't say much on this, someone else might be able to help.

    With a protected object you can *very easily* code a semaphore.
Probably only of interest for educational purposes and toy programs.

-- 
          ||            Manuel Carro -- DLSIIS            ||
          ||           e-mail: mcarro@fi.upm.es           ||
          ||      http://lml.ls.fi.upm.es/~boris          ||
          ||    http://clip.dia.fi.upm.es/Software/Ciao   ||
          || Phone: +34 91 336-7455  FAX: +34 91 336-7412 ||

Re: Can any one shed some light on this problem.

[Jeff Creem]

----- Original Message -----
>
> > > I have bean asked to examine the synchronization mechanism
> > > (rendez-vous) which is promoted by ADA and contrast this to the
> > > Dijkstra's (p/v) wait and signal mechanism.
> >
> > I'm afraid I can't say much on this, someone else might be able to help.
>
>     With a protected object you can *very easily* code a semaphore.
> Probably only of interest for educational purposes and toy programs.
>

Actually using a protected object to code a semaphore is also very useful
for creating portable semaphore packages to replace vendor specific
packages.

In Ada 83 it was somewhat common for people to use some sort of semaphore
package (either vendor specific or tied to the OS services) for signaling
and/or
mutual exclusion since the only built in language  feature for this was the
rendevous. I know of a
150K+ line of code Ada 83 project that was originally built for vxWorks with
a project
specific semi-thick binding on top of vxWorks semaphores that was ported to
run as
a simulation on an SGI (should work anywhere now) by replacing the body of
the semaphore
package with a protected object based implementation.

Re: Can any one shed some light on this problem.

[Wilhelm Spickermann]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset=us-ascii, Size: 957 bytes --]


On 03-Aug-01 chris.danx wrote:
> 
> "simon broadhead" <sabroadhead@yahoo.com> wrote in message
...
>> I have bean asked to examine the synchronization mechanism
>> (rendez-vous) which is promoted by ADA and contrast this to the
>> Dijkstra's (p/v) wait and signal mechanism.

Read "Concepts and Notations for Concurrent Programming" from Gregory R. Andrews
and Fred B. Schneider (Computing Surveys, Vol 15, No. 1, March 1983). It�s a
very good and condensed overview of synchronization problems and the different
approaches used in several programming languages and proposals to solve these.

The article is from 1983 and contains only information about the old Ada
standard. Ada today has the same rendezvous concept as in 83 -- but there is
an additional synchronization concept in the present standard: the Protected
Type. (It fits into the article at the place where Hoare�s "conditional wait"
and "Kessels� proposal" are discussed.)

Wilhelm

Re: Can any one shed some light on this problem.

[tmoran]

and read "Concurrency in Ada", Burns & Welling, Cambridge U Press