Re: Embedded Keynote Speaker Mentions Ada

[Cesar Rabak <crabak@acm.org>]

Warren W. Gay VE3WWG escreveu:
> Benjamin Ketcham wrote:
> 
>> Cesar Rabak <crabak@acm.org> wrote:
>>
>>> Even in a non commercial arena, why can't we convince Open Source
>>> enthusiats to write the system software in Ada and start to make a
>>> difference in this realm (my perception of their account on bugs is that
>>> they have a better response time, not intrinsical better designed in
>>> security)?
>>
>>
>> Well, this is a very good point.  I would suggest that a way to
>> make Ada much more popular, would be to recode the Linux kernel
>> in Ada.  
> 
> 
> I think your heart is in the right place, but I would suggest
> that an Ada clone of Linux makes it a wanna-be. In some senses
> it is probably true that Linux was a UNIX-wannabe, though Linus
> admitted that it was easier to use existing standards (POSIX
> for example) than to develop new one(s).

I've to agree with this point. The last thing we need a failing attempt 
to get at the market due an unsurmountable amount of work necessary 
before something sees the light of the day.

We can make paralells with Hurd, for example.

> 
> I would suggest that an Ada O/S with the same principles learned
> from UNIX et al. would be a good thing, but let's not do
> a "UNIX version in Ada" or a "Linux version in Ada".

Perhaps if we start small and create a kernel for appliance servers, 
which now have Linux or BSD as OS, and then progress onwards?

> 
> Radically new O/S research is welcomed also but the danger
> there is that something radically different is not likely to get off
> the ground. So I would suggest that a practical Ada O/S with
> "normal features" could be developed without the need to
> follow a standard.

Yes. Again, try to grab some kind of 'market' where still everyone is 
not the incumbent. Displace a 'winning' technology would require lots of 
energy.

> 
> Maybe this too is asking too much because it requires
> the development of both a new O/S and a new "standard"
> (and obviously a big porting headache for X11 etc).  Yet
> it would be real nice to have a natural Ada API base
> on which C programs had to adapt. ;_)

Yes! Besides even taking spite out of the equation ;-) the more an 
environment would be a better fit for Ada, the easier would be to have 
people to come to this language.

> 
>> How long does someone who actually knows Ada (not me) think it
>> would take to translate the kernel to Ada?  
> 
> 
> I think this process is the wrong way to put something into
> Ada. Ada can be made to look like a C program, but I don't
> see any benefit to that.

I cannot second this more.

> 
> Only by redesign, in Ada terms (not C ones), does such a
> project have any useful outcome. For this you toss away
> the C code, and design from the requirements.  And what
> I am suggesting is that you don't need the Linux kernel
> to dictate your requirements.
> 

I date to say we should be a little less entuasistic (or say more 
humble) and try to get a more small piece, like a software like bind or 
another (server) deamon and re-engineer it in Ada and start to quench 
these awful stories of 'vulnerabilities' and 'buffer overflow' problems.

--
Cesar Rabak