Re: seL4 as base of an AdaOS with some Spark proofing?

["Randy Brukardt" <randy@rrsoftware.com>]

<gvdschoot@gmail.com> wrote in message 
news:485a93a1-8920-4139-8563-591dba79d68d@googlegroups.com...
On Friday, August 1, 2014 9:42:03 AM UTC+2, kug...@web.de wrote:
...
>> > Here is another idea. Let's for instance instead focus on tools that 
>> > can automatically translate C source code to Ada.
>>
>> What I know about C, you won't and I'm very sure you can't translate it 
>> to Ada in an automatic way.

Of course you can translate C to Ada. There were various tools to do that 
back in the 1980s. The problem is that the Ada you get is terrible Ada (the 
one I saw was full of address manipulations). (That's true of any automatic 
language translator, BTW.)

>Just because it hasn't been done before doesn't mean it is impossible. For 
>instance the
> Go (golang) team in the person of Russ Cox is working on c2go and the 
> approach he
> takes is unconventional. There is presentation about this on youtube.

There's no problem doing such a translator (it's relatively easy to do). The 
problem is that the result is terrible Ada (little typing, lots of address 
math, etc.). In order to make *good* Ada, you need a redesign. That's also 
the case if you want to use SPARK or even the Ada 2012 contracts -- a 
translator can't create intentions that aren't in the source code. There's a 
reason such translators today stop at thin bindings -- that's because there 
is no point in going further without a redesign (and thin bindings ought to 
always be wrapped in a thick binding for Ada use).

                                  Randy.