Re: Web Development Using Ada?

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

"Jeffrey Carter" <spam.jrcarter.not@spam.not.acm.org> wrote in message 
news:kt119m$hln$2@dont-email.me...
> On 07/27/2013 03:35 AM, Aay Jay Chan wrote:
>>
>> What are the pros an cons of using Ada in web development? Is it feasible 
>> /
>> practical? What would be your recommendation?
>
> There are a number of sites out there with "Ada inside". One advantage of 
> using Ada is that many of the common vulnerabilities don't occur with Ada. 
> Brukardt discussed the Ada behind archive.adaic.com here in c.l.a several 
> years ago. In addition to obvious things such as no buffer overflows, he 
> mentioned that an attacker can get a server such as Apache to run a shell, 
> which the attacker can exploit, while the Ada equivalent uses a 
> non-exploitable Ada task to accomplish the same end. (I'm writing from 
> memory here.)

I'd be hesitant to call anything "non-exploitable", but there is much less 
attack surface with Ada. The vast majority of things that can go wrong raise 
an exception, which is logged in the normal web logs (and thus can be 
investigated at some future date) with the request being failed. Nothing bad 
will happen in this case, because it's all normal Ada semantics.

OTOH, if you execute a shell, if an attacker can find a way to pass 
information to that shell, they might be able to do anything. Apache has 
fixed many such bugs. It's better if there are no shell outs. It's even 
better if the capability to do shell outs isn't even in the code (since some 
attacks require executing existing code in unusual ways - if the process 
doesn't have any code that can shell out, such attacks can't shell out 
either).

Of course, you're still vulerable to problems from the host OS and from any 
interfacing that you have to do. And of course, from any compiler bugs. 
(Which is why I never claim that Ada is "non-exploitable".)

                          Randy.