Re: Heartbleed

[Alan Browne <alan.browne@FreelunchVideotron.ca>]

On 2014.04.12, 18:01 , Yannick Duchêne (Hibou57) wrote:
> Le Sat, 12 Apr 2014 06:58:29 +0200, Shark8 <OneWingedShark@gmail.com> a
> écrit:
>
>>> BTW, once you have people exposed to type safe programming, then maybe
>>> you can introduce them to Ada for the large projects. One of the major
>>> revelations for me over Heartbleed was seeing people discuss the need
>>> for a safer language and immediately jump to languages like Java.
>>
>> Hm, good point.
>> Ada has some *REALLY* good features when it comes to
>> programming-in-the-large -- the YF-22 integration is astounding:
>> 12 major avionics subsystems, across 650 Ada modules containing
>> millions of lines of code, coded in 8 geographically distinct
>> locations, took *three days!*
>>
>> Source: http://archive.adaic.com/docs/present/engle/comments/tsld033.htm
>
> While Ada certainly played an important role in this success story,
> another as much important role was certainly humans… how did they
> communicate between each others? I mean, what documentation techniques
> and systems was used? Was the style guide(s) homogeneous or
> heterogeneous? Did they used some kind of maps to handle the big lines
> before diving into the details?
>
> I would enjoy a lot to know the answers to these questions.

Ada is an element to the described success.  Before a line of code is 
written, requirements and specs must be written and ICD's agree'd upon; 
many, many, many details need to be agreed to before software code 
design begins.

Also, databuses have evolved tremendously - reducing the wiring and 
eliminating often bizarre timings on board aircraft/weapons systems and 
making everyone use similar data types and structures throughout.

Further, back in the 80's, the internet and e-mail were not yet common 
as a means to sharing requirements, specs, test plans and ICD's.   Now 
such can be edited by teams in realtime w/o worrying about which copy of 
the Word file is up to date - everyone sees one single evolving copy no 
matter where and when they are.

(I recall our one of nav systems integrated on a USAF aircraft where it 
had to output waypoint and position data in different formats to 
different systems over the same databus (using different A-429 labels). 
  But in one case two destination systems used the same data label but 
different representations of position (semi-circles for one of them and 
radians for the other IIRC).  Ugly solution: the LSBit of each lat/long 
pair (separate labels) was set to a 0 or 1 to distinguish which format. 
  The target systems would use that 'data' bit to accept or ignore the 
data. (The 2 bit sign matrix having been used (wrongly) to signal some 
other state)).

It only took a conference call to make the ICD change.  But down the 
pike, when our system and their systems were integrated into other 
aircraft, we and they had to change code again and again and again for 
each new integration.

Imagine the above for many dozen parameters on similar data buses ...

-- 
"Big data can reduce anything to a single number,
  but you shouldn’t be fooled by the appearance of exactitude."
      -Gary Marcus and Ernest Davis, NYT, 2014.04.07