[oclug] On certifying Linux [was: How open nature of Linux can work against it.]

[Stephen M. Webb]

On 14/08/06 09:31, Spencer Cheng wrote:
> The fundamental problem that Linux (and Unices and Windows and...)
> runs into in terms of certification or validation is a problem that
> is only solvable as the industry matures. The root problem is NOT
> that the underlying implementation changes but that the semantics of
> the interfaces are changed unpredictably (regardless of the best
> intent of the S/W teams involved).

Respectfully, I must disagree.

Oh, I agree that this is a problem, all right.  But it's not THE problem under 
discussion.  The problem is not in fact about certification and validation 
(there are many easily inspected robust certification program available for 
Linux distributions).

The problem is in fact one of who do you sue when something goes wrong, and 
who provides indemnification.

The big problems are not technical -- those are all solvable.  The problem is 
not one of standards  -- there are many well-documented peer reviewed 
standards in in in Linux.  The problem is not one of verifiability and 
certifications -- those have been solved over and over.

The problem is one of lawyers and insurance adjusters.  It's not that these 
folks are bad people (all cynicism and humour aside).  It's just that when 
someone gets killed or worse yet injured, who is going to pay?  The faceless 
crowds behind Rinkidinx Linux, or the shareholders of MSFT with their 
billions in equity?

For the best insight into this entire issue, read the license conditions you 
have to agree to before using the Java programming language.

--smw