Re: Re: Trusted OS...

[Bennett Todd <bet () rahul net>]

2000-03-31-13:41:37 Starkey, Kyle:
> I am with you on the fact that the TOS certification is a little
> much for the corporate standard, but I can not accept the fact
> that there is any other way to certify that the OS is truly
> tursted.  The OS and ALL of it subcomponents must be broken down
> and mathematically proven to adhere to the security structure
> that the OS was designed for.  Most of us have no time to read
> through 4000 pages of mathematical proofs, but to be a TOS you
> must be able to provide this document before I will accept that
> certification.

So far, I've not seen a certification I'd want --- i.e. I've not
seen one that has covered products that I'd want, or that would make
me want the covered products more.

Some certifications are so superficial or irrelevant that they don't
tell me anything I'd care about.

Others are so bulky, slow, and cumbersome that they are a complete
barrier to quick development, and so will never apply to anything
except really antiquated systems with depressingly archaeic designs
and a spartan lack of features.

I remain unconvinced that this gap will be filled in any time soon;
and it's between where the really spiffy engineering is happening.

As a specific for-instance: I trust the auditing done by the OpenBSD
folks more than I trust the code reviews performed by the TPEP. And
I'm _sure_ more inclined to use OpenBSD than any evaluated OS I know
of.

Of course I like Linux even better. Nothing like having a few orders
of magnitude more people working on the system to keep it in a
froth.

-Bennett

Attachment: _bin
Description: