[cap-talk] Failure isolation

[Pierre THIERRY]

Scribit Raoul Duke dies 17/03/2008 hora 16:45:
> i agree it is sad when a great idea is apparently difficult to achieve

I don't think it's difficult. Frankly, it didn't take me that long to
begin to think in terms of confined objects each time I design a system.
As for many other design disciplines, it even makes things easier to
reason about. The real problem is that POLA is a discipline, and as for
any other discipline, it's very tempting to avoid its burden, when
you're not really convinced by its virtue.

Many engineering domains have a stronger tradition to enforce
discipline, probably because people are used to some basic quality. You
don't expect your car or refrigerator to stop working randomly, and you
wouldn't accept if a bridge could collapse even in harsh conditions.

But people are now perfectly aware that computers are *meant* to be
unreliable and unpredictable. They rant about it, but accept the
situation as a fact.

Maybe creating a feedback loop is one key to the solution: create
secure, reliable and more predictable systems, which will make people
expect more of them, which will lead to more of them being made, and so
on...

Hoepfully,
Pierre
-- 
nowhere.man at levallois.eu.org
OpenPGP 0xD9D50D8A
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://www.eros-os.org/pipermail/cap-talk/attachments/20080318/34185a97/attachment.bin