[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Supporting POSIX *users*

From: Jonathan S. Shapiro
Subject: Re: Supporting POSIX *users*
Date: Tue, 01 Nov 2005 14:13:40 -0500

On Tue, 2005-11-01 at 17:48 +0100, Alfred M. Szmidt wrote:
>    Today it is the fault of the system architect, whose crappy system
>    design made this inevitable. It cannot be the user's fault, since
>    they had no authority or ability to alter circumstances.
>    Responsibility follows from authority.
> No, it is the fault of the user.  Take a capability based system, I
> give all programs the same capabilities, so it works like it does on a
> normal system, who is at fault?  The system architect for giving the
> user the ability to set the capabilities?

Absolutely. There is no conceivable justification for this bad design,
and no functional requirement for it.

>    What happens to your analogy when (a) there is exists only bad oil
>    in the world, (b) you need to get your kid to a hospital, and (c)
>    the auto vendor has designed an automobile that requires perfect
>    oil but converts better oil into bad oil before using it? Yes, in
>    the face of point (c) I would blame the car manufacturer, because
>    they have made it impossible for me to act sensibly. The design is
>    defective.
> If you put artifical restrictions like that, then you can make
> anything work as you want.  A is false, since there exists `perfect'
> oil in the world (C is also false for the record) so your whole
> version of my analogy is false.

Alfred, I did not introduce these restrictions. They (fairly) faithfully
capture the current state of commodity software.


reply via email to

[Prev in Thread] Current Thread [Next in Thread]