Bitching about what to do on error returns :-) (was Re: Trojan Horses)

[Boyd Roberts]

In article <8536:Nov600:37:2790 at kramden.acf.nyu.edu> brnstnd at kramden.acf.nyu.edu (Dan Bernstein) writes:
>
>You say that you *must* inform the user of an error. Have you ever seen
>the quota system in action? Did you notice that the kernel blares some
>obnoxious messages at the user to indicate quota-related write errors?
>

No, I've not seen the quota system in action.  From the documentation and
the bizarre `close() returning EDQUOT' scenario it sounds like a real crock.

We had a per-user limits system running at basser.  close() _never_ complained
about exceeding your disk limit.  When you wrote you got the error.  There were
also uprints() messages from the kernel, but you could turn them off (which
everyone did because exceeding your soft limit was not really a problem,
although the message was).

Look Dan, in the general case you should report all errors.  However, in some
cases it is not clear cut and requires a judgement call.  Note I said `some'.


Boyd Roberts			boyd at necisa.ho.necisa.oz.au

``When the going gets wierd, the weird turn pro...''