Re: [RFD] New driver system (was Adding new syscalls via modules)

Alexander_Maryanchick%STORACTIVE@storactive.com
Thu, 8 Jul 1999 14:36:10 +0400


>> Hooks and filters are innocent.
> Why are you so sure ?
I am 99% sure.
I wrote many drivers for windows and know, why they crashes :-)
For example, in NT you can safely filter OPENs.
But safe filtering CLOSEs (I skip the details) is not documented.
These situations (I have MANY examples) force developers to resort to
hacks.
I don't know any other drivers-related problems.

>> M$ problem is opacity and ugly initial design.
> And hooks will add opacity to mix really fast :-((
An unexpected idea %-/. Are you sure?
Hooks must be transparent by definition.
M$ hooks are opaque due to compatibility problems and dishonest
competition.

> And it's GREAT ! If feature is REALLY need it will be added eventually.
Cool :-)

> But if not then chance is high that it will not be added at all and we'll
> not have this ugly compatibility problems.
What compatibility problems?

> BTW if idea requires "fixing the
> whole kernel" then it's hardly possible to implement it via separate
module
> anyway. Hooks or no hooks.
experiment 1: Add new operator to the shell. (meaning 'hooks' - extensible)
Copy one file to /bin
experiment 2: Add new operator to VB. (meaning 'no hooks' - not extensible)
Err...
Feel the difference.

> And IMHO easy way to add new features will make it much uglier fast...
The hard to add new features system is M$-DOS.
Hmm... Hack's paradise.
The most extensible system is UNIX.
30 years alive - thousands of new features.

> I'm interested in OS to solve MY problems.
Who is not? ;-)

Best regards.

Alexander.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/