Re: utime()

[Stephen Usher <steve@earth.ox.ac.uk>]

>
>> 
>> Hi Julian,
>> I agree such a function should be added.
>> Maybe it should be a system call, though.
>
>That would be nicer, yes. But I don't think that this is the best moment
>to make changes in the kernel and in the interface between file system
>drivers and the kernel... (look at the version number :-).
>
>> Another wish of mine: a Ftruncate() call.
>> Also possible as an Fcntl() command, of course.
>> 
>
>Indeed. Let's define yet another standard opcode for Fcntl!

Why not make everything which is device dependent a Fcntl op code... not all
devices would need to support them, the default would be to return EINVAL.
If only all device drivers (including the standard TOS filesystem, serial
port support etc etc) were loaded at boot time, then the kernel would not
have to be changed just to fix, say, the tty driver. The kernel would only
use the ROM tosfs file system and hard disk driver at boot time. The
loadable tosfs driver could of course use the default routines if it were
written that way, but it wouldn't need to.

The whole upshot of this would be though that the kernel could be made
totally stable and relatively static whilst the device drivers could evolve.

Just think...  multi-threaded interrupt driven hard disk driver with nice
multi-threaded filesystem drivers over the top.. a fixed tty driver which
allows you to do a select on things other than the console.. a serial driver
which can handle speeds up to 57.5Kbps on the new serial ports... :-)

>
>
>-- 
>________________ cut here _________________________
>Julian F. Reschke, Hensenstr. 142, D-W4400 Muenster
>  eMail: julian@math.uni-muenster.de, jr@ms.maus.de
>________ correct me if I'm wrong __________________
>

Steve

-- 
---------------------------------------------------------------------------
Computer Systems Administrator, Dept. of Earth Sciences, Oxford University.
E-Mail: steve@uk.ac.ox.earth. Tel: Oxford (0865) 282110.