Neal H Walfield <[EMAIL PROTECTED]> writes:
> Messages do _not_ have to go to the same machine; they are sent
> to a port and the Microkernel worries about getting the messge
> to the receiver.
I thought Mach wasn't supposed to transmit messages over the
network by itself; rather, each end would have a proxy which
converts messages to network packets and vice versa. Tasks
communicating with the proxies would think they are the real
servers/clients and not know they actually forward messages.
Of course, I may have misunderstood this.
> For instance, one could,
> theoretically, have a single auth server on a hurd cluster.
Right. (Would that be a "collective" as mentioned in the Hurd manual?)
> It is very general; the whole hurd philosophy is based on putting
> everything into user space servers. Imagine accessing a disk... then
> try doing a ls -R or something similar. Slow!
Actually, file system servers access Mach devices directly if
possible. They get the device write port from the storeio
translator at /dev/WHATEVER with file_get_storage_info.
> Every time a new block needs to be read from disk, libc will need to
> contact the ext2fs translator which is done via messages and, therefore,
> implies some authentication.
Merely permission checking, which is done internally by the
translator and doesn't involve other servers. Real
authentication is done before the file is opened.
> No, but we are on help-hurd and not debian-hurd where most of the `talk'
> takes place.
As we should, since this thread is not Debian-specific at all.
