4 Mar
2006
4 Mar
'06
8:55 a.m.
"Zane H. Healy" <healyzh at aracnet.com> wrote:
When using CDRECORD on OpenVMS aren't you writing
to a MSCP-SCSI
device? Granted it's a lot more modern, and the SCSI controllers are
probably more advanced.
In that case the driver is the DK: driver, which supports
the IO_$DIAGNOSE QIO operation to do raw SCSI stuff. (I think
the GK: and DQ: drivers support this after a fashion too... isn't
DQ: the IDE driver or something now? I remember when it was the
IDC RL02 on a 11/730!)
Under VMS lots of devices that are not really MSCP can be served
through MSCP-type layers (for example serving disks across
a cluster is done through a MSCP interface, even if it's not a
MSCP drive). In the case of CDRECORD
it's more direct. (And you'll note that you cannot use it cross-machine
between two machines in the same cluster...)
The whole MSCP concept is really pretty powerful and most OS's do
not really use it in its whole elegance and glory (or maybe
"colossal inconquerable tower" is more appropriate!) But in this
case CDRECORD goes through a lower level.
This layering exists in the PC-clone world too: most IDE CD writers
are served up by ATAPI, a SCSI-like layer "riding" on the IDE interface.
(And for a while even if you had no SCSI hardware, to record to an
IDE CD recorder in Linux you needed SCSI support in the kernel...)
Tim.