sure, either's ok. Far as I know anyhow. I had flu Friday and still
have a number of aches, but am able to be in. If I get plenty of
sleep (which I intend to!) at night hopefully it'll get better.

There's one further bit that's needed in the iosubnpag stuff you saw
by the way...I'll show it to ya when you come by. I've had it ready
to check in for some time now as it was just an oversight. I'm
using it for multipath failover.

I had some questions about the snappy disk too...mainly had to do with]
the details of the meta filesystem you'll have underneath to manage
all the storage as more snapshots are done. Wot the hell...maybe I
can show you the working undelete I already have for VMS (implemented
it before coming to DEC, but I have a binary kit on the workstation
here...no reason I shouldn't be able to use my own stuff after all...)

Very little is needed, but status does need to be kept around. I've
been trying (so far no success) to argue for adding a context stack
to the end of the IRP to keep such things around. I don't believe anyone
copies whole IRPs around anyhow so that I couldn't track what's going
on...keep a copy of the IRP address at the top to validate if nothing
else'd work...
The point is that I'd like to make it more natural to do streams type
I/O in VMS, making the multipath intercept simpler but also facilitating
everything else.

It would be delightful to meet you. You pick however which time you prefer.
Glenn Everhart
zko3 office 4T35
1-1497