info-vax@ucbvax.ARPA (04/17/85)
From: *Hobbit* <AWalker@RUTGERS.ARPA> I've stumbled across lots of 'em since the last episode. Most of them involve third-party software, of course, but here are some DEC-only ones. Diskquota: The sequence USE <dev>, CREATE, REBUILD, REMOVE <the UIC's I didn't want included, including [1,4]> led to *no one* being able to write any files without EXQUOTA. SHOW QUOTA still insisted that there were no quota entries for the given UIC's that were having trouble. Why?? I see no subtleties in the Diskquota utility. REMOVE should perform as advertised; I suspect that it is not, and leaving rat turds around. Phone: It doesn't work at all. If I answer someone who is calling me, I get this trash about ''attempting to continue after a transmission error'', and more than likely an access violation. The calling end simply exits without a whimper. Or something like that. It's so f*cked up that I don't want to deal with it until someone tells me the stupid simple thing I haven't done, or I get a fix, or comfirmation that it is indeed screwed up beyond all belief. Will someone please put my mind at ease?? Mail: What the hell is MAIL/PROTOCOL all about? Why isn't it in the manual? Spawn: Why didn't they fix the cretinous stuff about associated mailboxes? $brdcst: Why does this *not* check for the existence of OPER privs? I have people with the usual two default privs using an old sender program with universal success, *without* my having installed it with OPER. What gives? General: Is there a way to *extract* what a CLD source looked like from DCLTABLES? *That* would be one way to find these undocumented commands... _H* [close to nuking Manyard] -------
info-vax@ucbvax.ARPA (04/17/85)
From: Mark Johnson <MHJohnson@HI-MULTICS.ARPA> Well, your problem with diskquota is not V4 specific. Since disk quotas have been available, the lack of disk quota entry implies NO QUOTA. From what I remember, you should modify the quota for [0,0] (?) to your system default before you do a REBUILD. Otherwise you get the (I assume hopelessly inadequate) default for every user. I enabled disk quotas ONLY to get a quick list of disk usage by UIC & set the numbers really big (100k per UIC on an RA81). Also, REBUILD creates quota entries only for users with files on that disk, it doesn't look at the UAF. About CLD sources, look at the source microfiche & have some poor student type all them in. I did that once for the V3 CLD files so I could create a restricted command environment. About undocumented stuff, examining the HELP library for commented sections and looking through the BLISS .REQ files in SYS$LIBRARY for the string "DOCUM" works really good for finding hints. There is an undocumented system service $GETQUI (Get Queue Info?) that you could play with. I assume that most of the undocumented stuff is there because someone coded it but couldn't document it in time for the release (not that I have ever done things like that...). About the other stuff, I can't say because I have run V4 only briefly due to lack of 3rd party software for it. I am going to skip V4 anyway, because my V4.1 release is already here. I need the disk fixes too much to be another guniea pig. --Mark
info-vax@ucbvax.ARPA (04/17/85)
From: LEICHTER <Leichter@YALE.ARPA>
I've stumbled across lots of 'em since the last episode. Most of them
involve third-party software, of course, but here are some DEC-only ones.
Diskquota: The sequence USE <dev>, CREATE, REBUILD, REMOVE <the UIC's I
didn't want included, including [1,4]> led to *no one* being able to write
any files without EXQUOTA. SHOW QUOTA still insisted that there were no
quota entries for the given UIC's that were having trouble. Why?? I see
no subtleties in the Diskquota utility. REMOVE should perform as
advertised; I suspect that it is not, and leaving rat turds around.
Someone else already answered this. Consider the sequence of commands you
used, though - how did you expect DISKQUOTA to figure out what quotas to
assign to all those UIC's it found on the disk?
Phone: It doesn't work at all. If I answer someone who is calling me, I
get this trash about ''attempting to continue after a transmission
error'', and more than likely an access violation. The calling end simply
exits without a whimper. Or something like that. It's so f*cked up that
I don't want to deal with it until someone tells me the stupid simple
thing I haven't done, or I get a fix, or comfirmation that it is indeed
screwed up beyond all belief. Will someone please put my mind at ease??
The most likely thing is that you haven't INSTALL'ed PHONE with the necessary
privileges. You don't say what kind of system you are running on. I haven't
checked the full VMS distribution, but in the MicroVMS distribution, the
installation of PHONE, and several other utilities, is done in
SYS$MANAGER:SYSTARTUP.COM, but it is commented out. Check the file to be
sure. BTW, if you adapted a V3 SYSTARTUP file, be aware that PHONE now needs
a new privilege. The correct privilege list is:
SYSNAM PRMMBX WORLD OPER NETMBX
Mail: What the hell is MAIL/PROTOCOL all about? Why isn't it in the
manual?
It's a hook to allow the user interface part of the MAIL program to use trans-
port mechanisms other than the MAIL-11 protocol running on DECnet. It's not
documented because it's not finished and not supported.
Spawn: Why didn't they fix the cretinous stuff about associated
mailboxes?
Because it's very hard to do. The assumption that a device has at most one
associated mailbox is buried in a lot of low-level code. It's still on
everyone's wanted list...
$brdcst: Why does this *not* check for the existence of OPER privs? I
have people with the usual two default privs using an old sender program
with universal success, *without* my having installed it with OPER. What
gives?
BUG; fixed in VMS V4.1. BTW, there is a related bug in REPLY which does not
allow you to REPLY/TERM to a terminal you own...also fixed in 4.1.
General: Is there a way to *extract* what a CLD source looked like from
DCLTABLES? *That* would be one way to find these undocumented commands...
No, there is no supported way. It's not that hard to figure out how to do it
from the fiche, and programs to do it (for V3) have appeared on DECUS tapes.
Be aware that "undocumented" means "unsupported" and "uncommited". Expect
anything you may glean from this to change, breaking your code, at the next
release.
_H* [close to nuking Manyard]
General comment: INSTALL THE VMS 4.1 UPGRADE!!! NOW!!! V4 represents a VERY
large change from V3; there's hardly a piece of code that didn't change.
There are a lot of bugs as a result. The 4.1 upgrade contains something like
67 patches. This despite two field tests at a LOT of sites for something like
9 months...
Oh, BTW, the 4.1 upgrade fixes up many of the standard CLD definitions - so
the .CLD's are right there in the update saveset. Wink wink, nudge nudge.
-- Jerry
-------