On 2012-06-28 15:45, Peter Coghlan wrote:
It is triggered by the setting of the todr at boot time. In principle,
it picks the mtime of / to get a rough estimate of what time it is at
boot time. It's done in inittodr, which is called from ufs_mount.
inittodr checks if the date is before 1975, and if it is, the it's
preposterous.
I've booted both VMS and OSF/1 or Tru64 or Digital Unix or whatever it is
called this week on the same Alphaserver 2100. The oses use the TOY clock
differently unfortunately.
When VMS notices that the time is "preposterous" is prompts me to enter the
correct date and time before allowing the machine to boot. Unix gives me the
"preposterous" message and advises me to fix the clock later while going on to
boot and put "preposterous" dates on various files :-( The VMS approach can
also be somewhat inconvenient when trying to do "lights out" operations.
A pity that a standard way of using the TOY clock could not have been agreed
for all the oses supported on a particular processor.
Note that Ultrix does not output the "preposterous" error message for any value in the todr. It's based on a date in the file system.
However, I agree that it's a shame that Ultrix and VMS is not compatible in their use of the todr. However, it would be extremely hard to do, since VMS keeps the todr in local time, while Ultrix keeps it in UTC.
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt at softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
On 2012-06-28 15:57, Sampsa Laine wrote:
Guys,
Do you think it would be possible to have a host that would somehow bridge HECnet with the Italian DECNET and then use the "poor mans routing" (i.e. HOST1::HOST2::<object) to pass e.g. mail across our two networks?
I have no idea how this works in practice, but just throwing the idea out there. From what I understood is that the Italians use many of the same nets / nodes as us (mainly network 1) which would make a network merger impractical.
I don't think that will work.
PMR is used to be able to forward traffic to a node, which you either do not know about, or else can't address (for example a Phase V address).
There needs to live a machine in the middle with connections both ways.
This means that you cannot have machines with conflicting node numbers available to that machine in the middle, but this is exactly what will happen with the italian network. :-(
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt at softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
On 28 Jun 2012, at 14:45, Peter Coghlan <HECNET at beyondthepale.ie> wrote:
I've booted both VMS and OSF/1 or Tru64 or Digital Unix or whatever it is
called this week on the same Alphaserver 2100. The oses use the TOY clock
differently unfortunately.
When VMS notices that the time is "preposterous" is prompts me to enter the
correct date and time before allowing the machine to boot. Unix gives me the
"preposterous" message and advises me to fix the clock later while going on to
boot and put "preposterous" dates on various files :-( The VMS approach can
also be somewhat inconvenient when trying to do "lights out" operations.
A pity that a standard way of using the TOY clock could not have been agreed
for all the oses supported on a particular processor.
The DEC 'Enter date and time DD-MMM-YY (later YYYY) HH:MM' prompt
pre-dates VMS, it is also present in several PDP-11 OSs (because many
PDPs didn't have a TOY clock). I think it was DEC's defacto way of
handling the date issue that got handed down to VMS.
--
Mark Benson
http://markbenson.org/bloghttp://twitter.com/MDBenson
Guys,
Do you think it would be possible to have a host that would somehow bridge HECnet with the Italian DECNET and then use the "poor mans routing" (i.e. HOST1::HOST2::<object) to pass e.g. mail across our two networks?
I have no idea how this works in practice, but just throwing the idea out there. From what I understood is that the Italians use many of the same nets / nodes as us (mainly network 1) which would make a network merger impractical.
Sampsa
It is triggered by the setting of the todr at boot time. In principle,
it picks the mtime of / to get a rough estimate of what time it is at
boot time. It's done in inittodr, which is called from ufs_mount.
inittodr checks if the date is before 1975, and if it is, the it's
preposterous.
I've booted both VMS and OSF/1 or Tru64 or Digital Unix or whatever it is
called this week on the same Alphaserver 2100. The oses use the TOY clock
differently unfortunately.
When VMS notices that the time is "preposterous" is prompts me to enter the
correct date and time before allowing the machine to boot. Unix gives me the
"preposterous" message and advises me to fix the clock later while going on to
boot and put "preposterous" dates on various files :-( The VMS approach can
also be somewhat inconvenient when trying to do "lights out" operations.
A pity that a standard way of using the TOY clock could not have been agreed
for all the oses supported on a particular processor.
Regards,
Peter Coghlan.
On 2012-06-28 12:09, Mark Benson wrote:
That's a UNIX tradition I think, lot of UNIXs I have used (IRIX,
Solaris, NetBSD to name a few) return the 'preposterous value in Time
if Day clock' or similar error if the date wrong. I think it's
triggered if the system date is prior to the kernel's date :)
It is triggered by the setting of the todr at boot time. In principle, it picks the mtime of / to get a rough estimate of what time it is at boot time. It's done in inittodr, which is called from ufs_mount.
inittodr checks if the date is before 1975, and if it is, the it's preposterous.
So, a simple way to trigger this error message would be to just set the date to sometime in 1972, shut down the system, and then reboot.
The initial time for a booted system should thus always be just following the last time the machine was shut down. This is then updated by the todr, if it exists, and have a valid value. But since the todr only is a running time that covers about 1.5 years, it cannot really tell the full time. But assuming you got it almost right (within 6 months or so), todr can then adjust the rest.
The code in current NetBSD is i principle identical to Ultrix, so the same is probably still true for all modern Unixes (I don't know about Linux though).
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt at softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
On 28 Jun 2012, at 11:26, Tony Blews <tonyblews at gmail.com> wrote:
Ultrix not liking the year 2000 means you need to be preposterous, sadly.
Use my strategy and use 1984 - the calendar is the same as 2012 and it
was a leap year :)
Slightly related: while poking around in the shed looking for a spare kettle lead, I found a shrink-wrapped copy of SCO Unix on 5.25" disks and a copy of the "Unix for VMS Users" book. I've moved house 7 times in the last 20 years. Why do i keep this crap?
If I knew the answer to that I wouldn't keep any of it either ;)
--
Mark Benson
http://markbenson.org/bloghttp://twitter.com/MDBenson
Ultrix not liking the year 2000 means you need to be preposterous, sadly.
Slightly related: while poking around in the shed looking for a spare kettle lead, I found a shrink-wrapped copy of SCO Unix on 5.25" disks and a copy of the "Unix for VMS Users" book. I've moved house 7 times in the last 20 years. Why do i keep this crap?
On Thu, Jun 28, 2012 at 12:09 PM, Mark Benson <md.benson at gmail.com> wrote:
That's a UNIX tradition I think, lot of UNIXs I have used (IRIX,
Solaris, NetBSD to name a few) return the 'preposterous value in Time
if Day clock' or similar error if the date wrong. I think it's
triggered if the system date is prior to the kernel's date :)
--
Mark Benson
http://markbenson.org/bloghttp://twitter.com/MDBenson
On 28 Jun 2012, at 10:53, Mark Wickens <mark at wickensonline.co.uk> wrote:
> On 28/06/12 10:31, Tony Blews wrote:
>> Hi.
>>
>> Has anyone had any success emulating Ultrix with DECnet under SimH?
>> I managed to find a copy of 4.0, and the "tape" claimed that it was installing DECnet, but I couldn't find any evidence of it anywhere on the system.
>> I may just be being stupid (this is not unknown).
>>
>> An aside: You've got to love an OS that gives you error messages containing the word "preposterous"!
>>
>> Tony.
> That would be reporting the TOY clock setting then?
>
> --
> http://www.wickensonline.co.uk
> http://declegacy.org.uk
>
Yeah, '70 was a preposterous year it seems. Most on the 70s were, thinking about it.
On Thu, Jun 28, 2012 at 11:53 AM, Mark Wickens <mark at wickensonline.co.uk> wrote:
That would be reporting the TOY clock setting then?
--
http://www.wickensonline.co.ukhttp://declegacy.org.uk
That's a UNIX tradition I think, lot of UNIXs I have used (IRIX,
Solaris, NetBSD to name a few) return the 'preposterous value in Time
if Day clock' or similar error if the date wrong. I think it's
triggered if the system date is prior to the kernel's date :)
--
Mark Benson
http://markbenson.org/bloghttp://twitter.com/MDBenson
On 28 Jun 2012, at 10:53, Mark Wickens <mark at wickensonline.co.uk> wrote:
On 28/06/12 10:31, Tony Blews wrote:
Hi.
Has anyone had any success emulating Ultrix with DECnet under SimH?
I managed to find a copy of 4.0, and the "tape" claimed that it was installing DECnet, but I couldn't find any evidence of it anywhere on the system.
I may just be being stupid (this is not unknown).
An aside: You've got to love an OS that gives you error messages containing the word "preposterous"!
Tony.
That would be reporting the TOY clock setting then?
--
http://www.wickensonline.co.ukhttp://declegacy.org.uk
