Re: ObjectAda - no clock drift!

Re: ObjectAda - no clock drift!

[Larry Kilgallen]

In article <6m6f0t$1ue$1@nnrp1.dejanews.com>, dennison@telepath.com writes:

> The Ada rm in D.8 paragraph 41-43 requires vendors to give "an upper bound on
> the drift rate of Clock with respect to real time." I was curious what my
> Aonix ObjectAda compiler said, so I looked it up.
> 
> The documentation Requirements section of the ObjectAda 7.1 NT docs (p 2-9)
> has the following statement:  There is no software clock drift.
> 
> What is meant by this? Surely not what I think it's saying!

Since you express surprise, I would concur that it must not be
what you were thinking, whatever that is.

I believe that Aonix is saying that the software they supply
introduces no drift, and if you see drift you should take it
up with Intel (or Cyrix, or AMD).

Since Aonix does not sell hardware, I don't know how much
further they can go (aside from saying you might also take
it up with DEC/Compaq or Samsung, but they don't support
Alpha the last I heard (drift-free or not)).

Larry Kilgallen

Re: ObjectAda - no clock drift!

[Corey Ashford]

dennison@telepath.com wrote in message <6m6f0t$1ue$1@nnrp1.dejanews.com>...
>Apparently, Aonix has made atomic clocks obsolete with the simple use of a
>Wintel PC.
>
>The Ada rm in D.8 paragraph 41-43 requires vendors to give "an upper bound
on
>the drift rate of Clock with respect to real time." I was curious what my
>Aonix ObjectAda compiler said, so I looked it up.
>
>The documentation Requirements section of the ObjectAda 7.1 NT docs (p 2-9)
>has the following statement:  There is no software clock drift.
>
>What is meant by this? Surely not what I think it's saying!

I think it's saying that the way the software is written, it will not cause
the current
time to drift as a side-effect of the way it keeps time the way some
algorithms do.  However, if the hardware clock drifts, it has no control
over that.

- Corey

Re: ObjectAda - no clock drift!

[dennison]

In article <6m7r5m$4gn$1@usenet.rational.com>,
  "Corey Ashford" <corSPAMey@rational.com> wrote:
>
>
> dennison@telepath.com wrote in message <6m6f0t$1ue$1@nnrp1.dejanews.com>...
> >The Ada rm in D.8 paragraph 41-43 requires vendors to give "an upper bound
> on
> >the drift rate of Clock with respect to real time." I was curious what my
...
> >has the following statement:  There is no software clock drift.
.
> I think it's saying that the way the software is written, it will not cause
> the current
> time to drift as a side-effect of the way it keeps time the way some
> algorithms do.  However, if the hardware clock drifts, it has no control
> over that.

Ahh. I think my confusion came from assuming they were answering the question
that was asked. :-)

The RM says "An upper bound on the drift rate of Clock with respect to REAL
TIME." (emphasis mine) Aonix gave me an answer with respect to the system
clock. Those are clearly not the same thing. So someone looking here for
Clock's drift rate with respect to real time will not get it.

In all fairness to Aonix, there's no way they could know what the clock drift
rate for the hardware in any particular PC is. But they should have said that,
rather than answering half the question in an incredibly terse manner.

I guess I was mostly posting this to see if this is a just a farcical non-
compliance with the documentation standard, or if I'm misunderstanding the
standard. There are several other examples I could site, but this one was the
funniest. From the seriousness of the responses I'm getting, it looks like
this level of "compliance" is probably typical. Its annoying to think that
I'm going to have to stop and analyze all their answers now to make sure they
are really answering the question that was asked.

T.E.D.

-----== Posted via Deja News, The Leader in Internet Discussion ==-----
http://www.dejanews.com/   Now offering spam-free web-based newsreading

Re: ObjectAda - no clock drift!

[Robert A Duff]

dennison@telepath.com writes:

> I guess I was mostly posting this to see if this is a just a farcical non-
> compliance with the documentation standard, or if I'm misunderstanding the
> standard.

I think the Documentation Requirements in the RM shouldn't be there at
all.  You can't legislate good documentation.  If you try, you get the
sort of documentation you're complaining about -- totally useless
information, intended purely to satisfy the letter of the law.

Note that the validation process makes no attempt to verify that
Documentation Requirements are met properly.  If the vendor had written,
"Humpty Dumpty sat on a wall" as the answer to that question, they would
still be just as validated.

- Bob

-- 
Change robert to bob to get my real email address.  Sorry.
-- 
Change robert to bob to get my real email address.  Sorry.

Re: ObjectAda - no clock drift!

[dennison]

In article <wccbtrrpwle.fsf@world.std.com>,
  Robert A Duff <bobduff@world.std.com> wrote:
>
> dennison@telepath.com writes:
>
> > I guess I was mostly posting this to see if this is a just a farcical non-
> > compliance with the documentation standard, or if I'm misunderstanding the
> > standard.
>
> I think the Documentation Requirements in the RM shouldn't be there at
> all.  You can't legislate good documentation.  If you try, you get the
> sort of documentation you're complaining about -- totally useless
> information, intended purely to satisfy the letter of the law.

I'm beginning to agree with you. I only spotted this because it was so
outlandish. There seems to be lots of good information about the compiler in
this section, but now I can't trust *any* of it.

> Note that the validation process makes no attempt to verify that
> Documentation Requirements are met properly.  If the vendor had written,
> "Humpty Dumpty sat on a wall" as the answer to that question, they would
> still be just as validated.

I guess that explains why they answered 18 of the doc req's as "To be
determined".

Surely that means any day now they will have all those things determined and
will send me an addendum to the docs. I'll think I'll go check my mail for a
package from Aonix right now! :-)

T.E.D.

-----== Posted via Deja News, The Leader in Internet Discussion ==-----
http://www.dejanews.com/   Now offering spam-free web-based newsreading

Re: ObjectAda - no clock drift!

[Larry Kilgallen]

In article <6m904m$vv0$1@nnrp1.dejanews.com>, dennison@telepath.com writes:

> In all fairness to Aonix, there's no way they could know what the clock drift
> rate for the hardware in any particular PC is. But they should have said that,
> rather than answering half the question in an incredibly terse manner.

Well, Aonix was describing a compiler, not a computing system.

If the RM applies to compilers rather than computing systems,
_it_ should provide the wording that excludes drift beyond the
control of compilers.

On the other hand, some of the responses in this group have
indicated that the intentional style of the RM is terse. If
so, Aonix would seem to be fully justified in responding in
the same style.

Larry Kilgallen

Re: ObjectAda - no clock drift!

[tedennison]

In article <1998Jun17.221533.1@eisner>,
  Kilgallen@eisner.decus.org.nospam wrote:
>
> In article <6m904m$vv0$1@nnrp1.dejanews.com>, dennison@telepath.com writes:
>
> > In all fairness to Aonix, there's no way they could know what the clock
drift
> > rate for the hardware in any particular PC is. But they should have said
that,
> > rather than answering half the question in an incredibly terse manner.
>
> Well, Aonix was describing a compiler, not a computing system.
>
> If the RM applies to compilers rather than computing systems,
> _it_ should provide the wording that excludes drift beyond the
> control of compilers.

Perhaps, but compilers for many single-vendor non-PC platforms certianly
could supply this information. It looks like the RM had just this in mind,
with compilers for PC's and some platform "families" being forced to use the
weasel words. Additionally, I could see where the the disctinction between
the two kinds of drift could get really blurry on platforms where Calendar
doesn't have OS support.

Even when the platform's drift isn't known, I could still get an answer like
"0 + F, where F is the clock drift of the hardware", or "10 + F + G where F
is the clock drift of the hardware in seconds per month and G is the clock
drift of the underlying operating system in seconds per month".

T.E.D.

-----== Posted via Deja News, The Leader in Internet Discussion ==-----
http://www.dejanews.com/   Now offering spam-free web-based newsreading

Re: ObjectAda - no clock drift!

[Christopher Green]

In article <6m6f0t$1ue$1@nnrp1.dejanews.com>,  <dennison@telepath.com> wrote:
>Apparently, Aonix has made atomic clocks obsolete with the simple use of a
>Wintel PC.
>
>The Ada rm in D.8 paragraph 41-43 requires vendors to give "an upper bound on
>the drift rate of Clock with respect to real time." I was curious what my
>Aonix ObjectAda compiler said, so I looked it up.
>
>The documentation Requirements section of the ObjectAda 7.1 NT docs (p 2-9)
>has the following statement:  There is no software clock drift.
>
>What is meant by this? Surely not what I think it's saying!
>
>T.E.D.

I *think* what this means is that there is nothing in ObjectAda to make the
clock drift on "Wintel" hosts any worse than it already is.

If clock drift with respect to real time is important to you, you should not
be relying on the unaided operating system clock of a "Wintel" host, or any
product that derives its clock from such a clock.

NTP has been ported to Windows NT.  I do not know how effective it is at
disciplining the clock on NT; it is generally effective on Unix hosts.
Visit http://www.eecis.udel.edu/~ntp for details.

-- 
Chris Green                                  Email cgreen@atc.com
Advanced Technology Center                   Phone (949) 583-9119
22982 Mill Creek Drive                                   ext. 220
Laguna Hills, CA 92653                       Fax   (949) 583-9213