Linux-Advocacy Digest #140, Volume #31 Sat, 30 Dec 00 14:13:04 EST
Contents:
Re: Is Windows an operating system like Linux? (T. Max Devlin)
Re: Is Windows an operating system like Linux? (T. Max Devlin)
Re: Why is MS copying Sun??? (T. Max Devlin)
Re: Uptimes (T. Max Devlin)
Re: Uptimes (T. Max Devlin)
Re: Uptimes (T. Max Devlin)
Re: Conclusion (T. Max Devlin)
Re: Conclusion (T. Max Devlin)
Re: Conclusion (T. Max Devlin)
Re: Conclusion (T. Max Devlin)
----------------------------------------------------------------------------
From: T. Max Devlin <[EMAIL PROTECTED]>
Subject: Re: Is Windows an operating system like Linux?
Reply-To: [EMAIL PROTECTED]
Date: Sat, 30 Dec 2000 18:38:42 GMT
Said John W. Stevens in comp.os.linux.advocacy on Fri, 29 Dec 2000
>"T. Max Devlin" wrote:
>>
>> Said John W. Stevens in comp.os.linux.advocacy on Tue, 26 Dec 2000
>> 17:05:40 -0700;
>> >"T. Max Devlin" wrote:
>> >>
>> >> Said Tim Smith in comp.os.linux.advocacy on 23 Dec 2000 16:37:31 -0800;
>> >> >Let's check Win95 against these:
>> >> >
>> >> >Scheduling: Handled by Windows, not DOS.
>> >>
>> >> DOS doesn't have scheduling.
>> >
>> >Actually, it did.
>> >
>> >DOS was designed to be a single tasking OS. IOW, DOS scheduled one and
>> >only one task (or, "process", if you prefer), allocating it 100% of all
>> >non-kernel cycles.
>>
>> That is not scheduling.
>
>Yes it is. It's single process scheduling.
There is no scheduling necessary or possible for a single process. This
is not scheduling; its lack of scheduling.
>> And at the highest, most practical level,
>
>Ooohhh . . . more of T. Max's "I, and I alone, determine what is
>practical".
>
>Yeesh!
No, but I seem to be the only one to be paying any attention to it.
>The "definitions" you make up are worthless for technical discussion.
>Which this is.
>You insult me for "redefining" . . . then go right ahead and do
>it yourself.
Yes, this is a common mistake. I do not redefine; I define. As
everyone does, when they use a word. Unlike most other people,
especially technical people, I pay attention to this, and try to stick
to a singular vocabulary which is as accurate, consistent, and practical
as possible, despite the fact that I work with people in disparate and
distinct specialties. Often, these distinct specialties use the same
word for two different things, or two different words for the same
thing, so I have a lot of practice out sorting out concepts from
terminology, and maintaining an accurate, consistent, and practical
nomenclature which "defines" terms minimally (existing definitions being
consistent with my usage) and maximally (the context and use provides
the conceptual understands of the term accurately).
Forgive me my small pride in this accomplishment. Feel free to ask
questions, though, if you're not sure how I'm defining a term.
Following the "accurate/consistent/practical" model, I find its very
easy to provide a textual definition whenever necessary which is
generally both concise and comprehensive.
>> it is a box with a name, and
>> possibly a price tag.
>
>Wrong. A box is not an operating system.
>
>Go ahead. Put the box on top of your computer, see how well it works
>as an operating system.
Go ahead, put a stream of ones and zeros on a shelf, and see how well it
sells. When a consumer buys an "operating system", they might very well
simply be buying a box with a 'license key'. The code they get to use
because of that purchase is what you want to call an operating system,
but the consumer, it is the box that they are paying for, because having
the box provides the benefits of an operating system.
>> Everything in between is the point of contention,
>> which would include an understanding that the nature of "software" is
>> somewhat amorphous, but that does not invalidate all discussion of
>> rational engineering and product design.
>
>Not everything is a point of contention, and trying the ole'
>"wining by redefinition" trick simply makes you look silly,
>here.
Just about everything I do makes me look silly, here. I'm quite
resigned to it, at this point. Were I ever to become a published
author, as I hope, or emperor of the planet, as I dream, anybody at all
could read all the crud that I've posted and might quite easily draw the
conclusion I'm a moron. It is a fact which I have learned to live with.
The point of contention remains, nevertheless, what precisely is an
"operating system", and whether DOS qualifies, and further whether
Windows qualifies, and most particularly whether both DOS and Windows
qualify. Once you are willing to stick to a definition of operating
system, DOS, and Windows, we might possible begin to examine the matter.
As it stands, it seems obvious that Windows is middleware, and DOS is an
operating system, but it is a very poor one.
--
T. Max Devlin
*** The best way to convince another is
to state your case moderately and
accurately. - Benjamin Franklin ***
Sign the petition and keep Deja's archive alive!
http://www2.PetitionOnline.com/dejanews/petition.html
------------------------------
From: T. Max Devlin <[EMAIL PROTECTED]>
Subject: Re: Is Windows an operating system like Linux?
Reply-To: [EMAIL PROTECTED]
Date: Sat, 30 Dec 2000 18:38:43 GMT
Said John W. Stevens in comp.os.linux.advocacy on Fri, 29 Dec 2000
>"T. Max Devlin" wrote:
[...]
>> I don't think so.
>The word "so" could be safely left off of your response, Max.
My, you are witty.
[...]
>> But most people don't mistake a telephone for a telephone network,
>
>Wanna bet?
Sure.
[...]
>> As for the computing power of the
>> telephone network, it may be big, but its anything but powerful.
>
>Wrong. You don't know much about the telecommunications industry,
>do you?
Guffaw. Big time.
>> Pretty
>> much capable of only a very few specific tasks.
>
>The number of tasks performed is irrelevant when discussing power.
I'm afraid I will have to disagree on that point.
>> It does them very fast,
>> of course, and incredibly reliably, but that's not power, that's just
>> speed.
>
>More "argument by redefintion", Max? The telecommunications companies
>buy a large number of some of the most powerful machinery built . . .
>
>Do you have any technical training at all?
>
>In anything?
Yes; I have a great deal of expertise in internetworking, including some
knowledge of telecommunications. I'm sorry; I'm being modest. I have a
comprehensive understanding, not just "some knowledge" of
telecommunications.
In regards to their purpose, telephone switching equipment is, indeed,
incredibly "powerful". To suggest that it is a powerful *computer*
network, however, is really a silly idea. Compared to general purpose
computer hosts, telephone switches are so functionally limited that to
call them "powerful" in comparison is really silly. Fast, yes, big,
yes, complex, yes, but powerful? No; its just a telephone system, not a
powerful computer network.
--
T. Max Devlin
*** The best way to convince another is
to state your case moderately and
accurately. - Benjamin Franklin ***
Sign the petition and keep Deja's archive alive!
http://www2.PetitionOnline.com/dejanews/petition.html
------------------------------
From: T. Max Devlin <[EMAIL PROTECTED]>
Crossposted-To: comp.os.ms-windows.nt.advocacy,comp.lang.java.advocacy
Subject: Re: Why is MS copying Sun???
Reply-To: [EMAIL PROTECTED]
Date: Sat, 30 Dec 2000 18:38:45 GMT
Said Steve Mading in comp.os.linux.advocacy on 30 Dec 2000 03:27:06 GMT;
>In comp.os.linux.advocacy T. Max Devlin <[EMAIL PROTECTED]> wrote:
>:>[...]Saying that new products
>:>(making new markets) come out sometimes and that when they come
>:>out there is a temporary situation where one company has the
>:>whole market is not "imaginary" here on planet earth where the
>:>rest of us live.
>
>: Yes it is, because it is premised on an abstraction. This abstraction
>: is known as a "market", and it doesn't exist until people are already
>: buying the product in the marketplace. Generally, the amount of time it
>: takes modern competition to present a product is far less than the
>: amount of time it takes for any particular consumer to become aware of a
>: product. The market for a new product is all the people who *might* buy
>: the product, not those who are already buying it (which is zero, since
>: the product is new.)
>
>By that reasoning, there can't be any computer monpolies as long
>as less than 50% of households actually own a computer.
The recurring discussion of what "monopoly" means is on another thread,
again. It does seem to come up a lot.
You're presuming, I guess, that every household might buy a computer,
and that any household would buy only one, and that households are the
only purchasers of computers. You see how easily gedanken experiments
are shown to be straw man arguments?
>This hasn't
>happened yet (but will soon, I think). If the "market" is all people
>who "might" buy the product, instead of all people who have bought
>the product, then nobody has an OS marketshare of more than 50% yet
>in the home computer market, not even Microsoft.
So? Monopolizing is defined as having the power to control prices and
exclude competition, not having more than 50% of the market.
>Granted, a lot of
>those people haven't bought computers because they are poor and have
>more essential things to spend money on, but to use that line of
>arguing you'd have to qualify the phrase "might buy" a bit better.
>I "might buy" a speedboat. Granted, it would eat a large amount of
>my savings, and it would be a stupid idea since I don't live on the
>shore of a lake or river or ocean, but hey, I *might* buy one anyway
>if I was being really stupid. Does that make me part of the "market"
>for speeboats?
It doesn't sound that way to me. Ever been to a boat show?
>:>
>:>I don't have time to recite a litany of everything that's been
>:>invented in the last 100 years or so just because you refuse
>:>to admit that it has happened.
>
>: Please, I only require that you cite one thing, and provide concrete
>: statistics, to show that this isn't an imaginary experiment you are
>: conducting.
>
>I did. The areoplane. You discounted it using a line of reasoning
>I don't agree with, based on the point covered up above in this post.
Well, IIRC, I pointed out that you hadn't provided any business returns
on the Wright Brother's enterprise, specifically, that might refute the
fact that "the areoplane" (sic) was not a market which was monopolized.
>(The notion that the "market" was more than just those few individuals
>who had bought a wright flyer made to order, that it consists of all
>potential buyers, including those who haven't even heard of the
>areoplane yet.) If I bring up another example, you will do the
>same again, so why should I bother? We don't agree on how the
>marketshare is measured.
No, we don't agree on whether it was measured, or can be measured. I'll
admit that I don't believe "the market", as an abstraction, can be
quantified for a new invention, and you, in parallel, have failed to
provide any actual measurements to support your claim that all new
products provide the producer with monopoly power.
>[snip]
>
>:>For example, if I walk to work, then the event
>:>of "drive to work" has failed to occur, as well as the event of
>:>"ride a bike to work", as well as the event of "take a sickday
>:>at home", and so on and so forth, yet this doesn't mean that
>:>the event of walking to work was a default event, even though
>:>"anything else" did indeed fail to occur.
>
>: I'm afraid it does, the way you've described it. If you don't drive to
>: work, ride a bike to work, take a sickday, or so forth, but walk to
>: work, then walking to work can obviously be considered, the default,
>: presuming that had you done any of the previous, you wouldn't be hoofing
>: it to the rat race.
>
>No, you don't get it. Until all those "if"s showed up, it was not
>the default action. The default action would be to NOT go to work,
>since that's what would happen if I just sat there and didn't do
>anything.
But I thought you said you didn't do that, and didn't provide any reason
for doing anything but "drive to work" unless prompted to some alternate
action. Forgive me if I was jumping the gun, but luckily your
contentiousness illustrates my point. You wanted to whine about what is
or is not "the default", so I thought it might be useful to subtly point
out that you can't ever really know what "the default" is,
philosophically, so your ranting about biking to work is just
arm-waving.
>:>You contradict yourself here. Make up your mind. If the words
>:>are amorphous, then how can one "misuse" them?
>
>: By not using them accurately (to mean precisely what you want),
>: consistently (the way others understand them), and practically (to
>: communicate conceptual information) as much as possible.
>
>You aren't using "monopoly" the "way others understand" it.
Only because most others don't understand it to begin with, but parrot
it from the media, which all too often reflects the same error that
those who don't understand the issue make.
It is the fact that the comment "monopolizing isn't illegal" has come
up, quite often, that leads me to engage in these discussions in order
to clarify the situation.
Monopolizing is illegal, as is attempting to monopolize, so obviously
there are no monopolies that are legal. Coincidentally enough, there
are other terms for what is often, inaccurately, called a monopoly;
public utility and regulated market being the two most common.
Thanks for your time. Hope it helps.
--
T. Max Devlin
*** The best way to convince another is
to state your case moderately and
accurately. - Benjamin Franklin ***
Sign the petition and keep Deja's archive alive!
http://www2.PetitionOnline.com/dejanews/petition.html
------------------------------
From: T. Max Devlin <[EMAIL PROTECTED]>
Crossposted-To: comp.os.ms-windows.advocacy,comp.os.ms-windows.nt.advocacy
Subject: Re: Uptimes
Reply-To: [EMAIL PROTECTED]
Date: Sat, 30 Dec 2000 18:38:47 GMT
Said Adam Ruth in comp.os.linux.advocacy on Fri, 29 Dec 2000 09:36:14 \
>> C'mon. Give me a field name that is referenced in an RFC spec for a
>> packet header. "Synchronization counter" isn't in there, believe me. A
>> random 32 bit counter doesn't help anybody put something in the proper
>> order.
>
>http://www.cis.ohio-state.edu/htbin/rfc/rfc793.html
>
>Read section 3.3
Adam, I've appreciated your contributions, and thank you for the link.
But I feel compelled to point out that, as I read it, and as I type
this, I am not connected to the Internet. (Pesky "off-line" readers;
how annoying.) But I am familiar with RFCs, and figure if you knew that
section 3.3 had the information I was asking for, it might have been
gracious, if a bit gratuitous, to post the information itself, and
simply mentioned that it was in RFC 793.
I will check the spec in just a few minutes, when I go back on line. I
suppose I could have delayed responding until I read it, but for some
reason I was feeling selfish.
Thanks for your time. Hope it helps.
--
T. Max Devlin
*** The best way to convince another is
to state your case moderately and
accurately. - Benjamin Franklin ***
Sign the petition and keep Deja's archive alive!
http://www2.PetitionOnline.com/dejanews/petition.html
------------------------------
From: T. Max Devlin <[EMAIL PROTECTED]>
Crossposted-To: comp.os.ms-windows.advocacy,comp.os.ms-windows.nt.advocacy
Subject: Re: Uptimes
Reply-To: [EMAIL PROTECTED]
Date: Sat, 30 Dec 2000 18:38:48 GMT
Said Erik Funkenbusch in comp.os.linux.advocacy on Fri, 29 Dec 2000 \
>"T. Max Devlin" <[EMAIL PROTECTED]> wrote in message
>news:[EMAIL PROTECTED]...
>> Said Erik Funkenbusch in comp.os.linux.advocacy on Thu, 28 Dec 2000
>> [...]
>> >Uptimes is a bit like an internet poll (which are historically
>inaccurate).
>> >Only people interested in participating are polled, and this in and of
>> >itself severely skews the statistics due to the habits of people that
>tend
>> >to participate in these kinds of things.
>>
>> Bullshit.
>
>What is?
>
>Uptimes.net is a self-selected poll. Only people that install the client
>are polled.
That doesn't make the metrics skewed in the matter at hand. We aren't
polling people's preference of OS, but merely observing the uptimes of
whatever OS is self-selected. You are correct that self-selected polls
skew statistics, but you don't seem to be aware of just which statistics
they skew. I suppose if you're soft-headed enough, just the fact that
"self-selection skews results" might be all that gets through your
filters. But the real world isn't that simple; it depends on what's
being selected, and what statistical results are being examined.
--
T. Max Devlin
*** The best way to convince another is
to state your case moderately and
accurately. - Benjamin Franklin ***
Sign the petition and keep Deja's archive alive!
http://www2.PetitionOnline.com/dejanews/petition.html
------------------------------
From: T. Max Devlin <[EMAIL PROTECTED]>
Crossposted-To: comp.os.ms-windows.advocacy,comp.os.ms-windows.nt.advocacy
Subject: Re: Uptimes
Reply-To: [EMAIL PROTECTED]
Date: Sat, 30 Dec 2000 18:38:49 GMT
Said Tom Hall in comp.os.linux.advocacy on Fri, 29 Dec 2000 17:40:54
>"Erik Funkenbusch" <[EMAIL PROTECTED]> wrote in message
>news:2m636.4445$[EMAIL PROTECTED]...
>> "T. Max Devlin" <[EMAIL PROTECTED]> wrote in message
>> news:[EMAIL PROTECTED]...
>> > Said Erik Funkenbusch in comp.os.linux.advocacy on Thu, 28 Dec 2000
>> > [...]
>> > >Uptimes is a bit like an internet poll (which are historically
>> inaccurate).
>> > >Only people interested in participating are polled, and this in and of
>> > >itself severely skews the statistics due to the habits of people that
>> tend
>> > >to participate in these kinds of things.
>> >
>> > Bullshit.
>>
>> What is?
>>
>> Uptimes.net is a self-selected poll. Only people that install the client
>> are polled.
>
>And apparently they are able to write in whatever uptime they wish into the
>http header :-(
Ooh, quick, lets spread that "fact" so we can pretend to have refuted
the numbers (again). It is really pretty amazing how far you
Micro-softheads are willing to go to spread dis-information.
For the record: No, it is not possible to write whatever uptime you wish
into the http header. The best that could be provided is arbitrary
results, which would appear on the graph as random data. (You forget
that uptime comes from the delta of the counter, not the absolute
value.)
--
T. Max Devlin
*** The best way to convince another is
to state your case moderately and
accurately. - Benjamin Franklin ***
Sign the petition and keep Deja's archive alive!
http://www2.PetitionOnline.com/dejanews/petition.html
------------------------------
From: T. Max Devlin <[EMAIL PROTECTED]>
Crossposted-To: comp.os.ms-windows.advocacy,comp.os.ms-windows.nt.advocacy
Subject: Re: Conclusion
Reply-To: [EMAIL PROTECTED]
Date: Sat, 30 Dec 2000 18:38:51 GMT
Said Chad Myers in comp.os.linux.advocacy on Fri, 29 Dec 2000 13:41:12
>"Adam Ruth" <[EMAIL PROTECTED]> wrote in message
>news:92hbg1$11j0$[EMAIL PROTECTED]...
>> > No the conculsion is real the numbers are bogus.
>>
>> The conclusion is correct in that there is nothing in the HTTP request to
>> provide uptime. There is, however, something in the packet to provide
>> uptime, so therefore the conclusion about the numbers being bogus is wrong.
>
>But who's sending out the packet? The firewall? The load-balancer? One of
>several clustered servers?
>
>Regardless of the method they choose, the numbers are completely irrelevant
>because in almost every common web-server setup scenario, it is impossible
>to determine the uptime of the machine or machines independently and
>therefore impossible to get an accurate, scientific result for uptime.
Well, gee, I'm glad we cleared that up. ;-D GUFFAW.
>The numbers are bogus and it's irresponsible and naive that Netcraft could
>think their cheesy setup could possibly be able to determine all these
>different setups.
You haven't shown one single example of Netcraft's method not working.
>It's obvious they can't because of the huge amount of screw-ups they have
>just simply determining the web server OS (IIS 4 or Linux, etc).
No, that doesn't make it obvious that their uptime values are not
correct, particularly given the fact that they are known to state that
sometimes they can identify OS but not uptime.
--
T. Max Devlin
*** The best way to convince another is
to state your case moderately and
accurately. - Benjamin Franklin ***
Sign the petition and keep Deja's archive alive!
http://www2.PetitionOnline.com/dejanews/petition.html
------------------------------
From: T. Max Devlin <[EMAIL PROTECTED]>
Crossposted-To: comp.os.ms-windows.advocacy,comp.os.ms-windows.nt.advocacy
Subject: Re: Conclusion
Reply-To: [EMAIL PROTECTED]
Date: Sat, 30 Dec 2000 18:38:54 GMT
Said T. Max Devlin in comp.os.linux.advocacy on Fri, 29 Dec 2000
16:16:32 GMT;
>Said Chad C. Mulligan in comp.os.linux.advocacy on Fri, 29 Dec 2000
> [...]
>>The point is that NO ONE has shown what is actually happening and the
>>numbers have been observed, prima facia, to be fallacious.
>
>Bullshit.
>
>--
>T. Max Devlin
> *** The best way to convince another is
> to state your case moderately and
> accurately. - Benjamin Franklin ***
>
>Sign the petition and keep Deja's archive alive!
>http://www2.PetitionOnline.com/dejanews/petition.html
I just had to repost that, and point out that I was, indeed, fully ware
of the irony of juxtaposition which is apparently made by my comment and
my sig. Yes, I believe that to be a moderate and accurate case, given
the circumstances.
--
T. Max Devlin
*** The best way to convince another is
to state your case moderately and
accurately. - Benjamin Franklin ***
Sign the petition and keep Deja's archive alive!
http://www2.PetitionOnline.com/dejanews/petition.html
------------------------------
From: T. Max Devlin <[EMAIL PROTECTED]>
Crossposted-To: comp.os.ms-windows.advocacy,comp.os.ms-windows.nt.advocacy
Subject: Re: Conclusion
Reply-To: [EMAIL PROTECTED]
Date: Sat, 30 Dec 2000 18:38:52 GMT
Said The Ghost In The Machine in comp.os.linux.advocacy on Fri, 29 Dec
>In comp.os.linux.advocacy, T. Max Devlin
>>Said Chad C. Mulligan in comp.os.linux.advocacy on Fri, 29 Dec 2000
[...]
>>>No the conculsion is real the numbers are bogus.
>>
>>Prove it.
>
>I could see several scenarios.
>
>Consider a website -- actually, a webfarm -- with a rotating DNS.
>One could measure whether the site is up by attempting to connect
>to it through port 80 or 443 (HTTP, HTTPS). Repeated connects
>through the same URL will connect to different machines -- which can
>be a problem if the webfarm isn't properly mirrored -- but the
>website wil almost always be up, assuming that the rotating DNS
>can take into account which machines are up (and that the rotating
>DNS doesn't itself go down). If a measuring system were to
>essentially ping every few seconds such a website, it would basically
>never -- or almost never -- be down, but what are we measuring?
>The individual machines could be very unreliable, but there's
>strength in numbers. (It's not perfect; a machine might go down
>while serving a user.)
That's why uptime measures uptime of a server, not of a site. The
clustering, or DNS rotary (which has never really been effective, for
esoteric reasons) would provide random results, since it is the delta,
not the absolute value, which provides uptime information. But they
would be obviously random results, and not inaccurate uptime values.
>Another scenario is using a timer -- which basically ticks away and
>is restarted upon bootup. However, because of various things, NT's
>timer cycles at 49.7 days, which means that, if NT stays up for
>more than 49.7 days, it cycles to zero because it ran out of bits.
>Therefore, the reported uptime using this method is questionable.
Not if it uses the counter as a delta value "continuity indicator",
rather than a clock. This would make sense, given that Netcraft says
they can't get valid data for NT beyond 49.7 days. After that point,
the TCP sequence number (specified to be an arbitrary unique integer, so
most systems just use the timeticks value) becomes random, so even
though the server is up, you can't track whether its "still" up.
>(Is it a bug? Is it a feature? Dunno.)
Neither do I, and that's saying a lot. There seems some respectable
reason to have a timeticks value with a 1 millisecond granularity. But
the industry standard was 10 milliseconds. Still, the purpose of
timeticks is a continuity indicator, not a clock, so it should make a
difference, as long as you know what the roll-over value is and your
sample rate is less than that. The fact that Netcraft numbers get
screwy is definitely a side-effect, neither bug nor feature. It isn't
really even related to the 49.7 rollover, though it obviously would have
never been noticed otherwise, since NT servers don't stay up for 497
days.
>I think this is the
>method Netcraft is using; what they're measuring is the maximum
>possible value of that timer, although if they bother to keep track
>they might be able to spot the rollover and compensate. (It's pretty
>obvious if one measures the system periodically, say, once a day:
>44.2, 45.2, 46.2, 47.2, 48.2, 49.2, 0.5, 1.5, uh ... what happened? :-)
>The main problem IMO is keeping track of who was measured and not
>misspotting the rollover; this requires a database of some sort, and
>things can get interesting with rotating DNS (as above) and DHCP;
>is that system you connected to using that IP addy the same one you
>connected to yesterday? Maybe one might have to use SSH host keys
>as well...)
Any non-atomic front-end (DNS rotary, clustering) entirely screws the
numbers. (I'll bet a lot of people are surprised Netcraft can get
numbers for so many sites, given this fact.)
>Still another scenario -- the most reliable one -- would have the
>system hold onto a timestamp -- "time last booted" -- and report
>it and its current time upon request (one can't depend on the system's
>idea of "now" to be identical to the requester's).
Which makes both the original time-stamp and the current time
meaningless to request; you don't know if either, or their delta, is
accurate. I'll admit, it seems that precision is being made more
important than usefulness, but that happens with computer systems, by
nature.
[...]
>The measuring system could compare the "now's" and get an idea of
>how off the clock is, then extrapolate as to when the system was
>booted in "measuring time" -- or one could simply subtract the reported
>boot from the reported now (of course, one could get very cute here,
>but that's nothing new) and report that as the uptime.
One could, yes. But the only authoritative indication of uptime is from
a remote system, using the delta of a continuity indicator to mark when
the system stops being the same instance of execution.
>Note that there are several operating systems which can be partially up.
>(Linux and most Unices supprot a "single user mode", in which the system
>is still running but most services are disabled -- including remote login.
>A system doing nothing might not be "up" in any useful sense, but the
>timer in the kernel is still ticking.)
Unfortunately, epistemologically, there is no such thing as 'partially
up'. In the current context, web sites, obviously it doesn't matter, as
the site must be both available and reachable ('up' and accessible) to
be considered in the reporting.
--
T. Max Devlin
*** The best way to convince another is
to state your case moderately and
accurately. - Benjamin Franklin ***
Sign the petition and keep Deja's archive alive!
http://www2.PetitionOnline.com/dejanews/petition.html
------------------------------
From: T. Max Devlin <[EMAIL PROTECTED]>
Crossposted-To: comp.os.ms-windows.advocacy,comp.os.ms-windows.nt.advocacy
Subject: Re: Conclusion
Reply-To: [EMAIL PROTECTED]
Date: Sat, 30 Dec 2000 18:38:55 GMT
Said Adam Ruth in comp.os.linux.advocacy on Fri, 29 Dec 2000 09:21:33
> [Chad:]
>> It's been shown numerous times.
>>
>> The OS cannot be determined accurately. Period.
>>
>> Cases in point: Several sites (listed in another thread on this topic several
>> weeks ago) show that they web server is IIS 4.0 running on Linux or BSD.
>
>Then please point me to that post. I have been asking for such data for
>weeks.
Indeed; I believe Chad is referring to an unsupported claim made earlier
in the thread.
>> In this case, it appears that there is a Linux or BSD firewall/load balancer
>> and that the web server behind it is NT/IIS. Now, which uptime do you think
>> is being reported? Is it the web server, or the firewall/load balancer?
>
>The firewall/load balancer, of course. OS and uptime will typically come
>from the front end machine (through network characteristics), the webserver
>comes from the http header strings. If the OS is coming from the firewall,
>so is the uptime.
Actually, I believe that the Netcraft FAQ stated that they determine the
OS from (unstated) characteristics of the TCP/IP packet, not the http
headers. I don't even thing HTTP "headers" include OS info, which isn't
to say that it isn't available during an HTTP "session".
--
T. Max Devlin
*** The best way to convince another is
to state your case moderately and
accurately. - Benjamin Franklin ***
Sign the petition and keep Deja's archive alive!
http://www2.PetitionOnline.com/dejanews/petition.html
------------------------------
** FOR YOUR REFERENCE **
The service address, to which questions about the list itself and requests
to be added to or deleted from it should be directed, is:
Internet: [EMAIL PROTECTED]
You can send mail to the entire list by posting to comp.os.linux.advocacy.
Linux may be obtained via one of these FTP sites:
ftp.funet.fi pub/Linux
tsx-11.mit.edu pub/linux
sunsite.unc.edu pub/Linux
End of Linux-Advocacy Digest
******************************
