Linux-Advocacy Digest #763, Volume #27 Tue, 18 Jul 00 20:13:05 EDT
Contents:
Re: Are Linux people illiterate?
Re: 486 Linux setup, 250 meg HD, which distro ??? (C Sanjayan Rosenmund)
Re: Star Office to be open sourced (phil hunt)
Re: Star Office to be open sourced (phil hunt)
Re: Linsux as a desktop platform ("John W. Stevens")
Re: A cute linux song (C Sanjayan Rosenmund)
Re: I had a reality check today :( (Tim Palmer)
Re: Anti-Human Libertarians Oppose Microsoft Antitrust Action (was: Microsoft Ruling
Too Harsh (Tim Palmer)
Re: Linsux as a desktop platform ("John W. Stevens")
----------------------------------------------------------------------------
From: <[EMAIL PROTECTED]>
Subject: Re: Are Linux people illiterate?
Date: Tue, 18 Jul 2000 16:20:54 -0700
Reply-To: <[EMAIL PROTECTED]>
Aaron R. Kulkis <[EMAIL PROTECTED]> wrote in message
news:[EMAIL PROTECTED]...
> > We may never know what the first language of Britian was. So let us
start
> > with the proto gaelic of the Picts and Celts. Then came the Romans to
> > Britonium (sp?) and their introduction of Latin. It was the Roman
presence
> ^^^^^^^^^
>
> The people were known as the Britons, and the land was called Britain
>
> (not Great Britain, merely Britain).
You are generally correct. Britian has been the name of the land for a long
time. It was to Romans who first tagged those barbarians with the name
Britons and others with then name Picts. Note here I am using the word
"barbarian" in it classical Roman meaning which is not of the empire. The
Britons and the Picts had the habit of tatooing their bodies in elaborate
patterns and then going around with very little clothing, compared to
Romans. So the Romans named these people as "the painted peoples" which is
what Briton is derived from. Their land was named became know as "the land
of painted people" which has come down to us as Britian. The part of
Britain that was a Roman provence was known as Britonium (sp?) (Got to look
it up.)
Over the years Britonium has be shortend and trasliterated into Britian;
just like the Roman city of Lundinium has been shortend and trasliterated
into London. Some english translations of the old Roman writings have
preserved the name Britonium while other have rendered that into English as
well.
> Hadrian's wall ?
Correct, that was the dividing line between the land that the Romans were
able to make a part of the empire and the land that they didn't. When the
Legions were no longer able to maintain the intergrity of that border, the
wall was built to assist the Legions.
> I was in Norway recently, and was surprised how many words I could
> recognize between my English and my study of Russian.
The Viking, controlled of Britian for about a couple of centeries. Many of
the British today are the decendents of Vikings, at least in part. So the
language assorbed much from the language of the Norsemen, and the affinity
between to languages are still evident today as you discovered. Even when
the words don't come across, the sound do.
> > When it evolved into old english it started to become a
> > writtenlanguage as well. But there were no standard spelling and each
> > author "wood rite tha words lik he thot waz rite".
>
> correction:
>
> "an eech awfor wood rite tha werds like Timmie Pawlmer duz."
Perfect example of writing English without having a standard speller.
> > When that became too much of a problem, the crown commissioned the
> > development of the standard speller an dictionary for the language of
the
> > English. Which is the source of the terms "the King's English" and "the
> > Queen's English". The person who was appointed to the task was not a
great
> > linguist, so he wrote the dictionary by examining various documents
written
> > in "english" and gathered spelling samples from those documets.
Although
> > rules were established as a part of that effort they were not applied as
> > they should have. The mishmash of spelling rules the we have to live
with
> > today got started that way.
>
> something is askew here.
>
> Words of Greek origin have consistant spelling rules with themselves.
> Words of Latin origin have consistant spelling rules with themselves.
> Words of French origin have consistant spelling rules with themselves.
> Words of German origin have consistant spelling rules with themselves.
> Words of Norse origin have consistant spelling rules with themselves.
Correct. Which is consistant with the mutt nature of the development of the
english language. When the standard speller and dictionary for english was
first being developed. English has still mostly a oral language. There was
the opportunity to develop the written form to have consistant spelling
rules with in English.
>From what I understand of the process that was actually used: As many books
and other documents written in English were gathered, wordlist were
generated from them. The word lists were merged alternate spelling of the
same words were folded together and one spelling was selected. From this
master word list, the final product was created.
------------------------------
From: C Sanjayan Rosenmund <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.hardware,comp.os.linux.setup
Subject: Re: 486 Linux setup, 250 meg HD, which distro ???
Date: Tue, 18 Jul 2000 23:40:38 GMT
Steve Martin wrote:
>
> Jim Cameron wrote:
>
> > >hacked every 5 minutes your on the internet. Choose Slack, Choose sleepless
> > >nights tearing your hair out, Choose Debian, Choose never being able to use
> >
> > I use Slack, and I have LOTS of hair. 8-)
>
> I started out with Slackware, and I have *no* hair! ('Course,
> I'm 45 years old, too... <g>)
I also started with Slackware, now I use Debian. Both work well for
me. On that small system, I would use Debian. Install just the base
package and add only the applications you *actually* use. . . ends up
with a VERY small install footprint, which gives you more room for
data.
--
Sanjay
[EMAIL PROTECTED]
Windows has detected that a gnat has farted near your computer.
Press any key to reboot.
------------------------------
From: [EMAIL PROTECTED] (phil hunt)
Crossposted-To: gnu.misc.discuss,comp.sys.sun.misc,comp.os.ms-windows.advocacy
Subject: Re: Star Office to be open sourced
Date: Wed, 19 Jul 2000 00:13:32 +0100
Reply-To: [EMAIL PROTECTED]
On 18 Jul 2000 11:01:24 +0100, Phillip Lord <[EMAIL PROTECTED]> wrote:
>>>>>> "KLH" == KLH <[EMAIL PROTECTED]> writes:
>
> KLH> Sometimes I don't understand how anything as complex as UNIX
> KLH> can be as stable.
>
> KLH> It's complexity is one reason I think it should be killed. Not
> KLH> that I know of any suitable replacement for a general-purpose
> KLH> operating system, but I don't think it is the OS I want the
> KLH> future to use.
>
> Unix is complicated I have to agree.
>
> Part of this is for legacy reasons, part because of
>creeping featuritis.
>
> The main problem though is I think that what we are
>asking the OS to do is a fundamentally complex thing, and the OS
>reflects that.
I'd agree.
> To me its much the same as the flavour of the month case
>tools that used to sprout out of every vendors catalogue. However much
>people have tried to decrease the complexity of coding a lot of it has
>remained.
I don't think case tools are particularly effective. The best way would
be to use a very-high-level language. For example, 100 lines of Python
might be equivalent to 300 lines of C++ or 200 lines of Java.
> Although I don't agree with what Churchill said, when he
>was talking about democracy, I think his statement can be applied
>nicely to unix, which is "its a worst possible operating system, but
>its better than all the others".
I certainly don't konw any better ones.
Unix is very flexible. Consider what it was originally designed for --
multiuser systems with teletypes or dumb terminals, & 1-2 MB RAM. Now
consider what it's actually used for: a very wide range of devices,
from PCs, graphic workstations, supercomputers, embedded control
systems, MP3 players, and (soon) palmtop PDAs. There's never been another
OS that covers even half as wide a range of uses. The fact that it is
so versatile points to it being well designed.
> It will be nice to see what OS's are like in 500 years time.
>Perhaps they will have advanced somewhat by then!
In 500 years time, they'll be so advanced as to be unrecognisable.
A more reasonable question would be 10-30 years time. or to ask: how
would you change Unix? Some changes I'd make.
1. Use + for arguments instead of -. So:
$ something +verbose
would turn on verbose mode, and
$ something -verbose
would turn it off
2. have more regularity in arguments, probably by having a standard parser
in the OS. At the moment, some use:
$ something -a=value
others use:
$ something -avalue
I'd standardise on the former.
3. for each command-line program, have a GUI version, with the same name
but prefixed with a 'g'. So grep and ggrep, ps and gps, ftp and gftp, etc.
4. (a big change), for each optional package that can be added to the
system, the package would go in directory /opt/package-version where
<version> is of the form 1.2.3. There would be a symbolic link, /opt/package,
pointing to the one in use.
For each package there would be a standard set of files, e.g.
/opt/package/help.html would be the root of the help information,
/opt/package/start would start it running, and /opt/package/stop would
shut it down gracefully, the source would go in /opt/src/package/ and would
have a file /opt/src/package/prog.html which would be information for
programmers, etc. (There would of course be a built-in web server to view
these files with).
There would also be a system whereby a package, newly installed, can
"sniff out" other packages that it is supposed to work with and automatically
connect to them. For example, if you have apache-3.1.0 installed and you
then installed php-5.2.0, then PHP, because it normally works with
Apache, realise that Apache is there and [handwaving] sort out how to
connect up to it, possibly asking the user first.
5. I'd use Python (or something like it) as the standard scripting language,
C or C++ as the standard low-level language, and proabably something Java-like
in between. There'd also be a built-in OO database, and all these languages
would have the built-in ability to store their native data structures in the
DB, in a way that the other languages could read them. The DB would also
have a standard file format for exporting and importing information, portable
across all implementations, which would also be the standard file format for
serialising data in all the standard languages. (It's quite possible that
this standard format would utilise XML).
The OODB would be interrogated via a TCP socket (again using a standard
protocol, which would incorporate the serialisation format defined above).
The DB interrogation protocol would be defined in terms of a standard
RPC protocol (which would do the same job as Corba or Java RMI, but in a
simpler way), which of course all 3 standard languages would understand, both
as servers and as clients.
--
***** Phil Hunt ***** send email to [EMAIL PROTECTED] *****
Moore's Law: hardware speed doubles every 18 months
Gates' Law: software speed halves every 18 months
------------------------------
From: [EMAIL PROTECTED] (phil hunt)
Crossposted-To: gnu.misc.discuss,comp.sys.sun.misc,comp.os.ms-windows.advocacy
Subject: Re: Star Office to be open sourced
Date: Wed, 19 Jul 2000 00:14:12 +0100
Reply-To: [EMAIL PROTECTED]
On Tue, 18 Jul 2000 21:58:59 +1000, Sam <[EMAIL PROTECTED]> wrote:
>>
>>IIRC they bought Star Office for $400 million.
>
>Yes but where are the earnings from it ?
Nowhere AFAIK.
--
***** Phil Hunt ***** send email to [EMAIL PROTECTED] *****
Moore's Law: hardware speed doubles every 18 months
Gates' Law: software speed halves every 18 months
------------------------------
From: "John W. Stevens" <[EMAIL PROTECTED]>
Crossposted-To: comp.sys.mac.advocacy,comp.os.ms-windows.advocacy,comp.unix.advocacy
Subject: Re: Linsux as a desktop platform
Date: Tue, 18 Jul 2000 17:38:25 -0600
"T. Max Devlin" wrote:
>
>
> My computer is incredibly stupid about what should and should not be
> given priority. I am constantly waiting for things to happen, and it
> has no way of knowing what they are, other than the possible (and
> flawed, you're entirely correct) method of which one I've got on top of
> my desk. You say that isn't a good system, because it requires all the
> applications to cooperate. I say, at least it allows *some* operator
> control, without going so far as to require explicit and
> engineering-level configuration in advance of a static algorithm
> scheduler. What do you say in response?
Hmm . . . you mean, the short, simplified course on scheduling?
Ok, first off, PMT is based on high priority hardware interrupts, and
involves a small number of special objects. These objects are:
Process - A separately scheduable entity that performs some kind of
work.
Time Slice - The minimum amount of time a running process will be
allowed to run before it becomes eligible for being switched out.
Environment - The complete description of those parts of the machine
environment that a process can see.
Priority - An integer number that is used to describe the relative
"importance" of a process in relationship to all of the other processes
present in the system.
Round Robin - A description of an algorithm used to fairly allocate an
atomic resource (in this case, the processor) among a set of equally
important processes.
Process State - Describes the state of the process in terms of: running,
runable, waiting, and some others we will ignore for now (initializing,
zombie, moving, etc.).
Timer Interrupt - A hardware interrupt generated on the count zero event
by a hardware timer. Some hardware timers are self-cyclying: they will
continually count down from a preset value, issue a count zero event,
reset themselves to the preset count value, and then count down again.
Other timers, once they count down to zero, will simply stop and must be
reloaded and reset before they will begin counting down again.
Runable Process List - Consists of a sorted list of processes, sorted by
priority, with all processes of the same priority grouped together in
order from least recently to most recently run.
Note that round robin fair scheduling can be achieved by always running
the first process in the runnable process list, iff the runable process
list is sorted from highest to lowest priority, and iff a process that
is being switched out is inserted into the runnable process list at the
end of it's equal priority sub list.
Wait List - A list of processes, sorted in priority order, that are all
waiting for the same resource to become available.
A very, very *SIMPLE* PMT algorithm consists of the following states and
processes:
Initialize:
1) Initialize a timer interrupt. Set the timer to a value equal to one
time slice.
Timer Interrupt:
1) Save part of the current machine state.
2) Perform scheduling.
3) If neccessary, reset timer.
4) Restore the saved machine state.
Scheduling:
1) Search the list of runable processes for the first process of the
highest priority. In a runable process list that is sorted from highest
to lowest priority, this will always be the first process in the list.
2) If this process is the same as the currently running process, exit
scheduling.
3) Otherwise: Save the complete machine state of the currently running
process.
4) Place the currently running process into the list of runable
processes at the end of it's equal priority sub list (thereby
guaranteeing round robin scheduling).
5) Remove the process from the the list, save a reference to it in the
location reserved for storing the currently running process.
6) Load the saved machine state of the new process into the machine.
7) Done with scheduling.
Wait for resource (sleep):
1) Place the currently running process into the priority sorted list
specified for the resource being waited on.
2) Perform scheduling.
Notify Waiting Processes (wake all):
1) Remove all processes from the priority sorted list of the resource
that is now available and insert them into the runable process list.
2) Perform scheduling.
Notify Waiting Process (wake one):
1) Remove the first process from the priority sorted list of the
resource now available, insert it into the runable process list.
2) Perform scheduling.
A process runs until a timer interrupt occurs, or a process makes a
system call that causes the process to be placed on a wait list.
As you can see, the highest priority runable process always gets all of
the time slices it wants, unless there are two or more processes that
have the same priority, in which case each process of that priority gets
to run in turn until all processes have run at least once, then the
cycle is repeated until a higher priority task becomes runable, or all
of the processes of that priority move onto wait lists.
Questions?
--
If I spoke for HP --- there probably wouldn't BE an HP!
John Stevens
[EMAIL PROTECTED]
------------------------------
From: C Sanjayan Rosenmund <[EMAIL PROTECTED]>
Crossposted-To:
comp.os.ms-windows.nt.admin.networking,comp.os.ms-windows.advocacy,comp.os.ms-windows.nt.advocacy
Subject: Re: A cute linux song
Date: Tue, 18 Jul 2000 23:53:28 GMT
Err, Aaron? you *claim* to be a UNIX Sysadmin, so why is your .sig 29
lines long? The standard is 4, ya'know. . .
"Aaron R. Kulkis" wrote:
<snip>
>
> --
> Aaron R. Kulkis
> Unix Systems Engineer
> ICQ # 3056642
>
> I: "Having found not one single carbon monoxide leak on the entire
> premises, it is my belief, and Willard concurs, that the reason
> you folks feel listless and disoriented is simply because
> you are lazy, stupid people"
>
> A: The wise man is mocked by fools.
>
> B: "Jeem" Dutton is a fool of the pathological liar sort.
>
> C: Jet plays the fool and spews out nonsense as a method of
> sidetracking discussions which are headed in a direction
> that she doesn't like.
>
> D: Jet claims to have killfiled me.
>
> E: Jet now follows me from newgroup to newsgroup
> ...despite (D) above.
>
> F: Neither Jeem nor Jet are worthy of the time to compose a
> response until their behavior improves.
>
> G: Unit_4's "Kook hunt" reminds me of "Jimmy Baker's" harangues against
> adultery while concurrently committing adultery with Tammy Hahn.
>
> H: Knackos...you're a retard.
--
Sanjay
[EMAIL PROTECTED]
Windows has detected that a gnat has farted near your computer.
Press any key to reboot.
------------------------------
From: Tim Palmer <[EMAIL PROTECTED]>
Subject: Re: I had a reality check today :(
Date: 18 Jul 2000 19:54:08 -0500
On Tue, 18 Jul 2000 02:44:36 -0400, Aaron R. Kulkis <[EMAIL PROTECTED]> wrote:
>
>
>Tim Palmer wrote:
>>
>> On Thu, 13 Jul 2000 10:12:15 -0400, Aaron R. Kulkis <[EMAIL PROTECTED]> wrote:
>> >
>> >
>> >Tim Palmer wrote:
>> >>
>> >> On 8 Jul 2000 07:20:33 GMT, Ray Chason
><[EMAIL PROTECTED]> wrote:
>> >> >Tim Palmer <[EMAIL PROTECTED]> wrote:
>> >> >
>> >> >>On 6 Jul 2000 03:40:57 GMT, Ray Chason
><[EMAIL PROTECTED]> wrote:
>> >> >>>Tim Palmer <[EMAIL PROTECTED]> wrote:
>> >> >>>
>> >> >>>>It won't healp LIE-nux anny. Nobuddy want's to reed HOWTO after HOWTO after
>HOWTO. You alreddy have
>> >> >>>>users reeding TOO HOWTO's PLUS the ones they alreddy half toreed to get the
>rest of CommyLie-nux working.
>> >> >>>
>> >> >>>Can't you set up your Windoze-based newsreader so it doesn't spew these
>> >> >>>mile-long lines?
>> >> >>
>> >> >>Cant you make your Generly Not Usefall (GNU) CommyLie-nux crap to handall long
>lines propperly?
>> >> >
>> >> >1) My newsreader is of my own design and handles long lines just fine,
>> >> > thank you very much...
>> >>
>> >> Proov my point again why do'nt you? In UNIX you half to rite your own programms,
>and your another exampel.
>> >>
>> >> >
>> >> >2) but others read news in university labs and such, using VT100 terminals
>> >> > with no GUI capability.
>> >>
>> >> Today's universitty's have Windows. If all your universitty has are UNIX, then
>your universitty is living in a cave.
>> >
>> >
>> >Caves like University of California, Berkely...MIT, Purdue, Carnegie
>> >Mellon.
>> >
>> >Yeah...some caves.
>> >
>> >>
>> >> >
>> >> >3) Then there are those who have to use large fonts just to read news at
>> >> > all. Some of them are even Windoze users.
>> >> >
>> >> >4) You could horizontally scroll but that's a PITA.
>> >>
>> >> ...only if you use SLRN. In Outlook its easie you just use scroalbar.
>> >>
>> >
>> >you are addled.
>> >
>> >
>> >> >
>> >> >5) Hence long-standing rules of netiquette call for lines to wrap in the
>> >> > low 70's.
>> >> >
>> >> >You piss and moan that Linux makes *you* work harder, yet you're perfectly
>> >> >willing to make *others* work harder to read your posts. Timmy-boy,
>> >> >you're not just a Wintroll. You're also a hypocrite.
>> >>
>> >> That only half to work harder becase they use UNIX and UNIX make's them work
>> >> harder. Thats' my
>> >> hoal point. UNIX blows. Windo's is miles ahed of UNIX and you peopal are still
>> >> acting like UNIX was
>> >> stait-of-the-art.
>> >
>> >lets see...
>> >
>> >for f in `cat [file with a list of files to process]`
>> >do
>> > echo "processing file $f"
>> > process_with_awk_script_to_edit and_rearrange_columns $f
>> >done
>> >
>> >
>> >
>> >Or, alternatively, using LoseDOS drool-and-click to process
>> >50 files by hand.
>> >
>> >Only a moron can fail to see that the LoseDOS interface
>> >is FAR more work
>>
>> Or drag and drop 50 fials onto an icon to do the same thing without wrighting shell
>scripts.
>>
>
>I can do the exact same thing on any modern flavor of unix, so, like
>do you have a point?
>
"Moddern UNIX", now thear's an oxymorron.
>
>> >
>> >>
>> >> >>>...which is why nearly every Linux newsreader has a decent killfile,
>> >> >>>unlike Lookout; why nearly every Linux newsreader honors user-
>> >> >>>supplied margins, unlike Lookout; why no self-respecting Linux mail
>> >> >>>client goes around spreading viruses, unlike Lookout....
>> >> >>>
>> >> >>>Oh, but Orifice does have that cute little paper clip. That paper clip
>> >> >>>must fascinate you, doesn't it, Timmy-boy?
>> >> >
>> >> >I see you couldn't address this point.
>> >> >
>> >>
>> >> I dont see any point to adress.
>> >
>> >Blindness is not victory
>> >
>--
>Aaron R. Kulkis
>Unix Systems Engineer
>ICQ # 3056642
>
>I: "Having found not one single carbon monoxide leak on the entire
> premises, it is my belief, and Willard concurs, that the reason
> you folks feel listless and disoriented is simply because
> you are lazy, stupid people"
------------------------------
From: Tim Palmer <[EMAIL PROTECTED]>
Crossposted-To:
misc.legal,talk.politics.misc,alt.politics.libertarian,talk.politics.libertarian
Subject: Re: Anti-Human Libertarians Oppose Microsoft Antitrust Action (was: Microsoft
Ruling Too Harsh
Date: 18 Jul 2000 19:54:19 -0500
On Fri, 23 Jun 2000 22:57:23 GMT, Henry Blaskowski <[EMAIL PROTECTED]> wrote:
>In talk.politics.libertarian Donovan Rebbechi <[EMAIL PROTECTED]> wrote:
>>>
>>>They produce a product. Other people want it. Does the other people
>>>wanting it imply that the other people own it? Or do you think the
>
>> I'm not clear on what your point is.
>
>The point is that the producer of a product should own it. Period.
>The alternative is to have the products politically owned, and
>the result of that form of ownership is well documented.
You can't tock to them. Their all commys.
>
>>>producer owns it and should be able to set the terms for it's release?
>>>will you apply the same principles to your life?
>
>> It's dangerous to allow the producer of a monopoly product to set arbitrary
>> terms of release. For example, ( the one Petrich already raised ), what
>> if $1- from every Windows purchase went to Gore's or Bush's presidential
>> campaign ? I'd have a problem with either scenario. I suspect you'd have
>> a problem with at least one.
>
>MS is not a monopoly, as I've repeatedly pointed out, since they
>have several strong competitors. They are not a monopoly in the
>strict definition of the word, or in any meaningful sense, since,
>if they started treating customers like crap, people would switch
>in a second. What's the lifespan of a computer now? 4 years, maybe?
>That's how long MS would last. That is not a monopoly in any sense
>except the political you're-not-a-favored-company sense.
>
>I have no problem is MS wants to give all their money to Gore or
>Bush. Once I, or anyone, gives them some money, it is MS's money.
>After that it's not my business. Curiously, the net result of
>this case is that more of MS's money IS going to political causes.
>That is clearly not an unintended result.
------------------------------
From: "John W. Stevens" <[EMAIL PROTECTED]>
Crossposted-To: comp.sys.mac.advocacy,comp.os.ms-windows.advocacy,comp.unix.advocacy
Subject: Re: Linsux as a desktop platform
Date: Tue, 18 Jul 2000 17:55:28 -0600
"T. Max Devlin" wrote:
>
> A good point. CMT doesn't use priority, though, so it is a rhetorical
> one. Processes in PMT which aren't active aren't using any CPU either,
> but the algorithm will still inevitably provide them with at least some
> level of priority, if only the lowest possible one. Thus, CMT has a
> marginal but true efficiency in this regard, doesn't it?
How so? Applications written for a CMT operating system must assume
that they are running in a crowd, so they must still make
YieldTheProcessor calls, even when no other processes are loaded on the
system.
In an intelligent PMT system, when only one process is present on the
runable process list, or by analysis it is clear that none of the other
processes on the runable process list can displace the currently running
process . . . the timer interrupt can be turned off, saving even the
miniscule cost of that interrupt.
> The argument
> that PMT systems have a 'cognizant' scheduler belies its ignorance in
> terms of whether applications might actually benefit exorbitantly from
> more explicit control of their loading, I think.
You seem to be using terms incorrectly, here. Loading is a separate
issue from CMT vs. PMT, unless you are talking about CPU loading . . .
in which case you seem to be creating invalid relationships.
> In less complex terms,
> if the fact that apps don't "want" to yield is used to discount CMT,
> then the same logic discounts PMT for providing unnecessary
> opportunities to take CPU resources when they don't benefit from them.
Nope. No CPU resources will be taken from a task to benefit a lower
priority task, so your characterization of "unnecessary" re:
opportunities to take CPU resources is incorrect.
> The problem is, possibly, that the goal of providing the lowest average
> loading is predicated on the idea that it will allow for the greatest
> number of processes.
This goal is not recognizably part of either the CMT or PMT Design
Pattern. Where did you get this goal?
> This true goal aside, however, there is equal
> value in allowing whatever processes are present at the moment to
> control their loading behavior,
Processes cannot control their loading behavior, as they are not running
until after they have been loaded. A chicken and egg kind of thing,
here.
> so that the value of the CPU's time most
> benefits the user, rather than the abstract and potentially contrary
> goal of minimizing demand conflicts by using an external control
> mechanism.
Nope. PMT's goal is not to minimize demand conflicts, but to maximize
throughput, which is exactly what the user wants.
> I don't want to care if my CPU is at 100% utilization.
Exactly. That's the job of the scheduler. It is, however, your job to
supply a weight that the scheduler can attach to the results of one time
slice worth of work on a process by process basis.
Scheduling, in short, is the dynamically adaptive process of maximizing
the return on your resource investment.
Since, of course, the value of the result is determined by the user, not
the operating system, it is the responsibility of the user to specify
that value by setting process priorities accordingly.
--
If I spoke for HP --- there probably wouldn't BE an HP!
John Stevens
[EMAIL PROTECTED]
------------------------------
** 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 (and comp.os.linux.advocacy) via:
Internet: [EMAIL PROTECTED]
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
******************************
