Linux-Advocacy Digest #298, Volume #27 Sat, 24 Jun 00 03:13:05 EDT
Contents:
Re: mind hours in development Linux vs. Windows (R.E.Ballard ( Rex Ballard ))
Re: What UNIX is good for. (Andy Newman)
Re: Microsoft Ruling Too Harsh (Donovan Rebbechi)
Re: Why Jeff Szarka Has Zero Credibility When He Claims Problems With Linux
----------------------------------------------------------------------------
From: R.E.Ballard ( Rex Ballard ) <[EMAIL PROTECTED]>
Crossposted-To: alt.os.linux,comp.os.linux.misc
Subject: Re: mind hours in development Linux vs. Windows
Date: Sat, 24 Jun 2000 06:04:23 GMT
In article <[EMAIL PROTECTED]>,
[EMAIL PROTECTED] wrote:
> O.K., I know this is probably an impossible
> question to answer, but what the hay.
Actually, it's really not that hard to answer.
> Would anyone care to compare--either quantitatively or
> qualitatively--the number of mind hours that
> have gone into developing
> Linux as an OS verus what has
> gone into developing Windows as an OS?
To really answer this question, you must go back to the beginning.
Microsoft developed MS-DOS as a minimalist operating system designed
to primarily "load the program from disk and execute it". The
application did nearly everything else.
Multitasking was done by the application (TSRs such as Sidekick).
Memory Management was done by applications (Lotus 1-2-3 himem).
Overlays, segmentation, and paging were done by the application
(WordStar, WordPerfect, Dbase).
In some cases, even the hard drive itself was manipulated by the
application (writing "bad blocks" to the hard drive).
The system was used by a single user and when the system failed
everything was fixed by the "three finger salute" <cntl><alt><Delete>.
Backward compatibility wasn't even an issue, but in many cases,
incompatibility was considered a strategic advantage. This meant
that code needed to be redesigned with each generation of MS-DOS
and later Windows.
When Windows 3.0 was released, it crashed as often as once an hour,
but no one cared. It looked like a Mac, it played most of the old
MS-DOS programs, and it ran some of Microsoft's little applets like
notepad and paint.
Keep in mind that from Windows 3.0 to Windows 98, Microsoft was playing
"beat the clock". The competition was percieved as moving faster than
Microsoft's programmers could move. At the time that Windows 3.1
was released, X11R4 was available, stable, and in source code format.
The big question was did you want the pretty Motif buttons (slow but
pretty), the OLIT buttons (fast and pretty, but harder to program),
or the Athena buttons (fast and ugly but very reliable and ran on
small machines).
> I'm writing a magazine article for a trade magazine and don't know
much
> about this stuff. I've heard people call Linux more reliable than
> Windows. If true, it seems to me that this could be because
> a) Linux is better designed
Remember that at it's heart, Linux started out with the goal of
being source code compatible with UNIX. Furthermore, when Linus
published the source code to Linux, the UNIX community was frustrated
by AT&T's insistance on a "floor" price of $700 for the base system
and nearly $3500 for a fully configured system complete with graphical
interface.
Linus quickly gained the support of numerous old UNIX hackers who were
thrilled at the ability to contribute the best and leave the garbage
out. Furthermore, both UI and OSF wanted to make sure that Linux users
could migrate from Linux to their machines and worked very hard to
make sure that they could provide source code compatibility under the
GNU C compiler.
UNIX had come from a history of a multiuser environment. It wasn't at
all unusual for a single VAX 11/780 running BSD 4.1 to have as many as
200 users using it concurrently from VT-100 compatible terminals and
terminal emulators. Needless to say, when a UNIX system went down,
everybody noticed. A 30 minute outage with 200 users idled could
cost an employer as much as $20,000. Since the average UNIX
administrator of that period only made about $40,000, two outages
were the equivalent of the administrator's salary.
Each administrator knew that there were others in the same boat.
They communicated with each other because they HAD to. They sent
e-mail via UUCP links that were linked in a huge store-and-foreward
network called usenet. In most cases, employers purchased the source
code to BSD because it was their best chance of getting this 200 user
system back up and running in the shortest possible time.
Needless to say, administrators got very effective at identifying
problem applications with memory leaks and other problems and became
very adept at fixing them quickly. Soon they could identify a problem
user or problem process even before the rest of the users noticed.
A key element was the ability to combine very small simple
processes called "filters" together into "pipelines". Since
each process in the pipeline was self-contained and very efficient,
the whole pipeline was more reliable. Since each process was
managed by the kernel (including the communication), it was
easy to create complex applications using simple functions.
Ironically, UNIX didn't have a relational database for several
years because it was easier to use the simple "pipe and filter"
tools available in UNIX.
Because UNIX was designed to run these little programs, it was
very efficient at switching between them. Modern Linux and UNIX
systems can perform as many as 1 million context switches per second.
Because each program runs for a relatively short time, UNIX was
designed to efficiently manage frequently used memory between
processes. In many multi-layer cache systems, the most commonly
used I/O functions are never swapped out of memory. The most
frequently used functions are kept together instead of with
their respective objects which means that even RISC processors
effectively become self-optimizing CISC machines.
Eventually, as Unix moved into more and more commercial uses,
it became very popular for mission critical systems. Eventually,
UNIX was adopted for TCP/IP and the ARPA-Internet. This was a
system designed to survive a nuclear holocaust, to prevent
interception from even the best cryptographers, and to prevent
cracking from even the most persistant hackers. Ironically,
the military tried to prevent the use of DES encryption and
Kerberos authentication in domestic commercial systems because
they were afraid they couldn't crack messages sent by enemy spies.
> b)it attempts to do less,
Linux was smart in it didn't try to reinvent the wheel. Java and BEOS
tried to go all the way back to the basic atoms and then generate a big
enough following to create a huge software base equivalent to the UNIX
software base. Linux simply started with the baseline software that
was already available and used that as the foundation.
Linux achieves a user-friendly look-and-feel by integrating
GUI front-ends written in Python, TCL/TK, or PERL/CGI to the
traditional scripted languages and commands such as ed, awk,
and grep.
The result is that one can generate highly leveraged code. Nearly
any good consulting firm will tell you the importance of their
intellectual capital repository. It lets them create applications
in less time than if they tried to create them from scratch using
standard C or BASIC compilers. Linux provides one of the largest and
most substantial intellectual capital repositories available for the
price of a utility or a game.
I have personally had cases where I had to estimate projects for NT
development. I found that if I could IMPLEMENT then ENTIRE PROJECT
in Linux in 10 days, that the first release of the NT version wouldn't
be stable in less than a year. The "Rapid Application Development"
tools provided by Microsoft made it easy to create forms, but didn't
handle the business rules, or manage the race conditions and deadlocks
involved in managing hundreds of concurrent threads.
With Linux, I could create an UGLY CLI working model in 7 days, spend
3 days editing a CGI form with HotMetal or Communicator, and have
a fully functional system by the end of the 10th day.
> c) more people have invested time in making it work and/or
UNIX has been around since 1968, has been used commercially since
about 1983 (divestature of AT&T), and has been taught to about 3
million people. The earliest UNIX programmers went to school for
5 years in electrical engineering, learning statistics, calculus,
and several programming languages as part of their training. They
worked on problems like missle guidence systems, air traffic control
systems, or other complex "do or die" environments before they were
turned loose on the business world.
Many of the programmers who contributed to the BSD 4.1 and later
releases of UNIX were MSEE and PhD students, not freshmen hackers
who were still trying to figure out pointer constructs. And many
of these students had been given the original AT&T Version 6 source
code, along with a detailed description of every major function.
> d)smarter people (and, hey, let's say
> better looking while we're at it) have
> invested time in making it work
> (I guess there's some overlap with "a)" here).
In a world of "freeze dried instant microwavable success", the Linux
supporter stands out as something extraordinary. The average Linux
user receives as little as 20 hours of direct support from an outside
source (friend, co-worker, business associate), spends as much as 130
hours just learning the basics of the Linux user interface and GUI
applications. Many Linux "gurus" spend over 1000 hours per year
learning new applications and technologies (20 hours a week for a year
less a 2 week vacation). Many UNIX professionals spend 2000 to 3000
hours a week on "original thought" - solving problems that have never
been solved before, as opposed to repetitive routine activities.
> I thought I'd make a meager attempt to
> evaluate the possibility of "c"
C is to the UNIX programmer what Machine code is to the Windows
programmer. You CAN go into debug and try to decode the instructions,
but you probably don't want to.
I would suggest an "introduction" exercise of writing a PERL script
that would be called by Apache using the Web browser as the user
interface. Start with a utility from the CPAN library and see
what you can add. From there, you can try Python, TCL, or Java.
You can use the GTK tools from most of these languages.
There are three primary construction pattern most frequently used
in Linux/UNIX programming.
The Procedural - which you are familiar with from VBScript, JavaScript,
Visual Basic or Visual C.
Object Oriented - which you are familiar with from Visual C++, Java,
and ActiveX.
Parser - This is where PERL comes in. You can also use YACC, and
the SAX XML parser. These look for particular patterns and execute
when matches or events occurr. This is a pure "Rules Engine" in which
patterns or tokens can be used to generate a valid set of rules.
There are a few good books on PERL, but I'd reccommend the O'Reilly
"Camel Book" as the best "first cut". It should take about a week
or two to learn (about 30 hours).
It's also a good idea to learn at least one "shell" language. This
is the easiest way to create simple pipelines. The underlying code
is fast, efficient C code, but the application itself is held together
by "glue" from the shell.
> --although if anybody wants to cast a vote
> or express a thought as to
Well, I've expressed several thoughts tonight. This is one of my
longer epistles. you seem like a genuinely interested person and
I felt you deserved the attention.
> the other options (or to propose alternatives),
I'd be interested to
> read. Thanks for any thoughts.
>
> Oliver Baker
>
> .
>
> .
>
> .
>
> .
>
--
Rex Ballard - Open Source Advocate, Internet
I/T Architect, MIS Director
http://www.open4success.com
Linux - 90 million satisfied users worldwide
and growing at over 5%/month!
Sent via Deja.com http://www.deja.com/
Before you buy.
------------------------------
From: [EMAIL PROTECTED] (Andy Newman)
Crossposted-To: comp.unix.advocacy
Subject: Re: What UNIX is good for.
Date: Sat, 24 Jun 2000 15:56:52 +1000
Reply-To: [EMAIL PROTECTED]
Dave Vandervies wrote:
>Editor, mail reader, news reader, web browser... Have they put in a
>compiler yet, or does it still call an external program for that?
Depends on the language :-Q Built in elisp compiler.
--
Chuck Berry lied about the promised land
------------------------------
From: Donovan Rebbechi <[EMAIL PROTECTED]>
Subject: Re: Microsoft Ruling Too Harsh
Date: Sat, 24 Jun 2000 02:35:31 -0400
[ posted and mailed ]
Hi. My news feed didn't get this, so I'm posting to cmop.os.linux.advocacy
and mailing. I hope this wasn't supposed to be private (-; I'm guessing it
wasn't.
On Fri, 23 Jun 2000, MK wrote:
> >The problem comes when the company in question has a monopoly and the
> >resources to hire every major vendor.
>
> No such thing can happen -- profitability of subsidized activity attracts new
> businesses, and if necessary, startups. Just like guys that got money from
> "National Cordage Association" were using the money to start new companies and
> suck even more money, until NCA collapsed.
The problem is that Microsoft are not "subsidizing" anyone. They are not
*bribing* the OEMs, they are blackmailing them. They do not lose money on
the "discount" prices of Windows. The marginal profit of a Windows OS sale
is not that much less than the OEM price.
> This, who subsidizes has to make money at some point. Which means
Again, as I've said, I don't believe Microsoft are "suibsidizing" anyone,
because the use of the word "subsidizing" would imply that they are making
a marginal loss per sale when they do these deals with OEMs. I do not
believe this to be the case.
> that he has to raise prices and stop subsidizing. Which immediately
> pisses customers off and attracts competition. MS has a lot of money, but even
> they don't have source of infinite money.
You don't need to have money to blackmail someone. They get profit (X) if
the OEM bends over, and profit (Y) if the OEM does not. Either way, they
do not make a marginal loss per OS sale.
> So? Lots of companies grow by mergers and acquisitions. I don't
> see any problem here, not in case of MS nor any other company.
Well, I do. Some acquisitions and mergers are not permitted because the
result is increased monopolization.
> In fact, it is regulation of industry that routinely results in worsening
> price/performance and quality of products and leads to practical
> monopolizatoin.
Huh ? How does preventing monopolies from forming lead to "practical
monopolization" ?
> in trying to achieve monopolization. It is naive to think that the law will
> be not used as weapon to destroy competition. In words of Sun executive,
> it is cheaper to use govt's lawyers.
I'm not sure why it's OK for a monopoly to destroy competition but not
the law.
> It is _lack_ of regulation that has
> to result in negotiations and cooperation.
Not always. Complete lack of regulation leads to coercion and
strongarming, and "survival of the biggest bully".
I believ ethe pro MS people mistakenly confuse "cooperation" and
"negotiation" with "coercion".
If I "negotiate" with you and I am pointing a gun at your head, it's no
longer "negotiation", and it's no longer "peaceful". The end result is no
longer "cooperation".
> I remember reading interview with David Miller where he claimed that
> Sun can not make up losses in software or hardware from the other
> one, bc they are separate companies. Maybe it changed from that
> time. Anyway, the distinction is arbitrary anyway -- it's pretty obvious
> who has software monopoly on Sun hardware,
AGAIN, let me remind you that Microsoft are not in court for having a
monopoly. They are in court for what they did with their monopoly.
> and Apple does
> for sure.
Yes, they do. But again, it boils down to what you do with your
monopoly. BTW, I believe that Apple, like Microsoft, have walked a pretty
fine line over the years. Half the problem with Microsoft is that they
took their chances by living on the edge, and when they got their day in
court, to put it bluntly, they just fucked up.
> I think what really pains some people is that MS does NOT fail its customers
The case is not primarily about whether or not they "failed their
customers". It's got more to do with restraint of trade.
> >Yes, but these companies do not have monopoly power.
>
> Neither does MS. No sensible definition of monopoly fits MS. The
THe legal definition and dectionary definitions equivalent to that
definition fit MS. THough I guess your dogmatic attitude will no doubt
dismiss those definitions as "not sensible".
> truth is that they're big and lots of people for some reason hate
> MS with vitriol, so they will accept ridiculous definition.
THe legal definition is not just something that someone made up one day,
and neither are dictionary definitions. Both predate the MS anti trust
case. You are simply wrong in your claim that these are hasty ad-hoc
definitions constructed in the heat of the moment by MS haters.
> Remember, the only way to achieve monopoly is _forceful deprivation
> of substitute_ -- otherwise the very high prices that monopoly tries
> to impose attract looters immediately.
It depends on how high those prices are.
> HP charges lots of money for its ink
> cartridges, while some company in Switzerland produces compatibles
> and sells for a lot less, adding the note "using this cartridge does
> not make you lose the warranty of equipment".
But HP *does* charge a lot of money, and presumeably can actually sell
them for a lot of money, right ?
> MS doesn't have monopoly power, bc it isn't monopoly -- you CAN
> use substitute, they are available. They're just not attractive enough
Irrelevant,. This is not the legal definition of a monopoly, it's just one
of many definitions that happens to suit your argument.
> >>done in conditions of informed consent. Everybody knows what they get
> >>and what they don't get. That makes deal fair.
>
> >If I point a gun at your head and say "give me your money and I won't shoot",
> >then you could also call that "informed consent";
>
> No you could not. You take totally ridiculous stance.
Read what you said. "Everybody knows what they get and what they don't
get. That's fair". You are completely ignoring the possibility of
coercion, and it's impact on the fairness of "negotiations" and the fact
that it violates any reasonable definition of "consent".
> >in this circumstance
> >everybody also knows what they get and don't get. But there is some
> >strong-arming going on.
>
> You know how it goes, "those who do not distinguish power of a dollar from
> power of a whip, eventually will learn the difference on their own backs".
And I suppose those who do will feel it in their pockets instead. But it's
hardly a pleasent choice to make, or a shining example of "informed
consent", huh ?
> The OEM does have a choice -- go niche. Which is sensible choice
> given how crowded the Windows market is. You seem to naively assume that
> everybody has to do what everybody else does.
HAHAHAHAHAHAHAHAHAHAHAHAHAHA
I could just imagine Michael Dell saying this at a share holders
meeting. Everyone at the meeting would start looking nervous, reaching for
their mobiles and telling their brokers "sell, sell, sell !". The ones who
didn't have cell phones would lynch Dell before he could get out the next
sentence.
Seriously, this is not only a non-viable option, it's an outright silly
one.
> >BeOS are not a monopoly. BeOSs "obstruction of trade" is unlikely
> >in practice to have a sizeable impact on the market.
>
> But it surely could be qualified as dumping and "intent to monopolize"
> by anti-trust. Why not? Bc it has small market share? Well, that
Because they are not suibsidising their dumping with monopoly profits ?
> Also, another charge was "bribing".
I think it was more like "black mailing". I've consistently argued that
coercion was involved, and the evidence put forth suggests this. The OEMs
were coerced and intel was coerced.
> Why BeOS is not charged with
> bribing? It's giving 100% discount, after all?
BeOS are not coercing anyone and they are not telling the OEMs to sell
their product exclusively.
> Nonsense. All of the business would have to be called
> "coercion". That is money talks, but not coercion. Coercion
> is when you _really_ can not choose another option without
> being subject to force.
In ijntel's case they would have been either out of business or in severe
difficulty if they didn't play with MS. The CEO certainly would have lost
his job after the stock dive.
Ditto with the OEMs -- what are they going to do, tell their shareholders
"We are going to 'go niche' because some guy on usenet said it would be a
good idea" ? The CEOs would lose their jobs for sure.
> >is that MS have a gun at their heads -- they can tell the OEMs "do exactly
> >what I say or else I'll blow your head off".
>
> Oh please. That's just figment of your wild imagination. MS has to sell to
> _somebody_. If they alienated their OEMs and indirectly OEM's customers, that
> would come back and hit themselves. Suppose MS wanting to control Dell and
> Compaq would deny them sales.
It wouldn't "deny them sales". It would be awfully good for gateway, who'd
grab the market share lost by those that were silly enough to "go niche".
> What a load of bullshit. You want me to believe that customers
> started to like Linux bc MS is in court? Or that nobody would
No, more like MS has behaved a little better because they are under more
scrutiny.
> preinstall Linux even if they had customers bc thugs from
> MS would come and shoot them?
Nah, they'd just increase the price of their OEM licenses to their
punitive defaults.
> cocaine market. You want me to believe that if there were money to be made on
> offering machines with preinstalled Linux, some people would not go for it?
Yes, some will, but the mainstream vendors have their hands tied.
> And that is DOJ that took the fear of doing it away? How ridiculous your
Since Microsoft have faced the DOJ suit, we haven't seen as many dirty
tricks from them.
> Weapon manufacturers can soon find themselves in court. Until they do, black
> markets in weapons don't exist or something?
Yes, but it's still difficult to buy them. Most buyers do not even know
where to start looking for them.
> >Microsoft seem to have quietened down during
> >the suit.
>
> What would they do to hardware vendor that wants to specialize
> in selling machines with preinstalled Linux? Blow them up? How
These are niche vendors. They're happy to control the mainstream, because
then they can misrepresent everything besides their own product as
"fringe" as opposed to "mainstream".
THe problem is that the Dells and Compaqs and Gateways cannot afford to
lose their Windows business. If they did lose it, it would kill them.
> What threats, again? Will MS burn your place or something? Or will
> it just _not grant you discount_ or at worst not sell you Windows?
If you're Dell, a threat to "not sell you windows" is a threat to put you
out of business. If you're the CEO, it's also a threat to put you out of a
job. If my boss threatened to fire me, I'd count that as a "threat".
> To me it doesn't -- what does it mean "mainstream". Does it guarantee
> huge revenues just bc you're doing what everybody else is doing? I'd say that
> it is opposite case. If vendor is small, it can easily go niche. If vendor is
> big, it pays a lot of money to MS for software, which means that what you call
> threat is rather empty.
It is not empty. If you are Dell, you are aware that Gateway and Compaq
are just waiting for the chance to steal your market share. There's demand
for computer hardware, it's not like Microsoft will just lose money
because Dell stop selling computers.
> Don't be an idiot. RH can grant the same kind of privileges to
> the niche vendors. It is much like franchising and it's not
> based on threats, but on _exchange of privileges_.
(a) It's not at all like franchising. MS doesn't own the OEMs and doesn't
pay them. What it does is threaten to jack up their prices if they
don't let Bill give them a facial.
(b) If you're a mainstream OEM, losing MS's business is *NOT* an option.
You will not be able to support yourself on sales of MS-free
machines.
> >why we have things like SEC regulations and corporate law.
>
> Which are excessive and idiotic. Overregulation is why corporations register
> overseas and the volume of trade on completely unregulated over-the-counter
> market is steadily growing.
That's funny, because the "overregulated" USA are still doing quite well
as far as living standards are concerned.
The SEC regulations are not only "good", they are necessary. They prevent
things like IPO scams and other forms of sophisticated fraud. Have you
seen "the boiler room" ? Well IPO scams are actually more sophisticated
than that, but fortunately, they are illegal.
> effects of regulations, Coase is unable to give one example of good
> regulation. And mind you, he started as socialist, not as flaming libertarian.
Environmental regulations are "good". They are good because corporation
(X) might "own" the factory and the land that it's on, but they don't own
the air I breathe, and they don't own the river or the sea. They
certainly shouild not be entitled to arbuitrarily trash shared reousrces.
Of course, if they want to seal off their chimneys and keep their smoke to
themselves, that's fine.
--
Donovan
------------------------------
From: [EMAIL PROTECTED] ()
Subject: Re: Why Jeff Szarka Has Zero Credibility When He Claims Problems With Linux
Date: Sat, 24 Jun 2000 06:54:34 GMT
On Fri, 23 Jun 2000 23:23:37 -0400, Jeff Szarka <[EMAIL PROTECTED]> wrote:
>On 24 Jun 2000 08:23:23 +0800, [EMAIL PROTECTED] (Terry
>Porter) wrote:
>
>>Well that must make my 17 year old son a "expert".
>
>17 year olds have lots of time to read How-To files.
More importantly, they are less intimdated by new things
and more open minded...
--
|||
/ | \
------------------------------
** 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
******************************
