Linux-Advocacy Digest #11, Volume #27 Sat, 10 Jun 00 16:13:06 EDT
Contents:
Re: Would a M$ Voluntary Split Save It? ("Daniel Johnson")
Re: Open Source Programmers Demonstrate Incompetence ([EMAIL PROTECTED])
Re: Linux faster than Windows? ([EMAIL PROTECTED])
Re: Linux faster than Windows? ([EMAIL PROTECTED])
Re: Linux faster than Windows? ([EMAIL PROTECTED])
----------------------------------------------------------------------------
From: "Daniel Johnson" <[EMAIL PROTECTED]>
Crossposted-To:
comp.os.os2.advocacy,comp.os.ms-windows.nt.advocacy,comp.sys.mac.advocacy
Subject: Re: Would a M$ Voluntary Split Save It?
Date: Sat, 10 Jun 2000 19:51:40 GMT
"Pascal Haakmat" <[EMAIL PROTECTED]> wrote in message
news:[EMAIL PROTECTED]...
> Daniel Johnson wrote:
> >> You must mean they are very good at bait-and-switch!
> >
> >No. I don't recall them advertising ont he basis of compatibility
> >either, so even if they didn't deliver it, it wouldn't be "bait and
> >switch". It would just be "switch". :D
>
> Sometimes I think you have a blind spot.
>
> http://www.microsoft.com/windows2000/guide/server/features/interop.asp
Well, I guess. They do have a page on their website listing all the
interoperability features; I 'spose you could call that advertisement.
But honestly, if that's "bait and switch", I suspect they'd have to keep
this a secret (shared only with their applications group :D ) to please
you.
[snip]
> >Well, I ask you how many *other* server OSes even go this far: how many
> >support Apple's protocols at all?
>
> Irrelevant to the possible issue of bait-and-switch, which we were
> discussing.
Well, we are only discussing that because you wanted to change the
subject, you know. I was saying how MS has lots of interoperability,
without reference to how they had advertised it.
Why don't I get to change the subject, too? :/
[snip]
> >If they are using VT100s, they are going to conclude that *anyway*.
>
> No, that is not true, and I don't see how you could possibly know.
Hmm. How could *you* know, if I can't?
> >I've used Window's telnet. It's pretty limited, but it is servicable.
>
> No, it's broken. The VT100 emulator breaks (among other things) the "talk"
> client and the telnet implementation cannot negotiate window size.
Hmm. You mean tcp/ip window size or display window size?
vt100's don't have any way to talk about their display size; it is
fixed at 80x25. Later DEC terminals do, but vt100s are pretty primitive.
The TCP/IP window should be handled in the TCP/IP stack, not the
telnet client. Or so I'd expect.
[snip]
> >No, it isn't. You may prefer one format. You may prefer one format
> >simply because you like its creator, or dislike the creator of the
> >other.
> >
> >But your preference in such things is no reason for *everyone* to use
> >your favored format.
>
> My preferences are not the issue. You were comparing the Word document
> format (a binary representation with largely ill-understood and hardly
> discoverable semantics)
Well, those of us who, unlike Leslie :D, are willing to read
MSDN and *believe* what it says can, in fact, discover the format.
> to the TeX document format (an ASCII representation
> with well-understood and easily discoverable semantics).
I personally do not know where to find the TeX document
format. This proves it is undocumented. :D
> Owing to their differing nature, the document formats sport differing
> characteristics. These characteristics are independant of my preferences.
Sure. But you haven't exactly made a case that Word's format is *bad*
on its 'characteristics'.
> In the hypothetical event that a fascist world leadership should arise and
> proclaim we must all use One Document Format, several characteristics of
the
> Word document format (storage of personal data, single-vendor
appropriation,
"appropriation"?
Is this just a fancy way of saying "Microsoft made it, and I don't like
that"?
> troublesome discoverability) make the TeX document format a better choice
> given the circumstances.
Well, TeX has its downsides as well. It is, for instance, a markup language
thing. Basically a text file with lots of tags in it. This is problematic if
you
have large files; to discover the formatting at any given point in the text
you must search the whole to look for tags that may surround it; they
can be anywhere. This pretty much means you have to bring the whole thing
into memory. Word's format does have this problem.
Also, Word uses structured storage, which means it is possible to
embed other (non Word) data in it, and even to extract that data again,
without having to interpret the word-specific stuff. Since many (non Word)
programs use binary storage, a text file format has a hard time dealing
with this.
Structured storage has other handy features. The structured storage format
is transacted, so that if you are working with a file over the network and
the
network goes down, your file is not corrupted. While a transacted filesystem
can provide this feature for all files, TeX documents included, these
are uncommon. Structured storage can do it on a FAT volume.
I do not say there are no nice things about TeX; only that it is not
transparently obvious that it is better than Word for word processing.
> However this is getting to be an increasingly irrelevant discussion.
> Firstly because I never claimed that everyone should use my favored
format.
> Secondly because if a fascist world leadership does arise, what document
> format it wants to use is going to be the least of my worries.
Well, I'm not sure what you want, if not that. Perhaps you are happy
the way things are; I was under the impression you though some
sort of villainy was afoot in this case.
If the wickedness is not "using a format other than my favorite",
I am at a loss to see what it *is*.
[snip]
> >It is still no reason for everyone to use TeX.
>
> It's your straw man. Pummel him as you please.
Ha! Take that you nasty, nasty straw man! <thwack> <crunch>
[snip]
> >> Not really. If I take an example like Coke or Microsoft, the bar of
> >> substitutability is so high not because of my standards, but because of
> >> their dominance in their respective markets.
> >
> >This sounds very like "I have a double standards; big companies
> >must meet a high standard because of this."
> >Did you mean it that way?
>
> It makes a lot of sense to distinguish between big companies and small
> companies or individuals. What this has to do with double standards
escapes
> me.
It *is* a double standard; you hold some to a higher standard than others.
It *could* be that this is well justified. But I think the case should be
made,
rather than assumed.
[snip]
> >I could make an argument that they *enhance* substitutability
> >by giving everyone a nice obvious target to try to clone.
>
> True.. In an effort to get away from the subject of economics as quickly
as
> possible I did make some unwarranted assumptions. Alas, I did not get away
> that easily..
There is no escape! We'll make an economist of you yet! :D
[snip]
> >No. If interoperability allows people to switch to the monopolies
> >product. That is why MS is so gung ho about it.
>
> Um. A monopoly seems to imply that most people are already using the
> monopolies product.
The conventional usage of the term would imply that.
But Microsoft is clearly not satisfied with 51%! :D
[snip]
> >Far, far better to head for new, untapped areas. Get innovative; come up
> >with something new. That's where the real money is, *and* where the
> >real excitement is.
>
> I understand what you are saying as an implicit admission that Microsoft
has
> cornered the OS market and there is no longer any excitement or money
there.
Sort of. It depends what you mean by "cornered". I suspect you intend
to imply that it is *bad* that they did this; I don't really agree with
*that*.
> "Let's move on" strikes me as a very pallid response.
Why?
> >What possible inducement could there be for a creative, driven,
> >ambitious individual to want to build a better Windows?
>
> None. Sad, isn't it.
Not really. It pushes them to head out to New Frontiers (tm),
and not just reinvent the bloody wheel over and over.
It encourages innovation- and I think that's good.
I don't see the point in doing Windows *again*.
One may worry that without this, MS will rest on their
laurels and stagnate. But this is not, in fact, what we
see happening. They seem to be afraid of becoming
obsolete.
[snip]
> >I very much disagree. Y ou have no right to make money selling it; that
> >necessarily implies that other people have an obligation to buy it and
> >pay for it, even if it tastes like Yoohoo, but looks like Mountain Dew.
> >
> >I do not accept that your 'rights' can impose such an obligation on
others.
>
> Well, let me rephrase that to read "you have the inalienable right to try
> and make money selling me sugarwater".
Okay. I'm not sure this is universally accepted; certainly *Microsoft's*
"inalienable right to try to make money" is not universally accepted. I'm
really not sure anyone's is, in the big wild world out there.
But I am prepared to accept it for the sake of argument. It agrees
with my prejudices. :D
> If you're going to be this literal, we are not going to get very far. Some
> days it is more difficult to find the right words than others and this
kind
> of response makes me wonder whether it is worth the effort.
I beg your pardon; but your comment made little sense. if interpreted
as "a right to try to make money". Microsoft hasn't even been accused of
making any efforts to stop you from *trying* to make money making and
selling OSes.
[snip]
> >"Decommoditization" favors variety; MS wants to be different
> >from the next guy, so you'll want their software instead of
> >his.
>
> We're not talking about the same thing. The variety comes about because
> widely disparate products can interoperate on the basis of a commoditized
> protocol.
I think you are mistaken: the variety does not, in fact, come about because
of that.
At least, I do not observer very much variety coming from such a source.
I see a lot more variety *outside* the open-standards world of Unix than
inside.
I see no reason to suppose that the use of common, standardized,
commoditized protocols and format will enhance variety anywhere.
> When you are in a position where you can more or less successfully dictate
> what protocols are being used, like Microsoft, decommoditizing the
protocol
> means that products can no longer interoperate, and the ensuing variety
> disappears.
Not actually. What MS wants to do is maintain the interoperability, and use
it to co-opt the customers of other products.
MS *could* modify the protocols they implement so they aren't
interoperable, but this is accomplishes nothing for them.
> IOW, when everyone wants to use Windows, how does this "favor variety"?
Well, the *reason* they want to use Windows is, in large part,
to expand the variety of peripherals and applications than they can use.
Now, the fact that "everyone wants to use Windows" does not
favor variety *in operating systems*, which I suspect is what you meant.
But I don't think I said it would.
[snip]
> >What popularity fo eye candy software?
>
> Screensavers, desktop animations, window dressing, skins, ...
Are those popular? I mean, compared to plain old vanilla Windows?
Sure, there's a market for them. There's a market for out-and-out
Linux, even. :D
[snip]
> >I don't either. MS is quite ruthless. But the *source* of this power is
> >MS's ability to woo customers (and developers) to Windows; only
> >once this is accomplished can threatening to withhold Windows
> >work.
> >
> >And after it is done, the rest is just a formality- Windows has won.
>
> Actually I think yours is an fascinating take on matters. I generally see
> Windows as an add-on to DOS and the transition from DOS to Windows as
> somewhat self-evident. But what you seem to be saying, namely that the
> transition was a battle that Microsoft simply fought well, is more
> interesting.
I'm glad I've entertained you. :D
At the time, it was OS/2 that was *supposed* to be the
"self evident" transition from DOS. These things are a lot
more 'self evident' in retrospect, really. :D
[snip]
> >Nope; the extortion (it's really a threat, not a bribe) has to come
*after*
> >you've got the customers. Otherwise Compaqs answer will be
> >"so what? We're shipping OS/2 anyway. That's what out customers
> >want."
>
> Well, Microsoft got the customers with DOS. I'll grant that this is only
> half the story, though.
OS/2 had excellent DOS compatibility, even better than
Windows, right across the board.
MS had to win customers *to Windows* to do what
they did. Winning them to DOS won't cut it.
[snip]
> >I think there *is* a market for a non MS OS that products like
> >Linux can fill; I just don't think it's the same market Windows 98
> >is filling.
>
> I don't think anything competes in the same market as Windows 98. People
> don't buy operating systems; they buy Windows. While there are some grey
> areas, generally speaking Windows 98 defines the market.
Windows 98 is the only OS that covers both the home and business
desktops the way it does. You can argue that it competes with MacOS,
but not really in the business world. You can argue that it
competes with BeOS, but not for home (ie, video game) users. And so on.
[snip]
> >Maybe.
>
> "Maybe" current software engineering practices cannot deliver a media
player
> for Linux???
Hey, I don't know all that much about Linux's internals.
> >Linux is not exactly a really media friendly sort of OS, when
> >you get right down to it. I'm not saying it's impossible, but
> >certainly it is a challenge.
>
> The problem as it stands is not so much display as decoding. It's not
> immediately apparent to me how that hinges on any real or perceived
> media-unfriendliness.
Decoding is, technically, usually pretty simple stuff. Legal
impediments are hardly relevant to the discussion we've been
having.
> >And there's only so much talent out there to meet that
> >and many, many other challenges.
> >
> >Why is that one the important one?
>
> Your question is a bit of a red herring.
IMHO, so was yours. :P
> It begs the question why, if talent is so limited and challenges so
> abundant, an incompatible and only marginally better solution was
developed
> in the first place.
Windows is fairly compatible, and *vasty* better than DOS.
A great deal of talent was used to make it so; but the need
was so blatant and so great that I am only surprised it
did not happen sooner.
[snip]
> >Forgive me, but I think you have confused them. You are saying
> >that Microsoft's "power"- whatever that was before Windows took
> >off- just suddenly *caused* Windows to succeed without so much
> >as a mechanism.
> >
> >Sure seems like magic.
>
> Microsoft could fumble about for years releasing at least two useless
> versions of Windows. Sure seems like power.
Seems like arrogance on the part of Apple, incompetance
on the part if Digital Research, and blindness of the part
of IBM.
There was a lot of time for some alternative to come along
that served the need acceptably well before 1991. It didn't.
The technical challenges are formidable, *not* insurmountable;
as I understand it there is *now* a product that can cope
with late eighties hardware: New Deal Office, which is
Yet Another Version of GeOS, the OS that *will not die*.
But it is much too late now; this product is only worthwhile on
very, very obsolete hardware.
In 1988, it would have taken the world by storm, especially
if they had released the API (which they don't now.)
[snip]
> >Thus, my "product quality" explaination can explain why *this*
> >Windows succeeded, and others had failed. I don't see how your
> >"power" explaination can do so.
>
> My "power" hypothesis does not need explanations. Instead it employs goons
> to beat up your hypothesis :)
:D
> Seriously, your product quality hypothesis does not explain why the
> Macintosh or the Atari or the Amiga or the Acorn, all quite capable
systems
> in their own respect, did not succeed.
The Macintosh failed because Apple priced it out of the market.
The Amiga was a glorified games machine that found a niche
in video editing and the like. But it did not solve or attempt to solve
the same set of problems that the Macintosh, and later Windows, did.
I do not know very much of the Amiga and the Acorn.
> One explanation is hardware. All the other systems were more or less
closed.
> The PC architecture has always been relatively open.
But not "open" in the same sense as in "open standards"; there is no
beauracracy in control of PC hardware, no committee to be
propitiated.
[snip]
> >They made a pretty darn clean break, actually. "You all have to program
for
> >PowerPC now. Have a nice day."
>
> Depends how you look at it. I was thinking more along the lines of clean
> break == dropping backward compatibility.
Well, I think the users would have found that rather messy. :D
[snip]
> >Yet they do worry; they bend into pretzels for backwards compibility.
>
> I think that is always true of any kind of backwards compatibility. It's
> simply hard.
Yes. But MS bothers. Surely with all their Power (tm) they could just
forget about backwards compatibilty and *make* everyone
upgrade to whatever product they want.
Like, say Windows NT.
[snip]
> >They were not able to. They lacked the power to make developers swtich
> >suddenly. They had to try to lure them over, and they ahd mixed success.
> >They were a *long* time wooing the game developers, in particular.
> >
> >In the meantime, they had to rely on compatibility.
>
> They lacked the power to make developers switch but they could afford to
> spend a "*long* time" wooing developers?
Yes.
There is no conflict in that statement, ya know. "Power" and "longevity"
are not the same thing. MS already had several cash cows: DOS,
and Macintosh applications like Word. They had plenty of time.
[snip]
> >I'm not that familiar with Atari's product. I recall vaguely that
> >they had a bad case of 'too little, too late'- they went to 16 bit
> >very late, and by then it was all old news.
>
> The Atari ST appeared in 1985 or the beginning of 1986. It was a sixteen
bit
> machine based on the 68000 chip.
That's kinda late. There may be more to this that I don't know about.
> >"It wasn't Microsoft branded" doesn't explain why the MAc did
> >so much better than the Atari, or what the Commodore Amiga
> >also did.
> >
> >I was thinking of GEM for DOS, of course. That was like
> >Windows. Like early, useless Windows. :D
>
> I understand. But there were alternatives. Heck, there was even a Sony
MSX2
> system that had a desktop, trashcan, floppy disk icons ...
Well, there are alternatives *now*, too.
There are usually good reasons why these alternatives stay
'alternative'. :D
[snip]
> >I dunno. Maybe there were too expensive too. :D
>
> No, they weren't. It's a pity I don't recall the going prices for PC's,
> Macs, Amiga's and ST's around that time.
I fear we are arguing in a vacuum of information on this subject.
[snip]
> >No, no, it was just inadquate, in the same way and for
> >the reasons Windows 1 and 2 were. It was dang near
> >impossible to write real software with the memory
> >constraints you faced.
>
> An architectural problem that did not exist on, for example, the Atari ST.
> Another blow to your product quality hypothesis :)
It didn't exist with Windows 3, either; that was the *big* thing that
changed:
Windows 3 fixed the memory problem.
It is true that earlier Windows were rather poorly off next to
anything with a 68000 in this area. But then, that's why
they couldn't give the fool thing away. :D
[snip]
------------------------------
From: [EMAIL PROTECTED]
Subject: Re: Open Source Programmers Demonstrate Incompetence
Reply-To: [EMAIL PROTECTED]
Date: Sat, 10 Jun 2000 19:54:20 GMT
Cihl <[EMAIL PROTECTED]> writes:
>Sorry to burst your closed-source bubble, but closed-source software
>is no better in this matter, and possibly much worse too, because
>nobody else but the company who made the program ever looks at them!
One example would be reading a directory in the Workbench under AmigaOS.
It works fine as long as your directories are small. But when they get
large.....
The reason is that AmigaOS tends to like linked lists, and in this case
sorted linked lists. Now, how to create a linked list of all the filenames?
Here is what they do (in pseudocode):
list=empty;
while (name=getNextFilename()) {
list2=&list;
while (*list2 && strcasecmp((*list2)->name,name)>0)
list2=&((*list2)->next);
Insert(list2,name);
}
Uh-oh --- an algorithm of order O(n^2). Maybe not such a good idea. And
it *really* isn't that hard to do the same operation in O(n*log n)/
Oh, and as for doing silly things --- the above left out the check for
whether a filename ends in ".info". If it does, it is compared to all
sorts of strings, including "CONSOLE:" (or something very much like that).
Now, how many strings ending in ".info" will ever match "CONSOLE:"?
Bernie
--
The world will never starve for a want of wonders, but only for want of
wonder
G.K. Chesterton
------------------------------
From: [EMAIL PROTECTED]
Subject: Re: Linux faster than Windows?
Reply-To: [EMAIL PROTECTED]
Date: Sat, 10 Jun 2000 19:54:22 GMT
[EMAIL PROTECTED] (Pete Goodwin) writes:
>One of the claims made here was the Linux file system is so much better
>than Windows - yet this test reveals they're about the same.
On neither system does the time measured by your "benchmark" include any
actual disk access --- unless you are *really* low on memory. I'd suspect
that almost all of the time is actually spent in the fprintf call, parsing
and preparing the string.
You can see this yourself --- simply replace the fprintf with an sprintf
into a previosuly allocated array. Now the filesystem is completely uninvolved.
See how little this changes the running time of the "benchmark".
Bernie
--
Few people can be happy unless they hate some other person,
nation, or creed
Bertrand Russel
British philosopher and mathematician, 1872-1970
------------------------------
From: [EMAIL PROTECTED]
Subject: Re: Linux faster than Windows?
Reply-To: [EMAIL PROTECTED]
Date: Sat, 10 Jun 2000 19:54:23 GMT
[EMAIL PROTECTED] (Pete Goodwin) writes:
>[EMAIL PROTECTED] wrote in <8hse0u$2tu$[EMAIL PROTECTED]>:
>>A lot of Linux zelaots may not realize it, but when you use
>>Linux, and have compiled your programs with GCC, you are getting much
>>less out of your hardware than if you the software you run is built
>>with a good compiler.
>That is interesting, Linux is _slower_ than Windows 98 SE.
Uhm, *that* is quite reaching.
Certainly gcc doesn't do a very good job for the Alpha, and using Compaq's
CCC instead gives considerably faster programs. But Win98SE doesn't run
on Alphas, and Compaq's CCC isn't available for it (it *is* available for
linux, though).
Now, on Intel, it seems things are much less clear. A while ago,
someone built the Bytemark benchmarks with some really fancy compiler
optimizations under NT, using some commercial compiler from either
Intel or MS (I could look it up if anyone needs more details). I built
the same code with a recent gcc. It turns out that for some of the
tests, the gcc code wipes the floor with the commercial compiler's
code, and for some other tests, it's the other way around. The
benchmark averages came out pretty much identical --- which just goes
to show that both compilers can still be improved upon.
Anyone around who could compile and run a bit of obfuscated C with commercial
in order to do some benchmarking?
>So much for the outrageous claims of a three fold speed increase!
I have been meaning to ask this --- could you provide a reference for that
claim? I don't seem to recall it, so I would like to check whether I missed
it or whether we interpreted something I saw differently. A message ID will
be enough. Thank you.
Bernie
--
Men enter politics solely as a result of being unhappily married
C.N. Parkinson
English writer, 1909--
Parkinson's Law, 1958
------------------------------
From: [EMAIL PROTECTED]
Subject: Re: Linux faster than Windows?
Reply-To: [EMAIL PROTECTED]
Date: Sat, 10 Jun 2000 19:54:25 GMT
[EMAIL PROTECTED] writes:
>Another interesting benchmark would be mail programs. I use Outlook,
>which has a sophisticated indexed system. My mail box has more than
>5,000 messages, and Outlook can open this mailbox (on a dual Pentium
>133, my main Windows workstation), in less than 2 seconds. Unix mail -
>which uses nothing but a huge text file, and has to parse through every
>byte (!) to open the mailbox, is very slow. On fast Unix systems, such
>as Sparc's, it takes several _minutes_ to open a mailbox of 200-300
>messages (compared to 2 seconds for a mailbox over 10x bigger on
>Windows).
Now now, let's be reasonable here. I don't know what mailer you use on
those "fast Unix systems", or whether you store your mail locally on the
machine with the mailer --- but things certainly don't have to be nearly
as bad as that.
My system used for outside interaction is a dual P150, i.e. very similar
to yours. For mail, it runs a recent version of elm. Starting it with
6438 messages in my mailbox, for a total of 37MB, takes 25 seconds,
measured.
Opening the same mailbox in mutt takes roughly 22 seconds, including the
sorting and threading.
>Mailboxes over that length are absolutely unwieldy for Unix,
>but no problem with a sophisticated mailer (security issues
>notwithstanding).
That is simply not true --- I only ran into trouble when my mailbox
approached 30,000 mails and 100+M. The reason I ran into trouble was
that elm occasionally copies the user's mailbox file, and the process
took long enough for elm to think that there was a filesystem problem.
In the 6438 mail mailbox, searching for a subject or author takes less than
half a second. Searching through all the bodies (i.e. 37M) for a 10 character
string takes about a minute.
Bernie
--
To win without risk is to triumph without glory.
Pierre Corneille
------------------------------
** 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
******************************
