Linux-Advocacy Digest #429, Volume #25 Mon, 28 Feb 00 14:13:05 EST
Contents:
Re: My Windows 2000 experience (mlw)
Re: My Windows 2000 experience (mlw)
Re: My Windows 2000 experience (mlw)
Re: Microsoft migrates Hotmail to W2K ("Drestin Black")
Re: Microsoft's New Motto (was: TPC-C Results for W2k!! (Jason McNorton)
Re: My Windows 2000 experience ("Drestin Black")
Re: Windows Network Configuratin? ("Drestin Black")
Re: Windows Network Configuratin? ("Drestin Black")
----------------------------------------------------------------------------
From: mlw <[EMAIL PROTECTED]>
Crossposted-To: comp.os.ms-windows.nt.advocacy
Subject: Re: My Windows 2000 experience
Date: Sun, 27 Feb 2000 11:16:19 -0500
Erik Funkenbusch wrote:
>
> mlw <[EMAIL PROTECTED]> wrote in message
> news:[EMAIL PROTECTED]...
> > However, SCSI ID 0 and 1 on BIOS additions provided by SCSI adapters are
> > defined as BIOS Drive 80 and 81 by convention, when no IDE drives are
> > installed.
>
> Irrelevant. We're not talking about the boot device.
Then, why would (the original topic of conversation) would the boot disk
be affected by adding/removing a zip disk?
>
> > The issue is when does an OS rely on the BIOS and when does an OS rely
> > on peripheral technology. The IBM BIOS is universally held to be
> > inadequate for anything but a boot loader. Every decent protected mode
> > OS circumvents the IBM PC BIOS because it can't run in protected mode.
>
> Every boot loader i've ever seen uses the BIOS to enumerate drives. LILO
> does this, IBM's Boot Manager does this. System Commander Does this. And
> NTLDR does this. The reason is simple. Individual controllers have
> different interfaces to them. The only way to guarantee enumeration is
> through the BIOS.
Boot leaders to not force an enumeration at boot time, the enumeration
occurs in the bios load of the peripheral. The main system bios reads
the first sector of the first bootable device, as defined by bios
settings. The OS loader has no way of effecting a change in this
process. At the point in which a boot loader is called, it is usually up
to the OS to decide the numbering scheme of the disks.
>
> > The BIOS number is irrelevant. Once booted, an OS that circumvents the
> > BIOS (NT, BSD, Linux, Win2K, Solaris, etc. Not 9x!) must load the
> > drivers for the installed SCSI adapters and enumerate all the devices on
> > the SCSI bus by ID. Then load the 'class' driver for that type of
> > device. An OS, keeping track of SCSI ID need not worry about additions
> > or subtractions in the bus.
>
> We're not talking about "once booted". We're talking about the boot
> process, specifically the boot loader.
The boot loader is called "after" the peripheral cards have enumerated
their devices.
>
> > On UNIX, one has a table, fstab, that tells how to mount drives.
> > "/dev/sd0a" represents SCSI drive 0, "/dev/sd1" it is unambiguous as to
> > how the driver are located and identified.
>
> Shouldn't that be sda0? and sdb0 for the second? And isn't it true that
> these names are equivelant to the BIOS enumerated drive# and not the
> physical SCSI ID#?
Yup, you are right, this is something that is broken in Linux too.
>
> > If you look into WinNT you will see they have a similar scheme. Why, on
> > earth, they will use BIOS enumeration over device enumeration is a
> > mystery. All I can think is that they want to be some what compatible
> > with Windows 98, but still, I would call this, on a workstation OS, a
> > bug.
>
> No. It's because Intel based hardware is not standardized. The BIOS is the
> only way to guarantee compatibility.
Actually the "bios" is probably the wrong way to assure compatibility.
If you have two ISA based SCSI controllers, it is on a first come first
serve order in which drive 0x80 will be, with IDE almost always winning
unless the SCSI adapter (Adaptec) allows one to override this behavior.
On PCI, with two different types of SCSI adapters, it is on a first
enumerated by system BIOS order. Then there is when you have similar
SCSI adapters that is a PCI enumeration order.
This is not a perfect system, obviously, but it does work.
FYW, The P.C. architecture is very poorly designed with regards to
peripherals. There are too few interrupts, the interrupt handling is
quite poor, the bios scan/load process is cleaver, in a 1970's sort of
way, and the memory bus designs are probably the worst in the high
technology arena.
I am frequently amazed at the lengths people go to do 'something' with a
P.C. that would be easier with almost any other design. It is just not,
technically, a great platform for doing some things. On the other hand
it has been a platform that has remained cheap, in spite of the constant
superhuman feats of engineering required to keep, what is effectively a
8 bit computer design, modern.
--
Mohawk Software
Windows 95, Windows NT, UNIX, Linux. Applications, drivers, support.
Visit http://www.mohawksoft.com
------------------------------
From: mlw <[EMAIL PROTECTED]>
Crossposted-To: comp.os.ms-windows.nt.advocacy
Subject: Re: My Windows 2000 experience
Date: Sun, 27 Feb 2000 11:28:06 -0500
fred wrote:
>
> On Sun, 20 Feb 2000 14:49:21 -0500, mlw <[EMAIL PROTECTED]> wrote:
>
> >How is that a mistake? I have written several SCSI drivers, and that is
> >what the SCSI ID is for. You do not know about what you are talking
> >about on this one.
>
> On the other hand, I think you're really reaching to try to prove a
> point.
>
> Operating systems on the x86 architecture abstract themselves
> somewhat from the type of drive controller when you are utilizing the
> SCSI BIOS.
>
> As far as Windows, or even OS/2 or whatever is concerned... you have
> Drive #1, Drive #2, Drive #3, etc. Whether it be IDE, SCSI, etc.
>
> That's the real issue here.
This is not true. Excepting Windows 9x, because it is DOS and can use
the BIOS, protected mode operating systems can not call the bios while
in protected mode. They have to either make bios class prior to the flip
into protected mode, or created a DOS virtual machine (>= 386) and wrap
a working real mode environment around the bios and "reflect" interrupts
in to it.
Typically what happens is that the system bios, after loading all the
peripheral bioses, calls the boot sector of drive 0x80 (which ever that
turned out to be) and then the OS will boot and enumerate the drives on
its own. A drive is typically defined by (adapter)/(id) or in NT,
(adaper)\(id) take a look at boot time.
--
Mohawk Software
Windows 95, Windows NT, UNIX, Linux. Applications, drivers, support.
Visit http://www.mohawksoft.com
------------------------------
From: mlw <[EMAIL PROTECTED]>
Crossposted-To: comp.os.ms-windows.nt.advocacy
Subject: Re: My Windows 2000 experience
Date: Sun, 27 Feb 2000 11:21:26 -0500
Drestin Black wrote:
>
> "mlw" <[EMAIL PROTECTED]> wrote in message
> news:[EMAIL PROTECTED]...
> > > Yes, it does identify the device. But such ID's are transient, since
> they
> > > can change. They do not *PERMANENTLY* identify a device.
> >
> > SCSI ID's are NOT transient. One has to assign or reassign a SCSI ID.
> > SCSI does not enumerate the bus and assign IDs dynamically.
>
> Is THAT your final answer?
Actually no, I was reminded of SCAM. I have to say, I was wrong about
the SCSI addresses being non-enumerated. I had forgotten all about it.
However, when I first saw SCAM, I put it in the same league as p&p isa.
A curious thing that could be useful for people who put things other
than disks on a SCSI bus.
I can honestly say I have never used it, and quite frankly will probably
never use it. SCSI is excellent for disks, however, and aside from the
obvious speed benefits, it is a poor interface for peripherals.
While I do stand corrected on the ability of SCSI IDs to be enumerated
and assigned automatically, I assert that it is not something one
would/should do with hard disks.
--
Mohawk Software
Windows 95, Windows NT, UNIX, Linux. Applications, drivers, support.
Visit http://www.mohawksoft.com
------------------------------
From: "Drestin Black" <[EMAIL PROTECTED]>
Crossposted-To: comp.os.ms-windows.nt.advocacy
Subject: Re: Microsoft migrates Hotmail to W2K
Date: Mon, 28 Feb 2000 13:56:16 -0500
"Wolfgang Weisselberg" <[EMAIL PROTECTED]> wrote in
message news:[EMAIL PROTECTED]...
> On Sat, 26 Feb 2000 22:15:22 -0500,
> Drestin Black <[EMAIL PROTECTED]> wrote:
> > OK, my reply at the top because there is no point in replying line by
line.
>
> Then you could have shortened my posting ...
>
> > Look, you are trying to say that simply because MS has not ported
> > hotmail.com to NT that NT is a failure.
>
> No. I am saying it's not as usable (and as much a success) as MS
> tries to make others believe.
I do not agree.
>
> > Can you not see the stupidity of that remark? What does one site have to
do
> > with the quality or success of NT? When you know that 1000s of other
sites,
> > many of which are much bigger than hotmail are running NT
>
> Well, depends on your measurement. But don't marginalise
> Hotmail: It's an MS-owned shop for two years now ... and still NT
> is not the primary server. Just one amongst others. That's
> undeniable, or is it?
"Just one amongst others." What others? Hotmail is a property owned by MS,
yes. But, when someone buys property it does not always follow that they
will then immediately take and strip it down to it's very core (OS and
Application), rewrite it all over again and carry on. If they had wanted to
do that, why not just build their own hotmail from the ground up? name
recognition is the answer to that. So, why convert at all? Let's put it this
way: Not everything that MS owns or has shares in runs Windows on every
single screen - hotmail is not an exception either. Gates owns 51% of a
naval shipyard, I'm sure there are non-windows computers in that operation
somewhere, does the fact that not 100% of this new aquisition of his
suddenly mean that his faith in his product is soiled? Of course not. So,
why apply this to hotmail? People are focused on hotmail not becuase it's
the biggest or best or fastest or anything - other than it's the only bigger
item MS bought that doesn't run an MS OS. Yet.
>
> The lesson here is: If you run Solaris and do something remotely
> similar to Hotmail, you don't need NT ...
I do not agree.
>
> > - what is the
> > point of bringing up hotmail again? NT is capable of handling MORE than
> > hotmail - but hotmail was bought in a working state by MS. "If it ain't
> > broke, why fix it?"
>
> Cause MS said they'll do so. Now, if they had said "Hey, it runs
> OK under Solaris or FreeBSD, no sense changing it!" I'd
> understand. (It wouldn't be too good for their marketing,
> though.) But they did not. They proclaimed they were migrating
> --- albeit slowly --- to NT. Now, i am just calling their bluff,
> so to say.
I think you'll find that bluff called... before 2000 is out.
>
> MS can now say one of:
> - There are other OS that can do this, actually we are happy
> enough with Solaris
> - We did not manage to port Hotmail to NT.
> - Porting Hotmail to NT turned out to be more hassle than it's
> worth.
> - Actually, we lied (again) ... we are going to do $WHATEVER.
>
> Somehow I cannot see them doing any of this. Can you?
Well, the first is possible but very unlikely. I agree.
>
> > Having said that I tell you that I expect hotmail to be running W2K
before
> > the end of the year.
>
> Wait and see. If the specialists need half a year, how muh longer
> will the average admin-team need?
Again - IF they had been actively TRYING to over half a year and were still
stuck - then, yes, there are problems. However, my information is clear: no
one has even tried at all. But, as I said... before 2000 is out... that's
all I can say.
>
> > But, again, what is the point? So, we announce: Hotmail
> > is running W2K and is now faster than ever before.
>
> And the W2k licenses alone cost more than the old licenses *plus*
> hardware, not to mention that you need 5 times the machines?
License cost is not a concern to MS - wouldn't you think? :)
5 times the machine? Not according to recent TPC benchmarks where using W2K
MS was able to extra 67% greater performance for 1/3rd the price over the
very best closest comers using any hardware and any os. All comers were
beaten by 67% for 1/3rd the price. And that is on their first attempt using
beta software. Wait till the bugs are worked out and debug code is removed -
expect a new higher record benchmark.
> (Note: I'm not saying that this *will* happen. Just that it
> could. As with any other OS, including Linux[1].) With enough
> money you can always be faster, but sometimes thats not the goal.
Not if you are Sun and Oracle - you can spend 3, 6 and even 24 times the
money and you will still be beaten by anywhere from 67% to 4400% by the
smaller, less expensive, easier machines.
>
> > What will be the reply?
> > "Oh my, that proves NT is perfect, I will abandon BSD/Solaris for NT
> > immediately!" - Of course not. So, again, what is the point? Why does it
> > matter?
>
<snip>
> > You are careful not to say: W2K can't handle Hotmail -
> > only that: How come MS didn't port to W2K yet?
>
> Yes, because unlike some I try not to say things that might turn
> out wrong ... NT probably can handle Hotmail. If HotmailA's
> still usable and how to administrate all the dozends of
> clients is another question.
I applaud you for not being a typical sing-a-long anti-MS type. However,
"administrate all the dozends of clients"? Huh?
>
> BTW: I do *not* wonder about W2K. MS probably knows that it's not
> tested enough (and by the collective previous experience with MS:
> debugged enough) to handle Hotmail. After all, SP1 is already
> out, isn't it? Give it half a year or 2 to ripen.
SP1 is not out. year or two will have another version out so that's silly.
MS already knows from the huge numbers of people already using W2K that it's
plenty tested and rebust and solid enough. Listen... What you are not
hearing are the bug reports flooding in. Why? There aren't any (well, you
know what I mean, there are some here and there, little one here and little
one there, no showstoppers thats for sure!). You'd think if there were
63,000 bugs someone would have found some of them by now eh?
It's a fact anti-MS types are having a hard time grasping and coming to
terms with. MS got it right finally. W2K really IS much more reliable and
stable and performs damn well! I have not heard of any BSODs anywhere from
anyone. Period. I have not heard of crashes. I do not hear of "reboot
daily." I'm hearing NOTHING like I heard when NT4 was first released (or
even up to SP4). People are upgrading NT4 boxes to W2K and things just
work... better... You don't think if there were a zillion bugs out there
that the vehmenetly anti-MS groupies would be screaming about them at the
top of their lungs? I mean we've got obvious trolls like petlion (or
whatever his fake name is) trying to make up SCSI stores when he doesn't
even know that the "bug" he describes exists on his own OS of choice? (cause
it's not a bug).
W2K isn't perfect. Nothing is. NT4 was good. W2K is much better. VERY much
better. It can still be better. It has bugs, like any software ever written.
I'm sure they'll find a security hole, I would be suprised if they didn't.
I'm sure they will find some ugly bug in there somewhere - but I'm equally
sure that MS will fix it quickly and it'll continue to work. And I'm sure
we'll see more and more W2K servers out there.
:()
------------------------------
From: Jason McNorton <[EMAIL PROTECTED]>
Crossposted-To: comp.os.ms-windows.nt.advocacy,comp.sys.mac.advocacy,comp.unix.advocacy
Subject: Re: Microsoft's New Motto (was: TPC-C Results for W2k!!
Date: Mon, 28 Feb 2000 12:50:57 -0600
In article Joe Ragosta, [EMAIL PROTECTED] says...
> Get $10 free:
> https://secure.paypal.com/auction/pal=jragosta%40earthlink.net
> Or get paid to browse the web (Mac or PC):
> http://www.alladvantage.com/home.asp?refid=KJS595
Pretty shameless. Can you squeeze anymore spam into your posts?
------------------------------
From: "Drestin Black" <[EMAIL PROTECTED]>
Crossposted-To: comp.os.ms-windows.nt.advocacy
Subject: Re: My Windows 2000 experience
Date: Mon, 28 Feb 2000 14:01:53 -0500
"mlw" <[EMAIL PROTECTED]> wrote in message
news:[EMAIL PROTECTED]...
> Drestin Black wrote:
> >
> > "mlw" <[EMAIL PROTECTED]> wrote in message
> > news:[EMAIL PROTECTED]...
> > > > Yes, it does identify the device. But such ID's are transient,
since
> > they
> > > > can change. They do not *PERMANENTLY* identify a device.
> > >
> > > SCSI ID's are NOT transient. One has to assign or reassign a SCSI ID.
> > > SCSI does not enumerate the bus and assign IDs dynamically.
> >
> > Is THAT your final answer?
>
> Actually no, I was reminded of SCAM. I have to say, I was wrong about
> the SCSI addresses being non-enumerated. I had forgotten all about it.
> However, when I first saw SCAM, I put it in the same league as p&p isa.
> A curious thing that could be useful for people who put things other
> than disks on a SCSI bus.
>
> I can honestly say I have never used it, and quite frankly will probably
> never use it. SCSI is excellent for disks, however, and aside from the
> obvious speed benefits, it is a poor interface for peripherals.
>
> While I do stand corrected on the ability of SCSI IDs to be enumerated
> and assigned automatically, I assert that it is not something one
> would/should do with hard disks.
Well, SCAM was intended when there was hope the general populous was moving
to SCSI, just before IDE got popular. They wanted a way to make it easy to
just plug devices into the SCSI chain and off you go.
Who cares about SCSI IDs if they are automatically assigned? (you can mix
fixed IDs in there too, BTW, because SCAM would presearch for empty IDs
before assigning them).
What is wrong with the idea of SCAM? If the device identifies itself to a
plug'n'play OS what does anyone care which ID the device uses? This is the
kind of info that can be safely hidden from the user.
Think about it: Users just plug SCSI devices in - no concern about
termination, where on the cable, what ID - just plug it in and the OS sees
it. What is wrong with that?
------------------------------
From: "Drestin Black" <[EMAIL PROTECTED]>
Subject: Re: Windows Network Configuratin?
Date: Mon, 28 Feb 2000 14:05:18 -0500
you could use hardware profiles and just pick the appropriate one when you
boot up.
Also, since you'll be using Windows 2000 Professional (of course) you can
just make the changes on the fly - without rebooting to doing anything other
than changing the values in the properties dialog and clicking on OK. W2K
requires no rebooting to change networking, easier than linux IMHO.
"Bill Shine" <[EMAIL PROTECTED]> wrote in message
news:00022721231100.02859@Penguin...
> I'd like to use Windows NT on a laptop, but I can't, because I can't
> reconfigure the network settings without a reboot and a lot of fiddling
around
> when I hook it up to different networks, because the IP address, the
gateway,
> and the DNS servers are different. This is very easy to do on Linux. How
does
> one go about doing it on windows?
------------------------------
From: "Drestin Black" <[EMAIL PROTECTED]>
Subject: Re: Windows Network Configuratin?
Date: Mon, 28 Feb 2000 14:05:40 -0500
why not ask how to do this in a Windows group? is this some form of reverse
troll? interesting...
"Bill Shine" <[EMAIL PROTECTED]> wrote in message
news:00022721231100.02859@Penguin...
> I'd like to use Windows NT on a laptop, but I can't, because I can't
> reconfigure the network settings without a reboot and a lot of fiddling
around
> when I hook it up to different networks, because the IP address, the
gateway,
> and the DNS servers are different. This is very easy to do on Linux. How
does
> one go about doing it on windows?
------------------------------
** 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
******************************
