Archived from groups: alt.comp.hardware.homebuilt (
More info?)
Michael Thomas wrote:
> On Wed, 09 Feb 2005 17:14:09 -0600, David Maynard <dNOTmayn@ev1.net>
> wrote:
>
>
>>Michael Thomas wrote:
>>
>>
>>>On Tue, 08 Feb 2005 21:24:51 -0600, David Maynard <dNOTmayn@ev1.net>
>>>wrote:
>>>
>>>
>>>
>>>>Michael Thomas wrote:
>>>>
>>>>
>>>>>On Tue, 08 Feb 2005 17:27:30 GMT, "joanne grint"
>>>>><joanne.grint@ntlworld.com> wrote:
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>>Hi, Sorry but I don't understand what difference would this card make to the
>>>>>>installed devices and drivers I have. Why does it allow you not to remove
>>>>>>things?
>>>>>>
>>>>>
>>>>>
>>>>>Windows NT/2000/XP/2003 etc. unfortunately doesn't use a generic disk
>>>>>driver when it loads,
>>>>
>>>>Actually, it does. And then switches to the installed driver when drivers
>>>>are loaded.
>>>>
>>>>
>>>>
>>>>>to completion.
>>>>
>>>What I meant to say was it doesn't use the generic driver to
>>>completion, not that it didn't use one at all. It obviously does or
>>>it wouldn't get to the point of booting the OS at all. I should have
>>>been more clear.
>>
>>I was just clarifying.
>>
>>
>>
>>>>There wouldn't be much use for optimized and hardware specific drivers if
>>>>it used the generic ones 'to completion'.
>>>>
>>>>
>>>>
>>>>>If it did, you wouldn't have to
>>>>>sweat any of this and it would work just like Windows 95/98/Me - as in
>>>>>finding the hard disk, booting off it, and plug-n-playing all the new
>>>>>devices.
>>>>
>>>>As far as starting with the generic driver, and then loading in the
>>>>hardware specific one when drivers are loaded, it *does* work 'just like'
>>>>Windows 9x.
>>>>
>>>
>>>
>>>Up to the point when it tries to load a specific driver and then
>>>chokes if it can't find it.
>>>
>>>
>>>
>>>>Windows 2000/XP, however, supports an infinitely more complex hardware
>>>>configuration than Win9x does. SMP, multiple PCI busses, device
>>>>serialization, just to name a few. And it's not feasible to dynamically
>>>>reconfigure the core components, as those are, on every boot. Especially
>>>>since that is a rare case.
>>>>
>>>>
>>>
>>>
>>>Too bad it's not smart enough to bypass the "specific driver" when it
>>>can't match one to the hardware.
>>
>>It isn't a matter of *just* 'the specific driver'. There's an entire system
>>configuration involved.
>>
>
> Are you saying that a generic driver wouldn't be able to boot the
> system all the way into windows?
I'm saying the system configuration is a heck of a lot more than an IDE
controller.
>>> It would save people a lot of grief
>>>when they change to new hardware.
>>
>>Which is a rare case and there is a perfectly good mechanism for handling
>>it: a repair install.
>>
>
> That's is where you and I differ, I don't consider a repair install to
> be saving people "a lot of grief."
I didn't say a thing about 'grief', one way or the other. I said it was a
rare case and a perfectly good mechanism exists to handle it.
>>Not to mention that simply 'bypassing' the specific IDE controller wouldn't
>>take care of it as the entire hardware configuration is (potentially)
>>different.
>>
>
>
> And I can tell you I've had great success moving a PCI controller/HDD
> over to a brand new, considerably different motherboard, and had no
> problems with XP configuring the new hardware.
I'm glad you were lucky.
>>Why would anyone go to extraordinary lengths providing a solution that
>>doesn't solve the problem when one that does already exists?
>>
>
>
> Because it works?
You're changing topics and pretending they're the same. Moving a controller
so you have an already installed driver is not the same thing as expecting
Windows 2000/XP to magically work with the wrong drivers.
> I don't see how it is "extraordinary" at all.
Write your own operating system and then get back with how 'un'
extraordinary it is.
> I
> consider a repair install to "extraordinary lengths."
Certainly nothing compared to writing an operating system so it
automagically reconfigures itself on the fly every time it boots on the off
chance someone picked today to completely change the entire configuration.
>>>>>NT/2000/XP/2003 uses specific disk controller drivers it has
>>>>>installed in \%windir%\system32\drivers. What adding the Promise disk
>>>>>controller does is give you a portable controller you can move from
>>>>>one system to the other, or from one motherboard to another to be more
>>>>>specific. This works because adding the controller into the existing
>>>>>system installs the driver on that system. Your system doesn't know
>>>>>beans about your new motherboards disk controller, and it will "blue
>>>>>screen" when it gets to the point of using the controller driver,
>>>>>doesn't have one, and can't see the hard disk anymore.
>>>>>
>>>>>I know, you ask: Why does it boot from the drive just fine, then all
>>>>>of a sudden decide to use a specific device driver in the middle of
>>>>>booting? You'll have to ask the guys in Redmond about that one.
>>>>
>>>>Simple. It starts with the generic driver, since at initial boot it doesn't
>>>>know what the specific configuration is, and then loads in the hardware
>>>>specific drivers along with all the other hardware drivers. If the drivers
>>>>do not match the hardware then it will hang when they are loaded because,
>>>>all of a sudden, there is no working IDE controller anymore: wrong driver.
>>>>
>>>>
>>>
>>>
>>>What I meant was - why Redmond left out the ability to continue to
>>>boot off the generic driver when it didn't find the specific driver
>>>was a mystery to me.
>>
>>Because, for one, it doesn't solve anything.
>>
>
>
> It would solve the original posters dilemma - since they no longer
> have a CD to do the "repair" install with.
And it would be nice if Windows would unscrew the old motherboard and
install the new one too but I don't expect it to nor do I expect it to fix
a lost or damaged install CD.
I feel for the guy but it isn't an O.S. 'problem'.
>>> It would save people a lot of trouble in the
>>>event of a motherboard replacement,
>>
>>No, because a repair install is still needed.
>>
>
> No it's not. There are other options - one of which is to buy a PCI
> IDE controller card.
You're assumption that because you got lucky that means it will work in
every case is rash and bad logic.
>>>or a corrupt driver for that
>>>matter.
>>
>>Mechanisms to 'fix' a corrupt driver already exist.
>>
>>
>
> Not for the OP. They've don't have a CD.
That's not an O.S. 'problem'.
>>>>>Another option I have heard works is to add the specific driver to
>>>>>\%windir%\system32\drivers your new motherboard will be using. Thing
>>>>>is, you better get it right because once you tear out the old
>>>>>motherboard, you won't want to put everything back in if it doesn't
>>>>>work.
>>>>
>>>>The thing to do is a repair reinstall so the hardware gets redetected and
>>>>configured properly.
>>>>
>>>
>>>
>>>That's your take on how to do it.
>>
>>Not just 'my take'. It's based on how HAL works.
>
>
> It's not the only way to handle the situation. If it was then every
> time you installed a new device, such as a sound card or network card,
> your advise would be to run a repair install.
No, because those are not core HAL components.
>>> A repair install may fix the
>>>problem, but it also requires you to install a service pack and fixes
>>>afterwards.
>>
>>Yep.
>>
>>
>>>Personally I've been very successful with using a PCI disk controller
>>>(the above mentioned Promise Ultra100/133tx2) to move the system to a
>>>new motherboard, and I didn't have to do anything except install any
>>>new device drivers that windows didn't already have.
>>
>>Then you were either lucky or things may not be working as well as they
>>might otherwise because there's a heck of a lot more to a motherboard, and
>>supporting drivers, than just the IDE controller.
>>
>
>
> I wasn't lucky
And what do you base that assumption on?
> and things are working just fine. Just because you
> don't agree with the method doesn't make it wrong in any way.
And just because it worked for you doesn't mean it works in all cases.
>>>The OP already has sunk 450 quid into this. Another 20 for a
>>>controller card isn't going to break the bank.
>>
>>He'd still need to do a repair install.
>>
>
>
> No he/she wouldn't and I've got more than one system that says so.
Want to bet your life that I can't find a system it doesn't work on?
>
>
>>>MT
>>>
>>><snip>
>
>
> MT