__/ [ Oliver Wong ] on Friday 15 September 2006 18:26 \__
> "Richard Rasker" <spamtrap@xxxxxxxxxx> wrote in message
> news:pan.2006.09.15.08.52.33.316222@xxxxxxxxxxxxx
>>
>> No matter how well Linux can extend the life span of older hardware, even
>> Linux-users need some new hardware from time to time. And ever more often,
>> the hardware comes with severe warnings: "WARNING! Install Driver Software
>> from CD FIRST, Then [device]!", often in big type on huge red labels
>> sealing the bag with the goodies. I've seen it with USB-printers and lots
>> of other USB devices, but also PCI-stuff.
>>
>> Alas, when I try to heed these warnings, it invariably turns out I can't
>> install the driver software from the CD, because there isn't any. For
>> Linux, that is. Luckily, when I hook the stuff up, it mostly works in one
>> go, so I never got round to researching exactly *why* these warnings were
>> given - until I got a call from a family member.
>> The man was trying to install a second hand PCI WiFi card (delivered with
>> driver CD) on a Windows box. He'd plugged in the card, and booted the
>> machine to install the drivers. Windows worked through its routine of "New
>> Hardware Found", he fed it the driver CD and he thought everything was
>> going fine. Not so. The card refused to work. Even worse, it wasn't even
>> recognized as such.
[commenting as I read, first pass]
"The Device is working properly" pitch, right? And the option for "reinstall
driver" -> "Have/from Disk". That never works. When Windows works, it might
work. When it fails, it falls down -- "SPLAT" -- face on the floor. And
there are many factor that can assist a failure, e.g. a Service Pack, a
patch, or even malware. And rarely will there be an indicative sign, other
than chronological reconstruction, to say what the culprit was, let alone
how to resolve the issue. That, among other reasons, is why so many people
reinstall Windows way too frequently. It's time-draining and it leads to
loss of data such as application settings.
>> He tried installing the drivers again, rebooted again - and that several
>> times. No go. The manual he downloaded from the Internet gave no real
>> clues too. Then he contacted the previous owner; he'd seen the card work
>> on a machine there, so it coudn't be broken, but perhaps the guy knew what
>> might be wrong. The man told him that, yeah, now he thought about it,
>> there'd been a warning on the package, something about installing the
>> drivers *before* plugging in the card.
>> OK, so the cause of the problem was more or less clear, but not the
>> solution. Searching on the Internet suggested anything from uninstalling
>> the drivers (impossible, since the hardware wasn't recognized properly, so
>> he had no access to its properties), hacking the registry (a sure recipe
>> for disaster), removing mainboard device drivers (yeah, way to go!), doing
>> a rollback (and lose many months of installed apps, not to mention work),
>> right up to the King of all Windows Solutions: a fresh reinstall of XP.
I guess this proves my previous point (hadn't read this before). And the "use
Google to solve Linux problems" myth is suddenly shattered, or at least
leveraged to align with Windows.
>> Alas, I couldn't help the man either, so now he's been busy for many hours
>> preparing for a roll-back, making back-ups, retrieving the software he
>> installed the past months. Yup, this is the famous Windows ease-of-use as
>> I know it. But the man's looking on the bright side: the machine had
>> slowed down considerably in the past months, and hadn't he seen some signs
>> of malware? So it was due a reinstall or a rollback anyway. Poor sod ...
The more complex the work environment, the longer it takes to 'reconstruct
'it (always with some loss assumed). Moreover, reconstruction of settings in
Windows is hard because not everything is a simple file in the home
directory, from which unwanted bits can be culled out. A complete image of
Windows before reinstallation is a loss cause as it contains the
'contamination' in system's state. That in itself is an issue that many
people face when a virus strikes. Their Norton Ghost image may already
contain a live (let us say "Frozen") virus, in which case the image is
'dirtified'.
>> Anyway, my technical curiosity wasn't satisfied yet - but after some
>> digging, I think I found out what's going on. And yes, as I expected, it's
>> Windows' "user-friendly" way of making things "easy", turning out to be
>> something between a nuisance and a disaster.
>>
>> What happened? As the man plugged in the card, Windows found it,
>> recognized it, and tried to locate and install drivers for it. All good
>> and well, and very user-friendly - if only Windows had recognized it
>> correctly. In fact, it didn't, and Windows set up the wrong drivers for
>> the hardware. But as the hardware turned a totally deaf ear to these
>> drivers, it didn't show up anywhere any more, so the user couldn't change
>> anything about it any more. The result: a one-way dead end street.
Imagine yourself installing SUSE, choosing GNOME as the desktop environment
and then finding yourself unable to use KDE. I suppose it's the way the
stack of drivers and components is constructed (one large blob rather than
hierarchical set of modules) which leads to such a trap. Software which is
designed by glueing bits, each of which is developed by a different team
without protocols/interfaces for communication and exchange, is bound to
become overly complex. And that's the issue Windows is facing.
>> Of course this wouldn't have happened if the user had observed the right
>> installation order. But he couldn't have known this - and besides, I think
>> it's very logical to plug in a device, then install drivers (if at all
>> needed). Often, it's even advisable *not* to install the manufacturer's
>> drivers at all - I once permanently knocked out an XP laptop within 15
>> minutes, by simply installing some drivers and tools supplied with a USB
>> stick.
>> So once again, crappy design makes users jump through all kinds of hoops,
>> with punishment for failing to do so ranging from mild annoyance to many
>> lost hours. And this is what all those trolls here call "excellent Windows
>> hardware support"? You can keep it ...
That statement assumes that you still have all these driver CD's somewhere at
the back of the drawer. Moreover, it assumes you don't 'tango' with several
different versions of Windows (yes, sometimes the drivers themselves vary,
i.e. are forked). Linux works out of the box and when it does, it's
painless. Windows is a different case altogether, despite the fact that
hardware manufacturer tailor hardware and software for that one platform and
work 'hand in glove' with Microsoft.
> Your post is a bit on the antagonistic side, but otherwise this is a
> good point and something Windows could be improved. However, I'm not
> certain that it's a design flaw on Microsoft's part. It could be poor code
> on the device manufacturers part.
Aye. But the customer is caught in a deadlock-type situation, which is partly
permitted due to bad O/S architecture and the ability to retract component
or reverse an installation.
> To your friend, if he hasn't reinstalled Windows yet, I recommend he
> try
> (physically) removing the device, and seeing if the drivers will install at
> that point, and then re-installing the device (physically).
A time machine may be a worthwhile route.
> I'm surprised and a bit skeptical that Windows installed the wrong
> drivers though. As I understand it, there's a standard (not sure if it's
> from Microsoft, or defacto, or from somewhere else) signature that all
> hardware is able to emit to the computer. It's something like an 10 bytes
> to identify the manufacturer, and another 10 bytes to identify devices made
> by that manufacturer. In my experience, when Windows can't find a match, a
> new entry is added to the device manager under "unknown device". If an
> incorrect driver WAS installed, the (phantom) device should show up in the
> device manager, with a yellow "warning" sign indicating that while the
> driver is functioning, the hardware is not. At that point, you can right
> click on the phantom device, and choose "Uninstall driver".
Interesting. I didn't know some of these things. And I suspect it was more
than 10 bytes when I worked with an ARM processor, communication from a
GUI-based front-end.
Best wishes,
Roy
--
Roy S. Schestowitz | Partition if an operating $ysteM must be set aside
http://Schestowitz.com | GNU/Linux | PGP-Key: 0x74572E8E
Swap: 1036184k total, 310188k used, 725996k free, 53980k cached
http://iuron.com - next generation of search paradigms
|
|
