From: Rod Speed on
Nick Andrew wrote:
> Frank Slootweg <this(a)ddress.is.invalid> writes:
>
>> Gary R. Schmidt <grschmidt(a)acm.org> wrote:
>>> And "released decent drivers" can be replaced with "released
>>> specifications" or "released information not under a Non-Disclosure
>>> Agreement that prevents an open-source driver being written" and
>>> variations thereon.
>
>> Even if the latter two requirements are met, you can and often still
>> will have a problem. Case in point: After a very long goose chase,
>> including 'help' from Linux zealots which said my problem didn't
>> exist, I *finally* found that my about-to-be-bought USB DVB-T tuner
>> was "supported", in the sense that that make and exact model number
>> had been tested. Just very shortly before buying it, I found out
>> that the current version of that tuner had a slightly different
>> chip-set, which made the device *not* work.
>
> Vendors often do this, apparently. Specifically, they reuse an
> existing USB vendor and model ID for a new device with different and
> incompatible hardware inside it. Vendors can maybe get away with this
> because the Windows end-user is expected to install the device driver
> which comes with the device, whereas OSS developers have to write a
> device driver which works for _all_ the variants of the hardware. In
> real life this may mean that the actual "device driver" code pulls in
> modules for the actual hardware support based on what is discovered
> about the hardware at module load or hardware insertion time.
>
> So at the root cause it's basically another instance of "lazy vendor
> and/or under-documented hardware". I know that doesn't help much when
> you the consumer want to buy a device. All I can suggest is buy an
> expensive device from a good quality vendor like Hauppauge rather than
> an el cheapo Taiwanese no name clone.

Or give up on Linux for more exotic hardware like DVB-T cards,
because it drastically reduces your choice of hardware.

You can be completely confident that the card manufacturer has made
it work with Win and if it doesnt, you are legally entitled to a full refund.


From: Frank Slootweg on
Rod Speed <rod.speed.aaa(a)gmail.com> wrote:
> Nick Andrew wrote:
> > Frank Slootweg <this(a)ddress.is.invalid> writes:
[...]
> Or give up on Linux for more exotic hardware like DVB-T cards,
> because it drastically reduces your choice of hardware.
>
> You can be completely confident that the card manufacturer has made
> it work with Win and if it doesnt, you are legally entitled to a full refund.

Be careful Rod! This looks very much like you agreeing with the
pseudo-kraut! I'm sure you don't want *that* to happen. :-)

But seriously:

Not to mention "exotic hardware" like mobile (HSDPA/GPRS) modems.
Mobile networking for a mobile computer, who would have thought of
*that*! :-(
From: Rod Speed on
Frank Slootweg wrote
> Rod Speed <rod.speed.aaa(a)gmail.com> wrote
>> Nick Andrew wrote
>>> Frank Slootweg <this(a)ddress.is.invalid> writes

>> Or give up on Linux for more exotic hardware like DVB-T cards,
>> because it drastically reduces your choice of hardware.

>> You can be completely confident that the card manufacturer has made
>> it work with Win and if it doesnt, you are legally entitled to a full refund.

> Be careful Rod! This looks very much like you agreeing with the pseudo-kraut!

Nope, you're agreeing with me, silly.

> I'm sure you don't want *that* to happen. :-)

Indeed.

> But seriously:

> Not to mention "exotic hardware" like mobile (HSDPA/GPRS) modems.
> Mobile networking for a mobile computer, who would have thought of *that*! :-(

Yeah, its only relatively recently that its even handled USB and wifi anything like as effortlessly as Win does.


From: Nick Andrew on
Frank Slootweg <this(a)ddress.is.invalid> writes:

>Nick Andrew <nick(a)spamtrap.nick-andrew.net> wrote:
>> Vendors often do this, apparently. Specifically, they reuse an existing
>> USB vendor and model ID for a new device with different and incompatible
>> hardware inside it. Vendors can maybe get away with this because the
>> Windows end-user is expected to install the device driver which comes
>> with the device, whereas OSS developers have to write a device driver
>> which works for _all_ the variants of the hardware. In real life this
>> may mean that the actual "device driver" code pulls in modules for the
>> actual hardware support based on what is discovered about the hardware
>> at module load or hardware insertion time.
>>
>> So at the root cause it's basically another instance of "lazy vendor
>> and/or under-documented hardware". I know that doesn't help much when
>> you the consumer want to buy a device. All I can suggest is buy an
>> expensive device from a good quality vendor like Hauppauge rather than
>> an el cheapo Taiwanese no name clone.

> My case wasn't a "no name clone", but Pinnacle, which is a major
>vendor of this kind of stuff. I would not call them "expensive", but
>buying an "expensive" device is rather silly, when there are perfectly
>good and working "cheap"er devices.

I've heard of Pinnacle, but can't comment on the quality of their
devices. If they're reusing IDs with incompatible hardware though,
it is a sign of cutting corners.

> Anyway, my main point is that add-on stuff - other than disks (and
>disk look-alikes) - is off-limits for a Linux desktop/notebook/netbook,
>unless Linux is *specifically* supported by the *device vendor*.

I don't think so. Very few device vendors specifically support Linux
but there are a large number of devices which just work. Vendors who
specifically support Linux that I can think of include HP, NVidia,
Intel - and this is for a limited set of device types: scanners,
video cards. Other vendors may work with device driver writers behind
the scenes but they don't put a penguin logo on their packaging.
Yet others (like Canon, in my experience) specifically refuse to have
anything to do with Linux. Yet linux works on hundreds of scanners,
even some Canons (but not the one I was thinking of buying). So if
you want to buy a scanner it pays to choose a model which is known
to work. Fortunately the SANE project supports hundreds of models
of scanners, so that is not too hard. Fortunately the CUPS project
supports hundreds of models of printers. But in WiFi and DVB,
new devices are being developed at a great rate and sometimes it
takes time for the open source drivers to support the hardware.
I have a m920x based DVB device (two, actually) and at the time
I bought them the device wasn't usable under linux. Within a
couple of months, it worked fine. Sometime later I obtained a
new AF9015 based DVB device and it didn't work under linux - the
AF9015 driver was still at a "pre-alpha" state. Considerable
work has occurred on the AF9015 driver since then and although I
haven't tested the device on a recent driver, I believe it will
work.

Nick.
From: Nick Andrew on
"Rod Speed" <rod.speed.aaa(a)gmail.com> writes:

>Or give up on Linux for more exotic hardware like DVB-T cards,
>because it drastically reduces your choice of hardware.

Exotic is a good word there. But having a working DVB-T is seriously the
least important part of my computing experience, so I wouldn't give up the
other benefits of Linux just so I can have a working TV.

>You can be completely confident that the card manufacturer has made
>it work with Win and if it doesnt, you are legally entitled to a full refund.

I'd put "work" inside double quotes there. I've heard of device drivers being
so minimal that they only work with a single device of some type. Bad luck if
you want to use two of them.

Nick.