Re: free software mini pc
On Wednesday 15 February 2012 2:01:22 pm green wrote:
> Mark Neidorff wrote at 2012-02-14 17:45 -0600:
>
> When you purchased the server on which you run Lenny, did you know for sure
> that the installation would go smoothly and all hardware would work
> correctly? What if today you needed another system on which to run Debian
> and knew that you did not have time to troubleshoot any hardware problems?
> You could get the same as what you have now, but what if it is no longer
> available? Wouldn't it be helpful to find a vendor that provided a
> hardware table for each system with information about Linux mainline
> kernel versions, drivers, and firmware? Like, "this SATA controller is
> supported since Linux v2.6.29 with the ahci driver". So in that case you
> could look at their site, compare with the kernel version in Debian
> stable, and know with reasonable certainty that this hardware will "just
> work" with Debian stable. Or that you need to consider a kernel in
> backports, etc.
>
Yes. I knew because, for a server, I bought slightly "behind the curve." For
the server, I knew that I didn't need the latest and gretest, so I was able to
look at hardware that had been on the market for about a year and check
compatibility easily. Then the install "just worked."
IMO, in getting "the latest and greatest" can be as much of an ego thing as a
productivity thing. Question is: what are your specific needs going to be?
That will determine the power and features that you need.
> Many vendors mention various versions of Windows on their hardware pages,
> but nothing about Linux. So as a consumer, do I just blindly assume that,
> although the vendor apparently does not care enough about Linux to even
> mention it, that it will all "just work"? Or those that mention Linux,
> but no kernel versions: will the kernel in Debian stable work? Or those
> with Linux drivers available for download, do I need to maintain
> out-of-tree drivers (remember I mentioned a maintenance burden)?
Here's another way of looking at the same thing. Other M$ require that
hardware goes through a certification process before it gets the "works
with..." sticker. They have a roll-out scheudle of once every few years. Is
that what you want? That costs the consumer $$$. Are you willing to spend
for that?
>
> Now, because of the implication that hardware (as with your server, Mark)
> will all "just work" with Debian (and that my post/research is just
> silly/trolling), I will quickly mention nvidia, fglrx, and ralink wireless,
> all problematic a while back. I have had a Thinkpad T61 with a PSTN modem
> for >4 years, it has never worked (Debian amd64); I hope to try again when
> I upgrade to wheezy. Okay, so now someone might say "well, of course
> video, winmodems, and wireless will cause some trouble sometimes". These
> mini-pcs... any of them have onboard video hardware? Or come with
> wireless hardware?
>
> And someone might say that many of the problems had in the past are
> resolved, and quite possible so. So if I need a functional device now, do
> I need to just purchase one and shelve it for a few years before assuming
> Linux will work? I understand that Linux has a history of better support
> for older hardware, and that is reasonable, but would that need to be so
> (as much) if vendor support was better? And the Intel GM965 video on my
> T61 still does not quite work correctly for 3d applications, even after 4
> years.
True, audio and video devices have been less than perfectly supported in
linux. Look at why. Video hardware goes through benchmark testing. The
"ed's choice" hardware does the best on the benchmarks and sells the best.
So, the hardware is built to work best ON THE BENCHMARKS, but not necessarily
in the real world. So what linux faces is hardware that is tweaked to do well
on benchmarks on a different OS. This has lead to hardware manufacturers not
releasing their code to linux, bucause they would reveal how they make the
hardware look good on the benchmarks. Audio is continuously being worked on.
It is another difficult area for similar reasons.
>
> Okay, I could look through the specifications carefully and research eg.
> the wireless hardware, but what about when vendors change the chipset
> mid-model?
Yep. that is always a problem with buying the "latest and greatest."
>
> Am I being demanding here? I want an absolutely functional Linux on a
> device, and I am willing to pay for it (I have mentioned no limit, though I
> do have a budget). For those assuming I am needing tens or hundreds of
> whatever mini-pc I choose, no. I only need a single mini-pc system. More
> later, perhaps. It is not for my own use, but at a location where tech
> support is not available, and where the system will quite likely be in use
> for 5+ years.
One question. Do you expect the device to continue to be 100% functional when
the infostructure around it will change over the next 5+ years? That is not
reasonable.
>
> So to recap my original post, the basic requirements are:
> - fanless mini PC
> - it will run Debian
> - production environment (reliability is important)
> - good Linux support to facilitate fast deployment and low maintenance,
> - avoiding non-free software (non-free firmware, out-of-tree kernel
> modules, ndiswrapper)
>
> and I mentioned also:
> - many devices with only partial mainline Linux support
> - unable to find itemized information about Linux kernel support
> - some devices ship with Linux (often Ubuntu) and use a custom kernel
>
> My original post did not mention this explicitly, but I would be pleased to
> find a manufacturer/vendor that is interested in supporting Linux users,
> and provides devices with 100% functionality using 100% free software.
> Perhaps that sounds a bit less demanding, while still being very closely
> related to the original.
>
> The response I expected to that original post, and would even have expected
> to the question in this previous paragraph, is that no, unfortunately there
> are no/few significant vendors that are interested in Linux users to this
> extent. I would be satisfied with this answer, though disappointed, and
> definitely interested in future developments of this sort. (I do not see
> any use in beating an old topic.)
>
> I was surprised with the responses I received instead; I hope this and my
> other succeeding posts help remove the obscurity.
Reply to: