SFLC: Microsoft confirms UEFI fears, locks down ARM devices

Posted Jan 16, 2012 12:36 UTC (Mon) by drag (guest, #31333)
In reply to: SFLC: Microsoft confirms UEFI fears, locks down ARM devices by fluxion
Parent article: SFLC: Microsoft confirms UEFI fears, locks down ARM devices

Non-MS netbooks were inferior. Almost all of them had shit or otherwise hacked up Linux systems that could not install any software or do much of anything. The only people that put out decent systems were Dell and Acer.

When netbooks were first released the Linux ones were the best sellers. That boat quickly sailed. Linux had it's chance and it was roundly rejected by the people that purchased them. There are a few here and there that liked them, but there was virtually no repeat customers.

Microsoft didn't 'manage to shift' anything. Blaming Microsoft for Linux Desktop's failure is just means that you completely ignore any possible lesson that can be learned here.

A very careful examination why Android has succeeded while the traditional "Linux distribution" approach to making a smart phone OS failed utterly is in order, also.

Oh yeah, the secret so deep, the secret so hidden, the secret so unbelievable… that everyone knows about it.

Posted Jan 16, 2012 17:59 UTC (Mon) by khim (subscriber, #9252) [Link] (4 responses)

A very careful examination why Android has succeeded while the traditional "Linux distribution" approach to making a smart phone OS failed utterly is in order, also.

Very careful examination? Didn't you mean "5 second glance"? You do remember this video, right? Where crazy Microsoft CEO dances on scene?

Well, he's right: developers make or break your platform. And developers want stability. Not UI stability, but ABI stability. They want to create binaries once and sell them for a long, long, time. What they absolutely don't want to do is to keep few versions of them, recompile them for all existing distributions, etc.

Linux (the kernel) actually is pretty good here: while it has "no stable API" policy this policy only cover kernel. User-space ABI is sacred and Linux developers take regressions very seriously. Thus, surprise, surprise, Linux (kernel) is used on billions of computers around the world - but only on tiny percentage of desktops.

Why? Well, desktop people (and ever in-kernel desktop-related people) are breaking everything regularly. It's not a new phenomenon (I've already discussed it few years ago) but it's still valid. When you read maemo 4.0.x is not API compatible with earlier releases you know that someone decided to shoot his foot again.

And if you shoot his foot again and again and again… then limping is kind of expected, right?

Note: stable ABI is strict requirement, but of course it's not enough to drive your platform to success. You need to do other things, too. But if you don't offer stable ABI then it does not matter what else you'll do.

Oh yeah, the secret so deep, the secret so hidden, the secret so unbelievable… that everyone knows about it.

Posted Jan 20, 2012 1:39 UTC (Fri) by sorpigal (guest, #36106) [Link] (2 responses)

+1 million, the truth.

IMO API stability would be enough to begin with, but nobody does that either.

I like to say it like this: Microsoft succeeds by thinking about the platform first. In Linux there is no platform above libc (except maybe xlib).

Oh yeah, the secret so deep, the secret so hidden, the secret so unbelievable… that everyone knows about it.

Posted Feb 10, 2012 11:47 UTC (Fri) by mfedyk (guest, #55303) [Link] (1 responses)

You forgot about eglibc, dietlibc, etc. And for xlib there is xcb and xcb-xlib as well.

On another note, when are we going to get x protocol extensions that integrate the nx protocol and a module that does compositing into xorg? These would solve nearly all of the issues that make people want to work on wayland.

Long live X!

Oh yeah, the secret so deep, the secret so hidden, the secret so unbelievable… that everyone knows about it.

Posted Feb 10, 2012 13:21 UTC (Fri) by khim (subscriber, #9252) [Link]

You forgot about eglibc, dietlibc, etc. And for xlib there is xcb and xcb-xlib as well.

Which does not change anything. EGLIBC strives to be source and binary compatible with GLIBC and dietlibc is rarely used as glibc replacement. Microsoft, too, offered numerous version of MSVCRT: Windows 7 includes three or four "out of the box".

The important thing is not to decide when to add something but when to remove something. And the typical answer: years after the replacement is available. FCB was introduced in MS DOS 1.0 (in 1981) and deprecated in MS DOS 2.0 (in 1983). It was supported till "grand unification" of Windows (in 2001). And still some people complained because Windows XP broke their beloved WordStar.

Now, some interfaces are abandoned much faster (think DirectMusic) and then Microsoft is [rightfully] hated - but these are rare exceptions, not rules. In Linux world... yes, we have kernel, yes, we have glibc and xlib... and that's about it. Well, GTK+ comes close. Everything above is subject to sudden breakage. And while compatibility is possible (as someone pointed out you just need to pull bunch of old libraries from older versions of the distribution) it's not automatic: user must manually find and install these libraries, etc.

At some point it just becomes too tiresome and people switch to Windows or MacOS. Where things work "out of the box" and you desktop looks like a desktop not as Tamagotchi.

Oh yeah, the secret so deep, the secret so hidden, the secret so unbelievable… that everyone knows about it.

Posted Jan 20, 2012 4:33 UTC (Fri) by elanthis (guest, #6227) [Link]

> Well, he's right: developers make or break your platform. And developers want stability. Not UI stability, but ABI stability. They want to create binaries once and sell them for a long, long, time. What they absolutely don't want to do is to keep few versions of them, recompile them for all existing distributions, etc.

So very, very true. I'd argue this extends beyond the library ABIs as well, of course, to things like the software installers. It's still very frustrating that a developer has to build like 50 packages of the same application for major distros on the Linux desktop.

FOSS is nice and all, but regular people want to just be able to click and go without needing to dick around with building source, and regular developers don't want to have to spend their time working around a broken distribution platform and breaking ABIs.

SFLC: Microsoft confirms UEFI fears, locks down ARM devices

Posted Jan 16, 2012 20:32 UTC (Mon) by fluxion (subscriber, #62978) [Link]

I think you're focusing to much on my usage of the word "shift". My concerns are entirely grounded in our mutual agreement that neither linux or Android is in a position to take the desktop market from Microsoft. If it was, we could rely on having a fair share of non-secureboot ARM workstations at our disposal.

My point is that this time around, we won't have a chance to even try before Microsoft completely locks desktop linux out of ARM with secureboot, and Android finds itself the only contender to push back into that space, a space that I worry it will never be targeted toward. This leaves the future of linux being overwrites on toy gadgets designed to running Android, which would be a sad state of affairs.

Netbooks and price points

Posted Jan 17, 2012 1:08 UTC (Tue) by man_ls (guest, #15091) [Link]

I agree with most of your post, except for one detail.

Microsoft didn't 'manage to shift' anything.

In fact they did: they lowered the price of Windows XP to a point where it was no longer a big % of the machines' prices. It is true that offered a choice of Linux and (similarly priced) Windows people went with the latter. It is no less true that when the Windows machine was $100 more expensive, people went with Linux. This is not blaming Microsoft or anything; they did the smart thing and lowered their prices to match the demand.