Linked by Thom Holwerda on Sat 29th Nov 2008 21:22 UTC
Linux Even though there are a lot of happy people using Apple's iPhone very happily, there's also a group of people who are not so happy, most likely because of Apple's rather strict policies regarding applications and developers. While most of these people would just jailbreak the thing, some take it a step further - by installing another operating system. Yes, Linux now runs on the iPhone (1st gen/2nd gen, and the 1st gen iPod Touch).
Thread beginning with comment 338777
To view parent comment, click here.
To read all comments associated with this story, please click here.
RE[3]: Wha?
by silix on Sun 30th Nov 2008 23:06 UTC in reply to "RE[2]: Wha?"
silix
Member since:
2006-03-01

The reason you don't have a stable ABI in Linux is that the kernel folks do not want binary-biased driver integration.

exactly... the sad thing is, nobody realizes that making kernel APIs stable (for at least a reasonable assured time-/version- span duration, say... 3 years? ) and encouraging HW manufacture to open their drivers or at least their specs, are different and in fact orthogonal problems...

Which means that driver support will always be second-rate in Linux.

... and that it's not that easy, or reasonable, or "normal" for a HW maker, to even want to care for the users of an OS, that make less than 1% of the market that company caters to (not that they are really obliged... if i state a graphics card of mine specifically is a "DX10 gaming card for windows", if someone manages to use it on maya under linux good for him - but support for unintended use is not due)

nor, in case they care, to entrust some random developer (totally uninvolved with the design of a HW product; and, for what they may assume, uninvolved with the development of similar devices, or drivers for them, or driver development in general) with the IP related to their product, just to let him (try to) write a driver for that device...

the most suitable ones to write device drivers are those working for the manufacturer (and for as overly exceptionally complex a device can be, as in the case of a modern vga, one will just not suffice)
this is a simple but often overlooked fact that does not deny the obvious, is that peer review of code will be important for system level integration, auditing and debugging (but only at that point)

Imagine how much better things would be if you had a common ABI across distributions. You could use drivers from ANY distribution for your processor indiscriminately.

or, you'd have a chance to use a device (e.g. a webcam) as soon as a driver for it (for as broken as it may be) is made available, without the need to draw in a newer kernel
being able to resolve problems caused by the new driver just uninstalling /updating it, instead of reverting the whole system to the previously used kernel, and to avoid the possible regressions or new vulnerabilities inherent to a less mature kernel, may be invaluable...

But, no, the ABIs are constantly in flux, creating an environment which is brittle and unstable.
[...]
C'mon, folks, let's be honest here: this is RIDICULOUS.

it's ironical, even more so when they attempt to pass a political decison for a techinical one, pointing at GCC's incompatiblity with itself across major versions as a major cause of ABI instability in the kernel, as if it were something impossible to obtain... (of course ABI incompatibility between GCC ports for different architectures is understandable and somewhat obvious - though i've often seen it mentioned to justify the above (!)) ...

Somebody should let the grown-ups take over. Oh, and regarding LSB, it's a joke.

i used to have faith in the LSB but then i stumbled upon this: http://www.linuxfoundation.org/en/LSB:PM:Best_Effort_Dynamic_Linkin...

<< Right now, the LSB requires a different dynamic linker than the rest of the system. This linker is often not provided at all on non-LSB systems, and cannot be guaranteed to be available even on distros that can be LSB-compliant (if the LSB environment is not installed).>>

when a spec which is expected to be a common denominator and something all (compliant) linux distributions would be based from an "architectural" perspective, but (by words of its very proponents) turns out to be an optional side by side "environment", loses a lot of credibility in my book...

Reply Parent Score: 1