Slashdot Mirror
Hardware Companies Team Up To Fight Mobile Linux Fragmentation
Nunavut writes with news that a number of hardware companies have banded together to battle the fragmentation of the mobile Linux market. ARM, Freescale Semiconductor, IBM, Samsung, ST-Ericsson, and Texas Instruments are forming Linaro, a nonprofit organization that plans to focus on "low-level software around the Linux kernel that touches the silicon, key pieces of middleware that enable new markets, and tools that help the developer write and debug code."
"Linaro's chief goal is to reduce the time that it takes to bring a new ARM-powered product to market with Linux. This effort is largely neutral with respect to what software environment and components individual vendors choose to run in user space. Linaro will not compete with existing platforms such as MeeGo and Android. Instead, it will attempt to improve the shared underlying software components that allow those platforms and others to run on ARM SoCs. In principle, this could actually reduce fragmentation at the lower levels of the Linux stack."
enable new markets
This probably will not go well.
they'll be yet another fragment
FTFA:
Linaro is closely aligned with Ubuntu's Linux on ARM initiative. Many elements of the Linaro project will be managed through Canonical's Launchpad collaboration platform, though a lot of the code will be managed and contributed upstream. Much like Ubuntu, the Linaro project will adhere to a six-month, time-based release cycle. This will allow for rapid iteration and predictability. Ubuntu founder Mark Shuttleworth discussed certain aspects of Linaro in a blog entry that was published Thursday. He views the project as a win for consistency and a constructive way to reduce fragmentation without eroding the diversity of the mobile Linux ecosystem.
I've participated in a few industry wide organizations like this. They can be somewhat effective, but even then, they move very, very slowly.
Jibe!
... or compete with Qualcomm? How many Android devices don't have Qualcom hardware?
What in god's name are you blathering about? This is a non profit organization that intends to deal with low level hardware libraries. It has nothing to do with "apps" and they have no intention of shipping a product. All of this is right in the summary.
Congrats!
You've won the coveted "You Retard!" slashdot community award.
Do you have any thing to say about this prestigious win?
Nokia & Intel.
But with added committees.
HTH
Deleted
Finally you will be able to defragment your Linux!
Like Extending Extended Extensions? Then Get the Extended Extension Extender Extended edition with Extra-Extended Extreme Extension Extending Extenders.
After having read the FAQ on the Linaro web site and a blog entry by Mark Shuttleworth, I get the feeling that this is an initiative coming from Canonical. If they will be driving this forward, with support from the hardware vendors of course, it might not become your garden variety standards organisation. In that case, the key issue will be to keep the commitment from the hardware manufacturers. But I guess it could work out alright considering Canonical isn't a direct competitor to any of them.
Just tell him that they're recommending duct tape to prevent mobile linux fragmentation - he obviously only reads the headlines while trying to get a first post :-)
And it eliminates the need to defrag (I know, linux doesn't need defragging :-) - just add a new layer of tape every month.
And if you want, Monica Lewinsky is now offering a line of Duct Tape Nobile Phone Defragmentation Skins - because you never know where someone's going to stick their phone ...
What is this about improving the ability to bring new Linux-based devices to market? There are scads of them already, and that's part of the problem.
The fragmentation these devices represent mainly occurs above and below the kernel level: Above in the sense that the libraries and UI kits that are included change greatly not only between devices, but also between iterations of a single product; Below in that there is no (realistically desirable) minimum hardware spec for a Linux-based device in any given class like smart phones.
So this Linaro effort doesn't seem to help out where help is most needed IMO: Enabling creative types who work above the hardware level to identify and easily make use of a robust and attractive platform... something to confidently write apps on. As with the PC revolution, again its the apps that will win user interest.
Linaro maybe more of an attempt to get the likes of Google to heel with its forking of the Linux kernel for use in Android. But I would say that Android itself is the better focus for a standardization effort: it is full-featured so the roles that any particular components will be expected to satisfy can be looked at much more holistically when features are being written or debugged. And without that holistic view -- like what happened to Desktop Linux as a vague concept -- components that are written or standardized outside of a specific user-targeted platform (like Android, or iPhone for that matter) are unlikely to give rise to a coherent and satisfying user experience.
IOW, design affects all layers simultaneously and what works for people administering servers does not work when people expect to create/sell/share software applications on a user-oriented platform.
Very good idea. Code reuse is always good. However, one minor point about the summary: fat chance of Android either helping or being helped by this - AFAIK, they've already messed up their Linux-derived kernel to the point where you can't assume that modules from actual Linux will work with it.
# cat
Damn, my RAM is full of llamas.
I once had the good fortune to work on a project where the standard proceeded so much faster than capability that for 6 months we were the world's only supplier of a standards-compliant product, though a small one. Believe me, it was worth the effort.
From scarped cliff or quarried stone she cries "A thousand types are gone, I care for nothing, no not one."
I would suspect that the only people unhappy with this particular initiative are those who compete with linux, for obvious reasons, and possibly those companies who currently specialize in doing bespoke embedded linux work for hire. MontaVista, for instance, has basically made a business of doing the (often obnoxious and tricky) work of turning "linux runs on ARCHITECTURE" into "Linux is now running on CUSTOMERS_HORRID_EMBEDDED_BOARD". An industry consortium dedicated to giving that away probably isn't ideal for them.
Presumably, any hardware manufacturers who treat their BSPs as a source of profit(or consider their BSPs to be superior to such a degree that they serve as a key differentiator from competing products) from would also be displeased; but the impression that I get is that most of them view hacking together the BSP as a necessary evil to be endured in selling the board(even when they do charge for them, it's more of a cost recovery/keeping out the riff-raff sort of thing), so most would probably prefer a strategy that makes it happen as cheaply as possible.
Why "Fight Fragmentation" rather than "Promote Unity"? We're all on the same team right?
The SUMMARY? LOL
Back in the day... we used to ridicule people who actually read the STORY but didn't also cross reference it from another media source!
Give the guy a break -- look at his USER ID. You can't possibly expect a Generation Z'er to read the SUMMARY, can you? Trust me, it's a lost cause.
Just accept it will only get worse.. I expect the next generation only posts words without capitalizing the first word of each sentence, all words without vowels and zero punctuation.
Now get off my lawn!
generation z is the last one dude the world ends after this so quit wasting time reading summaries and punctuating how much did you pay for 4551 anyway question mark
interface for their damn AIC33 audio chips, I'll be happy.
Just go to the configuration, and check the box at "enable new markets".
Unless your carrier has removed "enable new markets" from the menu, as AT&T appears to do.
That this will work about as well as the various "let's unite Unix" corporate drum circles from the 80s and 90s.
They did eventually get things to a point where a subset that wasn't entirely useless could usually compile pretty cleanly, most of the time, sort of, in most places. So long as the code wasn't concerned with whatever the market was concerned with five years prior.
Linus did more to unite Unix than any of the cross-company bodies.
To be more fair, moving the compatibility bar forward, even slowly, is not a bad thing. I just don't think it will do much to shape things in any meaningful way for quite some time.
I forget what 8 was for.
I will FORK you motherfucker! motherfucker!!
I've participated in a few industry wide organizations like this. They can be somewhat effective, but even then, they move very, very slowly.
I hope it turns out better than the old "Project Monterrey" thing...
Canonical started this a few months ago.
I work for Canonical (not in this project tough) and follow it with a personal interest.
Scientia est Potentia
What's wrong with http://en.wikipedia.org/wiki/Carrier_Grade_Linux
I'd like to buy homeland for our 10 million people. http://twitter.com/mahadiga
I wonder if they'll utilize OpenEmbedded since it already takes care of a lot of this type of thing.
I gotta hand it to the MeeGo folks. Their project has goals like
1) Keep it FOSS. All of it (in the core distro)
2) Upstream code whenever possible
Even if you don't use it as a mobile OS, the work being done on it by Intel, Nokia, etc... is going to benefit pretty much every Linux-derived distro out there.
If Linaro wants to join the party and throw time/money at improving Linux-y software running on ARM chips, that sounds pretty darn good to me!
coding is life
I've participated in a few industry wide organizations like this. They can be somewhat effective, but even then, they move very, very slowly. - by Foredecker (161844) * on Saturday June 05, @10:33AM (#32468588) Homepage
On the note of "moving slowly", you're not one to talk. See here:
http://slashdot.org/comments.pl?sid=1630116&cid=31975424
How many months ago was it you'd stated you'd find answers to the HOSTS file dilemna noted there? 7 months now. You stated you'd get back to that poster in far less (2 months or so).
Pertinent material, requoted from the url above, now once again below:
***
"Be patient :) Ill get to this. I just dont know when. I think I can get back to you by mid February, but it may be March." - by Foredecker (161844) * on Saturday April 24, @01:42PM (#31968126) Homepage
That quote of your words is from here -> http://slashdot.org/comments.pl?sid=1495166&cid=30715150 back in January (10th of Jan 2010)...
Once more, to refresh you on it:
This is again, in regards to HOSTS files in VISTA, Windows Server 2008, & Windows 7 being unable to use the smaller & faster + more efficient "0" blocking "IP Address" (vs. the larger, slower, & less efficient on filesize & read/write time 0.0.0.0 (or, worse yet, 127.0.0.1 "loopback adapter IP address") which are STILL useable in Windows VISTA, Windows Server 2008, & Windows 7!).
However/Again (refreshing your mind on the particulars/details), before MS "Patch Tuesday" on 12/09/2008 though? Well - You could STILL USE THE SMALLER & FASTER 0 blocking address in HOSTS files, vs. the larger & slower + less efficient 0.0.0.0 or worse still, the 127.0.0.1 loopback adapter address in Windows VISTA, Windows Server 2008, & Windows 7 (for blocking out KNOWN BAD sites &/or servers)...
Using 0 yields increases in speed + efficiency & due to FAR LESS FILESIZE involved for reads inside the file and reading the HOSTS file as a whole (smaller = faster), especially!
----
E.G.->
HOSTS using 0, with 840,000 blocked KNOWN BAD sites &/or servers entries in it blocked = 18,430 kb size
vs.
HOSTS using 0.0.0.0, with 840,000 blocked KNOWN BAD sites &/or servers entries in it blocked = 23,338 kb size
vs.
HOSTS using 127.0.0.1, with 840,000 blocked KNOWN BAD sites &/or servers entries in it blocked = 24,975 kb size
----
As you can see, 25%-35% approximate filesize diff.'s in using smaller vs. larger preceeding blocking addresses in front of bad sites/servers domain-hosts names manifest themselves ("do the math" etc.), & thus? Using 0 as a blocking address indeed DOES MAKE A DIFFERENCE here, for performance sake!
(Which is YOUR division @ MS you head, correct? In the "Windows Client Performance Division" so, this ought to interest you some, & hopefully enough to find out WHY the IP Stack Team has taken out the fastest & smallest + most efficient entry of 0 for blocking in HOSTS files... makes NO sense that they did, because of the evidences above!)
Funniest part is, the Windows 2000, Windows Server 2003, & Windows XP still can use the smaller, faster, & most efficient 0 blocking address (vs. the larger/slower 0.0.0.0 & worst of all, 127.0.0.1)... but, MS inserted the ability to use 0 as a blocking IP address back as far as Windows 2000 (not its original OEM pre-service pack/hotfix release, but, somewhere in between SP#1 - SP#4 for Windows 2000... this is a BETTER STANDARD, one that MS set no less, because it yields a smaller & faster read HOSTS file, period!)
The physics of it all back me on this, & so does the math.
Especially when populating either the DNS ClientSide Cache service, OR, the loca