Linked by Thom Holwerda on Fri 15th Jan 2010 23:06 UTC
Ubuntu, Kubuntu, Xubuntu "The Ubuntu development community announced today the availability of Ubuntu 10.04 alpha 2, a new prerelease of the next major version of the Ubuntu Linux distribution. This alpha is the first Ubuntu release to completely omit HAL, a Linux hardware abstraction layer that is being deprecated in favor of DeviceKit."
Thread beginning with comment 404386
To view parent comment, click here.
To read all comments associated with this story, please click here.
RE: Deprecation
by Ed W. Cogburn on Sat 16th Jan 2010 16:45 UTC in reply to "Deprecation"
Ed W. Cogburn
Member since:
2009-07-24

we no sooner get one subsystem that works reasonably well when the wheel is re-invented yet again.


That's called learning from your mistakes, and trying to do things 'better'.

HAL had problems, but they weren't *really* noticed until people started to use it for lots of things it wasn't originally designed to handle, aka. 'feature creep' (and using XML for the config files didn't help any).

The people behind the *kits and u* packages are the same ones that were behind HAL, they aren't inventing a new wheel (where 'wheel' here refers to the general idea/function of the software), just refining the old one (different API because the old one was, in hindsight, broken, and making it more modular, breaking the old monolithic package into smaller more flexible pieces). HAL was simply trying to do too much.

Is it any wonder that most commercial software developers don't target it?


Most software apps wouldn't need to interact directly with HAL. They'd use DE hooks, or an xplatform lib, rather than talk to HAL directly. I don't think thats really significant.

They don't target Linux because it doesn't have much of any market share. Over the years, Windows has had various warts and ugliness that coders targeting it had to deal with, but that didn't stop them, they went to all that trouble anyway because of Windows's market share.

It always boils down to just the size of the market...

Reply Parent Score: 3

RE[2]: Deprecation
by strcpy on Sat 16th Jan 2010 18:55 in reply to "RE: Deprecation"
strcpy Member since:
2009-05-20


That's called learning from your mistakes, and trying to do things 'better'.


Alternatively, it can be also called NIH, lack of planning, and cowboy coding.

What was it? XXX: Doing the same thing over and over again and expecting different results.

Reply Parent Score: 6

RE[3]: Deprecation
by Ed W. Cogburn on Sat 16th Jan 2010 21:08 in reply to "RE[2]: Deprecation"
Ed W. Cogburn Member since:
2009-07-24

Alternatively, it can be also called ...


It can be called a lot of things. Opinions are plentiful, because everybody has one...

Reply Parent Score: 1

RE[3]: Deprecation
by Bill Shooter of Bul on Sun 17th Jan 2010 00:16 in reply to "RE[2]: Deprecation"
Bill Shooter of Bul Member since:
2006-07-14

No, it can't. Something can be NIH, but if it is, then its not the same thing as learning from your mistakes. You can't learn from mistakes you don't understand.

NIH is all about not learning the existing solution and implementing your own because you understand your own code more than anyone else's.


Learning from mistakes is all about understanding the benefits and limitations of existing software.

Reply Parent Score: 3

RE[2]: Deprecation
by darknexus on Sat 16th Jan 2010 20:25 in reply to "RE: Deprecation"
darknexus Member since:
2008-07-15

It's nice in theory, but realistically the hooks are often broken from version to version, one subsystem to another. Not that I can really expect much, if the kernel team won't bother to maintain stable APIs why should anyone else? You can replace subsystems to your heart's content as long as the APIs don't change. No one, however, seems to care about maintaining a stable API for desktop Linux. They just don't see it as important for some reason.

Reply Parent Score: 3

RE[3]: Deprecation
by Ed W. Cogburn on Sat 16th Jan 2010 21:04 in reply to "RE[2]: Deprecation"
Ed W. Cogburn Member since:
2009-07-24

if the kernel team won't bother to maintain stable APIs


I'm sure your aware some folks have a different take on that:

http://www.kroah.com/log/linux/stable_api_nonsense.html

Besides, what does this have to do with HAL? It is/was a userspace app...

Reply Parent Score: 2

RE[3]: Deprecation
by abraxas on Tue 19th Jan 2010 00:17 in reply to "RE[2]: Deprecation"
abraxas Member since:
2005-07-07

It's nice in theory, but realistically the hooks are often broken from version to version, one subsystem to another. Not that I can really expect much, if the kernel team won't bother to maintain stable APIs why should anyone else? You can replace subsystems to your heart's content as long as the APIs don't change. No one, however, seems to care about maintaining a stable API for desktop Linux. They just don't see it as important for some reason.


That's bunk. The only reason for stable in-kernel APIs would be to allow outside coders easier maintenance for their patchsets. The Linux developers don't care about that. They want development in-tree. If you don't want to get your stuff into the tree then they don't care about you. Too bad. All kernel -> userpsace interfaces have been very stable. As for desktop APIs, they are also very stable. How long has GNOME 2 been around? 8 years. How long did KDE 3 stick around before KDE 4? 6 years.

Reply Parent Score: 2

RE[2]: Deprecation
by nt_jerkface on Sun 17th Jan 2010 06:06 in reply to "RE: Deprecation"
nt_jerkface Member since:
2009-08-26


It always boils down to just the size of the market...


No it doesn't, it's just a factor in the decision to port.

If porting to Linux from OSX was as easy as setting a compiler flag then it would have nearly the same library.

However the situation is the exact opposite where the cost in porting to Linux well beyond what it should be for its size.

Linux is not a stable platform for commercial developers. It isn't even a single platform. It's a bunch of operating systems that share the same kernel and have software distributions designed around open source.

As I have pointed out before it's far easier to build your own Linux distro that contains your proprietary program than it is to support a single distro. The people that build the distros don't at all care about attracting commercial developers. They also don't care about being compatible with other distros.

6 months after the iphone was released it had better support from game developers than Linux even though it had a fraction of the market size. Market share is only part of the equation and doesn't matter much when the people behind an OS could care less about the market.

Reply Parent Score: 6

RE[3]: Deprecation
by Kyuubu on Sun 17th Jan 2010 12:46 in reply to "RE[2]: Deprecation"
Kyuubu Member since:
2007-09-07

6 months after the iphone was released it had better support from game developers than Linux even though it had a fraction of the market size. Market share is only part of the equation and doesn't matter much when the people behind an OS could care less about the market.

While I may agree with some of the above, here I think you're mixing things up. The iPhone opened a new kind of market for very casual games, all bundled with hype. Of course it appealed many developers. It's not a part of the old gaming market, not yet at least, and on the beginning didn't play by the same rules (again, I feel hype played its role).
And i'd be curious to see some numbers about both (phones/linux) market sizes... We may be surprised.

Reply Parent Score: 3