Linked by Thom Holwerda on Mon 14th Mar 2011 18:59 UTC
Talk, Rumors, X Versus Y And over the weekend, the saga regarding Canonical, GNOME, and KDE has continued. Lots of comments all over the web, some heated, some well-argued, some wholly indifferent. Most interestingly, Jeff Waugh and Dave Neary have elaborated on GNOME's position after the initial blog posts by Shuttleworth and Seigo, providing a more coherent look at GNOME's side of the story.
Permalink for comment 466278
To read all comments associated with this story, please click here.
RE[4]: F**k this shit!
by oiaohm on Tue 15th Mar 2011 21:37 UTC in reply to "RE[3]: F**k this shit!"
Member since:

Interfaces are not supposed to change rapidly. It's a key software engineering concept ... the interface remains the same the implementation behind it changes.

If the interface does need to change you depreciate the old one and give people time to move over.

The only reason they keep changing it is either poor design or it is a deliberate attempt to force other devs to open source their drivers (if this is true it is another case of "freedom but as we tell you).

Funny Linux did support Unified Unix Driver Standard. No drivers were made for it. It was binding up a kernel mode ABI for no gain. Developers insisted on calling the internal API's they could see not the stable ABI. Was even supported in the 2.6 line. To state what the hardware makers said repeatly for not using the Unified Unix Driver Standard was that the performance hit(less than 0.01 of a percent) was too great.

People can bash Windows all they like but a driver written for Windows XP in 2001 will still work with Windows XP today, the same is true also with Drivers between Solaris Versions.

Pardon. Windows XP drivers written in 2001 still work with Windows XP today. This is freeze kernel progression. Yes that was released is the 2.4 tree that was first released in 2001 that has a stable kernel abi across the 2.4 line has basically no closed source drivers.

Now lets move on to vista. Vista just like Linux kernel 2.6 is pushing lots of drivers to user-space for the same basic reasons no need to tie hands behind back.

Linux dev's really have no excuse.

Problem is people like you are blind. Please note the kernels tagged longterm. They will be API compatible for as long as XP if not longer. Why not ABI. Something interesting. Turns out you must use the same compiler to have ABI. Reason why MS shipped driver development kits containing a different compiler to the normal Windows SDK. If you don't have the same compiler you must wrap the API that does cost performance.

Anyone who builds open solaris themselves with different compilers has also found out that from time to time solaris closed source drivers don't run stable either. So this is a selection between stable and not stable.

Userspace is already wrapped with the syscall framework. Userspace is simpler to provide compatibility libraries. Something people are not aware is some of the old syscalls on linux called from userspace are not processed by the Linux kernel but redirected to userspace libraries. So historic compatibility does not mean kernel bloat.

A userspace driver frameworks its far more stable. Drivers written in them like cups drivers you can pick up cups drivers from 1880~ from a few different unix systems and use them on current Linux by using loaders. 1993 from Linux system and use them as well.

Basically userspace proper solves the kernel to kernel issue. And driver support from hardware makers has been as bad as it always was.

I am sorry but the userspace framework on scale of stable is massive far passing the time frame any Kernel base ABI could offer. Its done in such away they never need to be revised in a way to break backwards compatibility. At worst redirect some syscalls to userspace for userspace handling.

Reply Parent Score: 3