On TV.com: 5 SCARIEST Episodes in TV History
BNET Business Network:
BNET
TechRepublic
ZDNet
TalkBack 9 of 22:
Next »
« Previous
Let's put the REAL issues on the table
Linux (or Unix) was not designed to run unknown binaries. Unlike MS Windows (or other proprietary OS), out-of-the-box compatibility was not the intend. If I want to optimize my binaries, I customize my gcc on MY machine and compile it to MY specifications.

Here are the real issues: 1) more and more people expects Linux/Unix to run like MS Windows; 2) less and less people knows what they're doing when they dive their heads into Linux; 3) hardware developers got used to the proprietary model and refuse to release their code; 4) more and more software developers (on Linux side) do not want to release their code so they can make some money. Not releasing the source isn't the problem, it's that they also refuse to (or can't) compile binaries on a per customer bases.

Creating a common driver interface sounds simple enough, but it's next to impossible on Linux. An interface that encapsulate all possible combinations is going to be bulky as hell, and I won't want any part of it near my kernel.

Unless the interface is "hard-wired" to the kernel, there'll always be others like me opt not to use it, thus defeating the whole idea behind the word "common".

Besides, the bulkiness of a common driver interface would kill any kind of speed optimizations on the hardware side. Not all hardwares have low data rates like a mouse, you know. wink

So it comes full circle - to optimize the driver, it needs to be compiled on the machine. There are just too many sacrifices to be made behind the word "common".
Posted by: toomuchgreeatea@...   Posted on: 05/07/05 You are currently: a Guest | Members login | Terms of Use

Alert moderator to an offensive message

Subscribe to this discussion via Email or RSS

Great to have the best and brightest working on the compiler.  DonnieBoy | 05/06/05
Another thing, we need binary compatibility accross distributions/kernels  DonnieBoy | 05/06/05
One would think that there would be a way ...  George Mitchell | 05/06/05
The problem is venders have to supply 10 or so drivers.  DonnieBoy | 05/06/05
And I'm telling you that the answer is a wrapper  George Mitchell | 05/07/05
Only if you go lowest common denominator  Knorthern Knight | 05/06/05
This is true but  hipparchus2000 | 05/07/05
Kind of like the idea behind NDISwrapper  Hugh Jass | 05/07/05
Let's put the REAL issues on the table  toomuchgreeatea@... | 05/07/05
Those are only issues for *YOU*...  figgle | 05/09/05
I'd rather have a slow driver than none at all...  figgle | 05/09/05
Binary compatibility...  Henry Miller | 05/07/05
This guy must be a newbie in Linux  toomuchgreeatea@... | 05/07/05
If you care about optimizations  balsover | 05/07/05
Improving Optimization in GCC.  mobrien_12@... | 05/07/05
No it is not a waste of time  balsover | 05/08/05
GCC can specialize . . .  CobraA1 | 05/08/05
GCC tries to specialize  balsover | 05/08/05
CobraA1, you just got spanked.  No_Ax_to_Grind | 05/08/05
What optimisations?  Richard Flude | 05/08/05
Interesting Linux win over Windows ...  George Mitchell | 05/07/05
More of a IBM over Dell story  balsover | 05/08/05

What do you think?

SponsoredWhite Papers, Webcasts, and Downloads

advertisement
advertisement

Meet Doc

  • Here to help you with your Document Management Needs
  • Doc is an enigma. Born to a Russian ballerina and a German electrical engineer, he grew up in various locations in the United States. He’s seen the insides of more brands, versions, and generations of printer and printer-related hardware than almost anyone.
  • To learn more about this mysterious figure check out his blog on ZDNet and his Workspace on TechRepublic. You’ll be glad you did.
  • Produced by
    ZDNet and