[opensolaris-summit] driver limits [was Re: My comments (very subjective) on proposed Summit topics]

[Philip Brown]

On Wed, Sep 26, 2007 at 08:18:21AM -0700, Garrett D'Amore wrote:
> James Carlson wrote:
> >....
> > Thus, DDI isn't a mandate.
> >   
> 
> Historically, believe I had the only DDI complaint ethernet drivers that 
> were to be found... either inside ON or outside of ON.  I believe that 
> this signaled both deficiencies in ON, and in the various unbundled drivers.
> 
> I believe that sticking to the DDI, while not a hard requirement for 
> drivers in ON, is a worthwhile cause even for the drivers in ON.

Gaaaah... 

DDI *should be* a mandate, even for drivers in ON.

Why? because it ensures that the DDI level stuff, stays actually USEFUL.

Multiple times, I think, I was trying to write a driver, and looking
through the DDI for nice clean ways to do it.. 
and, there just wasnt a way to do it!
There were existing sun drivers that did similar things, and they got the
job done just fine... by going around DDI.
But for external people, "you cant get there from here".


So, the answer to the long-standing question/gripe of,
"why are there so few drivers by non-sun people out there?"
would be at least in part,

"Well, gee, becuase sun didnt provide non-src-privileged people a means of
 writing them!"


So, I doubly underscore the later paragraph from Garrett:


> 3) By using the DDI, you know that you're using the same APIs that we 
> recommend to external developers.  That means you help to uncover any 
> limitations in the DDI.



Even in this modern day and age, where people now have "access to the
source"... there is something very positive to be said about having an
official DDI for things, that must be adhered to.. because then,
(ideally), the interfaces would actually get DOCUMENTED...? 

Believe it or not, open source developers DO actually appreciate
documentation (even if they rarely write any themselves ;-)