Re: comedi as a POSIX standard

On Sun, 13 Apr 2003, Steven Jenkins wrote:

> Herman Bruyninckx wrote:
> 
> > Yes, but why choose POSIX then? Why not the Open Management Group, or
> > some other standards body?
> 
> I'm not arguing against any other body, I just happen to be more 
> familiar with both the process and the product of IEEE. The point is not 
> to get some organization's imprimatur, it's to get a really good 
> standard. I think the IEEE's process leads to a pretty consistenly good 
> product.

Anyway, I guess the Comedi project itself has to do most of the
preparation work. And in fact, the DAQ API standard is almost there:
it's not so big after all.
> 
> But to address OMG specifically, OMG is not a standards body. It's a 
> consortium, largely-vendor driven, that publishes specifications. I'm 
> not well-versed in their more recent specs, but the early CORBA specs 
> were mostly marketing fluff. Being CORBA-compliant didn't mean your 
> application your interoperate with multiple vendors' ORBs, nor that the 
> ORBs would interoperate with each other. In fact, they didn't. Things 
> may have improved since then, but I'm skeptical.

And what's better with POSIX? You pay tremendous amounts of money to
get your implementations certified, and that's not IEEE who's doing
the certification.

> You don't have to be an IEEE member to participate in (or even lead) a 
> POSIX working group. (I think. But even if you do, it's around $100 per 
> year.) It appears to me from OMB's web site that you can't get voting 
> privileges in an Task Force for less than $2200 per year. OMG doesn't 
> have individual memberships, as far as I can tell. IEEE has nothing 
> *but* individual memberships.
Good point!

> The Open Group would be a better option. I think their track record for 
> publishing real open specifications that mean something is better. They 
> have a Real-Time & Embedded Systems Forum that might be interested, and 
> they do have some precedent for publishing their specs on the web.
Indeed. 

So, let's get concrete :-) What would be the minimum starting effort
that could lead to a standardization? Describing the Comedi API as in
done in the current documentation? As I said before, that's not a big
job to do. (And I was planning to do it anyway, because I'm involved
in two other projects that would indeed love to see some "standard"
way to interface DAQ cards :-)

> P.S. I guess my email address suggests a disclaimer. I have no 
> connection with the IEEE other than being a member.
I'm a member too :-)

Herman
-- 
  K.U.Leuven, Mechanical Engineering, Robotics Research Group
<http://people.mech.kuleuven.ac.be/~bruyninc> Tel: +32 16 322480

Received on 2003-04-14Z05:05:58