aac support

[Theo de Raadt]

> There has got to be a better way to work with the vendors in order to 
> get the support we need. It just seem to me that the "screw you guys, I 
> am going home" stuff just does not work.

Well, there is.

We do it all the time!

We mail a vendor, and then we start a frank dialogue.  I (or some
other developer, maybe even Bill Paul from FreeBSD
(Mr. Ethernet)... anyways, people like that.. ) explain the business
case to the vendor.

They almost always understand, and then give us documentation.

Sometimes they open the documentation wide up!

Sometimes they are willing to give us documentation as long as we do
not distribute it too far, and we are willing to do that.  We normally
share it with, say, 3-4 developers, to ensure that the job gets done
and that there someone can fix it later.  This also ensures that the
documentation stays around in someone's hands even if the company goes
away (like Adaptec might after the FTC gets finished with them?)

I spend a LOT of time explaining the business case.

When vendors do not work with us, they are the odd vendors.  Normally
they are companies with strong USA stock profiles.  I don't know if that
has something to do with it, but I suspect it does.  And normally they
are ones that people, down underground, know produce crap.  This also
ties into how sometimes it is very hard for us to support their hardware.

But, and I must emphasize this, 90% of companies *do* come around.

> The vendors need a business 
> case in order to do things - they are in business to make money and I 
> can agree with that. Maybe we can do some sort of list of companies or 
> OpenBSD people that use or would use the cards - along with number and 
> install base study of the number of sales they would get and give it to 
> them. We should work on some sort of cookie cutter type setup that 
> tracks the interest and $$ with a product that we can compile and be 
> sent to the vendor in order to get support. The data needs to be 
> correct and true and presented in a business case manor. The one-off 
> flock of emails just do not work. I would be happy to help with this 
> and pursue this if there are others that think it is a good idea.

I have thought about doing this, but it is a lot of work.  I think we
all know what needs to be done to make this accurate.  It is a very
big job and it needs one passionate person to run it from start to
end.  It cannot at this time be me, sorry.

> Also is there needs to be a stock form that is send the vendors that 
> covers in detail what we ask for.

I do not think so.  I write each mail to the vendors individually, taking
the situation and the market into account.  I research the market at the
stock level, and I ask people in various parts of the world to help me
form a profile of what chips are showing up there.  If not done carefully,
they will be right to take me for a crank.

> Some that can be vetted by their lawyer that they would be OK with.

When the lawyers get involved, that is when the companies make bad
decisions and lose.  OpenBSD 3.7 will ship with aac off.  Adaptec just
lost.  No matter how they sell it within their own ranks, they just
lost.  Unless they have something to hide, like crap cards with
hundreds of unrepearable bugs and a history of selling crap to
customers after knowing that their product was not meeting the
promises they make.  But what do I know for sure.  I do however
believe they are balancing two choices of reality.

> We need to work with the vendors in 
> a clear, clean business like manner and leave emotion and philosophy 
> out of it.

I do.  It is hard.  I do it every day.  Last week we got Ralink
documentation.  I am working on Realtek for their 802.11g docs now.
And in a few days, if Realtek keeps stalling me, you will hear from me
as to where to send your mails.

And then we can get further at supporting a chipset.

One way or another, at some point we must get *ahead* of Microsoft
at supporting new hardware products on the market.

(Show this previous line to your Linux friends)