> > mmmh ... you've got a point ... you just opened my eyes ... documentation > > is > > pointless when I have a blackbox doing the work. > > Maybe I'm missing something, and if so, I do apologize to those on these > lists that I may have offended ... but ... having clean source code to a > driver is not equal to a black box ... is it? I don't believe I've once > advocatged 'binary drivers' ...
Approximately six years ago Intel gave the *BSD projects a driver for the Intel gigabit cards, the so-called em(4) driver. They refused to give documentation to the developers then, and they still refuse to give documentation to the developers today (A few people have documentation, but these are rare individuals, typically operating in some sort of support role with a large customer of Intel). That driver was garbage then. Today, six years later and 146 revisions later (in OpenBSD) that driver is still largely terrible. We know why it is terrible. The hardware is not that great. We know one reason why we never got documentation. Bit by bit more information has come out to show that the hardware design is an embarrasment and there are countless bugs and shortcomings. Marc, you cannot program, which is why you will never understand. You should just drop this entire discussion because you are completely wrong. In fact, it is bleating from people like you that, over the years, has let the vendors off the hook when they refuse to supply documentation. OpenBSD people ask vendors for documentation, and FreeBSD people attack us, and the vendor goes "phew, we can get away with not giving them any documentation". Don't just blame (some) Linux developers for signing NDA's. A lot of that also happened in the particular BSD community who you are associated with. And when that guy who signed the NDA (Hi Scott, Hi others) stops giving caring (like the vendors), the driver dies. You all know this is true. You've all seen it happen countless times. And you know there is a better way.
