[Prism54-devel] Re: wireless API semantics when netif down?

[Jean Tourrilhes]

On Mon, Dec 01, 2003 at 01:00:18PM -0500, Luis R. Rodriguez wrote:
> 
> I personally feel that we shouldn't be burdening Linux wireless driver
> developers with "hackish" workarounds just so that drivers work flawlessly
> around Distribution network scripts.
> 
> Why not get things straight and define semantics that *should and are
> proper* ways for a driver to work and then just have the Distributions
> fix/update their scripts?
> 
> 	Luis

	Please reread my initial answer. I explain in detail what is
the proper semantic and the reason why it is done this way.
	http://prism54.org/pipermail/prism54-devel/2003-November/000088.html
	For you information, most distributions scripts are doing the
right thing, so they are not going to change. If you don't agree with
the reasons things are done this way, we can discuss it, but don't
claim it's wrong.

	Now, there is multiple way to implement this semantic, some
simple and some more complex. Caching the firmware MIB is not
mandated, it's only an optimisation, and I can point out a few drivers
that don't do it. If you don't like, don't do it.
	Don't be too pedantic about the corectness of your
implementation, I think you are going a bit overboard. For example,
the Orinoco driver will report outdated settings in iwconfig when the
card is "down", because it make things simpler ; so far, I had 3
people complaining about it. The Aironet driver doesn't perform
Scanning when the driver is down, I had 1 person complaining about it.

	Have fun...

	Jean