Re: [Madwifi-devel] Re: Why Madwifi isn't integrated into kernel?

[Jim T. <ji...@ne...>]

On Feb 17, 2005, at 1:51 PM, Chris Wedgwood wrote:

> On Thu, Feb 17, 2005 at 09:35:22AM +0100, Michael Renzmann wrote:
>
>> But at this very point this discussion is useless. Madwifi isn't
>> ready for kernel inclusion, we don't even have a single release
>> currently. So this should be the first milestone to achieve, before
>> thinking about how to get madwifi into the kernel.
>
> Rightly or wrongly, the networking people don't want yet another
> 802.11 stack in the kernel.  They would prefer to see the hostap code
> cleaned up (this may have been done by now, but it's certainly not
> merged whatever the case).
>
> The hostap code is an 802.11 stack and people can (and do) write
> drivers for specific hardware/chipsets to use this. I would expect to
> see this merged in the next few months with support for some prism,
> centrino initially and probably atheros and realtek a bit later on.

Note that the realtek rtl8181 driver already makes use of the 802.11 
stack that is used with madwifi.

The hostap code is not as good a start on an 802.11 stack for linux as 
is the stack used with madwifi.

If this really is the opinion of the "networking people", I'd love to 
know who they are, so I can offer verbose
opinion on the subject.

If you didn't know, the Prism2/2.5/3.0 chipsets are, for all intents 
and purposes, EOL.   Connexant is cleaning house,
and the 802.11MAC group (formerly Digital Ocean, in San Antonio) are 
all out looking for new jobs.   It would be a mistake
to base the future of linux's wireless networking support on the hostap 
code.

Jim