[pkg-wpa-devel] Bug#489920: Bug#489920: madwifi no longer supported, ath_pci doesn't work anymore

[Kel Modderman]

Hi martin,

On Wednesday 09 July 2008 21:13:55 martin f krafft wrote:
> forcemerge 487585 489920
> thanks
> 
> also sprach Kel Modderman <kel at otaku42.de> [2008.07.09.0227 +0200]:
> > Where do you get your information from? Madwifi and ath5k support
> > the WEXT backend, it just so happens that ath5k doesn't work at
> > all very well yet.
> 
> I've been happy for years with the ath_pci driver and madwifi
> backend. When I try the wext backend, the driver cannot make an
> association, wavemon keeps showing an empty ESSID, and the invalid
> network ID counter keeps counting upwards. If I revert to the
> madwifi private backend, it works fine.

Did you agree with the discussion put forward in referenced thread and #487585?

> 
> > I suspect this is a duplicate of #487585. Please confirm, so that the bug
> > reports may be merged.
> 
> Yes looks like it, sorry.
> 
> I now wonder how to move forward. I don't have connectivity on my
> laptop anymore and have to type this elsewhere, so getting debug
> information is going to require a lot of extra work. I'd appreciate
> if my hand was held a bit. :)

Section 4. Troubleshooting of /usr/share/doc/wpasupplicant/README.modes.gz
is a starting point. A debug log of wpa_supplicant may be valuable should you
report problems to madwifi.org bug tracking system.

http://madwifi.org/wiki/TicketSubmissionGuidelines

These existing tickets may be related:
http://madwifi.org/ticket/275
http://madwifi.org/ticket/1817

Madwifi also has a mountain of debugging knobs to play with, but I am not
familiar with them. The wiki on madwifi.org, or support channels (eg. irc,
mailing list) may be of value.

http://madwifi.org/wiki/Resources

Also note that the most active Madwifi upstream developer, Matthew aka "mentor"
Bell is also assuming an active role in Debian maintaining the pkg-madwifi
packaging. It may be worth persuing this problem via a bug report against the
madwifi-source Debian package to open dialog with him.

I think I am correct in saying that Matthew would prefer to debug WEXT/scanning
related problems than reintroduce and depend on the legacy crap. Having said
that, #487585 does contain an invitation for outside help should the copy of
madwifi headers be re-introduced into wpasupplicant package for the purpose of
re-activating the driver_madwifi wpa_supplicant backend.

(I am also going to have less time to spend maintaining wpasupplicant package
soon, so extra help would be appreciated regardless of this issue...)

> 
> PS: I went a bit berserk after seeing that the latest wpasupplicant
> in sid has UNRELEASED in the changelog.
> 

I am sorry about that, but I do not have the rights in Debian to tag releases,
and only caught that error after it had been dispatched by the uploader.

Thanks, Kel.