Hi Kel, (mostly) just commenting on two aspects here, bcc:ing Felix again, so he can comment on the technical details.
On Wednesday 17 September 2008 00:45, Kel Modderman wrote:
> If forced between maintaining an awful snapshot of dead end crap for the
> lifetime of Lenny just because I am not allowed to have it removed for what
> I feel valid and compelling reasons, or getting to update the package
> version (at an unsuitable time of release phase) to something that is
> likely to be easily maintained for Lenny shelf life, I would probably take
> my chance with a new version, even though I cannot properly test it.
Great.
> However that version would (most likely) be based on 0.9.4 and not
> supporting new chips, like for example the one that Rene owns, or the chips
> that Julien claim to be unsupported by ath5k, because the 0.9.4 branch is
> the only branch on madwifi.org that functions with some integrity, but it is
> based on an old binary which dictates hardware support.
I have no idea on this one.
> In fact I prepared an upload of 0.9.4.0 (which would supercede the current
> SVN snapshot) and requested it to be uploaded but my request was ignored,
> possibly because of pressure to keep a version of madwifi that supports
> atheros hardware in recent mac's and eeepc hardware in the archive, I don't
> know.
Neither.
Well, I have some idea: I know that Felix/the Openwrt branch is based on
madwifi-0.9.4+r3314 from madwifi and then has approx 100 commits on top,
incorporating the (useful) stuff from the forthcoming and parallel 500
commits in madwifi trunk. (Based on/written for the new binary Felix
released.)
But Felix needs to comment what that means for macs and eeepc (and other
hardware (but I _assume_ basically all wlan capable hardware is supported by
openwrt, as are the archs which matter for debian)).
> The reaction against my proposal pisses me off, no-one comments on the
> other reasons i put forward which make madwifi unsuitable for a long term
> stable release, instead they just complain that there hardware won't be
> supported by some non-free m-a package. They could at least offer to help
> out with further action that may be needed during the Lenny lifetime or
> something, to give me confidence enough to believe that the packages would
> be providing good service to the users of Lenny. My last few requests few
> new uploads simply fell on deaf ears, one may search the pkg-madwifi
> archive on alioth if they need evidence of this.
Well, on my usual atheros hardware I run openwrt currently and the time I had
a thinkpad with atheros hw and running debian, I also used a madwifi driver
from openwrt ;-) Thats why _I_ didn't comment earlier. (And I knew a bit
about the driver mess and the new hal on the way...) Anyway.
And as said in the previous mail, I think this hal can and will be maintained
by Felix (and the madwifi stuff by all madwifi maintainers in Openwrt), so I
see this as a good reason for letting it in and switching, even now.
(At least compared to the other two options, removal or keeping the current
version.)
> BTW, I have pointed the kernel team toward a patch series which improves
> the ath5k driver in linux 2.6.26. I also have facilitated the removal of
> wpa_supplicant's specific madwifi driver wrapper some moths ago. I am not
> mucking around here, if someone want to support this madwifi stuff let them
> step up, it will not be me, by myself, any longer.
I'm considering stepping up (not sure yet if I want to help out or really join
the team...), if it makes sense for lennys lifetime...
regards,
Holger
pgpIHLc14R6Cp.pgp
Description: PGP signature
