Removal request: madwifi, madwifi-tools

[Santiago Garcia Mantinan]

Hi!

Sorry to jump on the discussion so late, I only realised about the discusion
today when Kel pointed me to it after I had filled more info on bug #492251
because of the bad state current package is on.

I see two issues here:

- Maintainability of the packages we "ship" with (even this is not main but
  non-free)
- Support of old and new hardware

I see that the current code on Debian doesn't meet any of the two points, it
fails to meet the first one because it is a snapshot taken out of a
development tree on which work seems to be at least dying, and it fails to
meet second point because if it does support new hardware it does so by
causing a lot of hangs on both this new hardware but also the old one (see
my latest post on bug #492251).

So... as I already sugested on that bug, for me the ideal thing would have
been to stick with the stable madwifi (with the necesary patches for current
kernels) and then add a new madwifi package with support for the new
hardware, this second package should now be based on the code from the
dd-wrt and openwrt guys, as it seems quite stable to me (only had it for the
last three days but on this time it already hanged my AP which has etch with
0.9.4 while the client based on the the new hal didn't hang at all).

I don't know if it is too late for all this, given the usual rules for a
release schedule it would be, but as this is non-free stuff and we have a
real problem with current code, maybe it is not that late.

What we win with having two packages, one based on 0.9.4 as Kel wanted, and
another one based on the new hal?

We have supported code: I mean that if madwifi continues to live they should
support 0.9.4 which is their latest stable version and also the openwrt guys
will support the code they use or plan to use on their distro (current code
seems not supported by anybody).

We have supported hardware: The stable version of madwifi supports well the
old hardware giving support for features that are not on the free drivers,
and the new hal support the new hardware (correct me if I'm wrong, but seems
stable and if not, it will be supported during lenny).

The cons will be that we have to give support to two packages instead of
one, but it seems that these two packages would be better supported upstream
than current one, and that we wouldn't be promoting too well the free
alternative, which is false as the free alternative is the one that comes
activated by default and in the main part of Debian.

So... I'd like to know how would people feel about this two packages scheme,
which would mean to go back to the old stable code like kel wanted and have
a new package around to support new hardware.

Regards...
-- 
Manty/BestiaTester -> http://manty.net