Removal request: madwifi, madwifi-tools

[Kel Modderman]

Hi Felix,

On Thursday 18 September 2008 02:27:04 Felix Fietkau wrote:
> Kel Modderman <kel <at> otaku42.de> writes:
> > the following facts are true (correct me if wrong):
> > 
> > * we must revert to a snapshot of madwifi.org trunk at svn revsion 3314,
> >   madwifi.org trunk is at revision 3856, the package in Lenny is based
> >   on a branch of revision 3772. This means discarding a few hundred commits
> >   to some point in time when the fork was taken.
> See below.
> 
> > * we must apply all patches: [...]
> >   that's a lot of patches! who has reviewed them? who has tested them?
> They haven't received much public review yet, but I did receive some feedback
> from various random people looking at them.
> For me it boils down to this: What do you prefer, something which quite a few
> people successfully use on embedded hardware and which maybe has a chance
> of working well for laptop users as well, or something which is already known
> to be broken for many people?

I prefer neither. I prefer the madwifi packages to be removed. After time,
when/if the code stabilizes as part of a group effort including the madwifi.org
project, then there would be possibility to provide it again, though I expect
the free drivers to be better by the time that is a relevant topic again.

This entire thread is in the context of a Debian distribution release that must
happen too soon for any of that to occur. Meanwhile, someone can keep what ever
version of madwifi in the 'unstable' archive as they like.

> 
> > * we must trust a binary HAL that was only announced to the public on the
> >   16th of September 2008 by Scott Raynel, who announced it because Felix
> >   couldn't really be bothered.
> There were multiple parties involved in the HAL release. The announcement on
> the DD-WRT site was done by DD-WRT people, the announcement on the OpenWrt 
> site was done by me (OpenWrt developer). The release date was communicated to
> the madwifi project in advance and the announcement itself was posted as news
> entry with little delay.
> Since I had lots of other stuff to deal with (just moved from Hamburg to 
> Berlin yesterday), I chose to leave MadWifi's announcement part to the MadWifi
> project.
> Anyway, so why do you put the HAL binary trust issue in the same category as
> the lack of a real announcement to the list?

Blowing off a bit of steam maybe. Sorry.

> 
> I have been actively working on fixing whatever HAL issues were reported from
> MadWifi users, and I will continue to do so, until ath5k is at the point where
> it fully replaces MadWifi.
> 
> > I do not like the idea for _Lenny_. I do not like that this is a fork of the
> > existing project. It indicate that Felix does _not_ want to put effort into
> > supporting madwifi (which is not just users of it, but developers of it too),
> > he only has time to put effort into Felix version of madwifi, which is
> > primarily targetted at users of embedded devices and primarily serves the
> > needs of Felix and his clients.
> Sure, I did put effort into supporting madwifi. When I started building my
> patch stack on top of r3314, I gave several madwifi developers all the info
> that they neeeded to evaluate my patches and decide whether or not to commit
> them into the madwifi svn. I specifically told most of the active developers
> what to merge, what to leave alone until it stabilizes, which patch implements
> which functionality and why, etc.
> I only stopped when I found that the MadWifi developers didn't really have
> time or didn't care to review my stuff - and I couldn't stay in sync with
> trunk because of constant random breakage going on there, which would have
> used up more time than I could spend on this driver.
> Of course, my primary intention was to have a stable wifi driver for OpenWrt,
> but please don't accuse me of not putting any effort into supporting MadWifi,
> which is something already disproved by the fact that I worked hard on making
> this coordinated HAL release happen. I could easily have turned it into
> something OpenWrt and DD-WRT specific, but I really wanted the MadWifi project
> and its users to have their share of the fun as well.

If what you say is true, someone can provide madwifi for Debian distribution
with confidence in the future, bit as it stands it all a bit up in the air
wouldn't you say? If I told you I had been stabilizing a snapshot of madwifi
for the last few months, and that you should put it in your next release of
OpenWRT which is due in weeks/months would you do it?

Look, I saw that there is a huge problem with current madwifi in our
Distribution, I did the gentlemanly thing and requested this bad stuff be
removed. It is not possible to change what is in the archive now with
new versions, those are the release teams rules. These other people try
now and I wish them luck but think it is a foolish idea.

1) current madwifi is screwed it is unfit for release
2) cannot put massively new versions of madwifi into archive at this time
   (especially from totally new upstreams)
3) this is at this stage a distribution specific issue, which is not relevent
   to you unless i missed your participation in Debian somewhere
4) getting sick of telling people points 1 and 2 so fast it is not funny

I don't think there is much point replying to me Felix, this topic i believe to
be a waste of your and mine time. Sorry to have stirred you up.

Thanks, Kel.