Re: [madwifi-project] Death to MadWifi!

Re: [madwifi-project] Death to MadWifi!

[Luis R. Rodriguez]

On Mon, May 11, 2009 at 10:27 PM, p0g0 <p0g0@madwifi-project.org> wrote:
> Hi Luis,
>
> As one of the Madwifi team members when you first joined, I'd like to speak
> to your refreshingly honest admissions.
>
> Luis R. Rodriguez wrote:
>>
>> Its no surprise I hate MadWifi.  And its no surprise I've wanted to
>> kill it since I joined the project.
>>
>
> You hatred is one thing, but by your own belated admission to murderous
> intent, your willful manipulation of the Madwifi project to your own
> advantage is a lesson for us all. Cui Bono? Luis, how do you like your shiny
> new job working for Atheros?

Where I work has to do with choosing to work on what I think is best
and doing it right.

> As the guy that co-wrote the part the Madwifi mission statement that sought
> to promote _all_ linux drivers for Atheros gear, you were not honest with me
> or others when you joined the team. How should those of us who spent years
> volunteering to promote Atheros hardware under Linux take your hatred and
> inhuming intent?

Now I'm inhuman? Heh.

> You have, years after the fact, admitted you were not Open
> (as in FOSS) about your ambition to hijack an ongoing project so you could
> fly it into the ground and trash the good work done by dozens of people, to
> "kill" it's progressive initiatives, and forge your credentials by claiming
> to represent that project's interests. Remember how forcefully you wanted us
> to join your lawsuit against Atheros? You wanted Madwifi's help, even though
> you intended to kill the project you were asking to help you. It's not a
> pretty lesson for volunteers wanting to support Linux.

I made my intentions very clear -- MadWifi was just not the right way
to do a Linux driver and that's what my focus has been on  -- on
upstream drivers.

> Duplicity is generally not seen in a good light, and most employers and
> co-workers would have second thoughts working with someone that they knew
> operated that way. Had you genuinely embraced the principles of FOSS, the
> freedom and the openess, and shown a willingness to compete on the merits,
> you would have not hidden your motives or intentions, and we'd have heard
> you say then that you joined our team with the intent to kill the project.

The "project" was to get devices properly supported in Linux, I joined
to help fix that. There are still some who keep on using the old
driver but by no means was keeping an ancient relic the goal of the
project. I'm not trying to kill the MadWifi driver -- that's already
done, and by no means am I trying to kill the "MadWifi" project, which
is what you try to allude and which is very different. The project has
served its purpose as a home for a place to find drivers for Atheros
devices for Linux. As time goes by that place has now shifted to
upstream and support comes out of the box. So there are some obvious
shifts in what it means to support Atheros devices on Linux. If to you
that whole shift is killing "MadWifi" then yes that was my intention
all along and I think I've always made that very clear.

> I'd guess that you knew that such honesty was going to interfere with your
> hatred and your plans.
>
> It may seem that success belies all, and ends justifies means, but in fact,
> many of the things you pursue were already under way, and they would have
> born fruit. You can make a valid claim to success, but you have no basis to
> claim our initiatives would fail. I'd further note that you, sub rosa,
> worked hard to assure that some Madwifi team efforts, extant when you
> joined, would fail. But Madwifi has not sought to kill your project.

Its not my project, its upstream, and um... upstream.. is umm.. Linux...

> In
> fact, we codified in our mission statement that we'd openly support
> competing drivers, and we've delivered.
>
>>  I think its bled to death now but
>> for some reason I keep seeing people giving it CPR. That's fine and
>> all but I keep getting random complaints from people about "not
>> supporting devices upstream yet". Of course these are just comments
>> from users not so well informed but at least one user has one point:
>> its not their fault they're not well informed.
>>
>
> Luis, you approached the Madwifi project with knives and zero sum thinking,
> presumed that all the developers would only support one codebase, and that
> they would have to quit Madwifi so that you could capture their volunteered
> work. Surprise! You only had a handful of Madwifi developers quit and move
> over to your project (though it appears you did convince some to just quit
> developing linux wireless drivers): it's not a zero sum game, and the AthX
> developers came from a different pool than the Madwifi developers for the
> most part.

I like to think of it more I was trying to beat explain and convince
why upstream is the proper way to support drivers in Linux, some did
understood it, some others didn't. Oh well.

> Developers, old and new, should work on the Athx code, it has
> merit. So does Madwifi. Some Madwifi developers have good reason to carry on
> with ongoing work and support. The two drivers are not mutually exclusive.
> There is no "must" there.

Sure, but its a small group with special interests and the intention
of my e-mail is to propose to explain that on the front page. Users
looking for drivers should just upgrade their kernels or use things
like compat-wireless.

> Please compete on the merits. You have the high ground: AthX code is
> distributed as part of the Linux base and that is plenty of advantage. Quit
> with the "Madwifi must die," it begs the Shakespearean notion of protesting
> too much.

Heh sure, this is true, point taken.

> If Madwifi is due to die of natural causes, fine, but you are not
> showing confidence in the AthX work when you worry about the competition so
> publicly.

Trust me, I'm not concerned at all -- reason for me sending this
e-mail was having to hear a rant from a user about MadWifi being such
crap and about devices not being supported properly. They were simply
unaware of upstream drivers. To the novice user who rarely upgrades
kernels and who apparently install Debian Lenny which is using 2.6.26
they don't get ath9k and to them support doesn't exist.

> Given that folks can now chose, your intention to deny them a
> choice of drivers makes me curious.

You missed it, its alien technology:

http://groups.google.com/group/linux.kernel/browse_thread/thread/8a17cfefaf142655/0ac3713c2fc47f18

> If I follow the money, the only benefits
> I can see to cornering the market on FOSS Atheros software goes to your
> employer.

Seriously though -- its kind of funny the type of things some people
will think of when a company starts supporting things upstream or
while they are actively supporting.

Its very simple: proper Linux support is upstream and companies slowly
start realizing this. There is just a big catch up game to be played
and it takes a while. There is nothing obscure about it or some hidden
reason why to do this. Its just the right way to support Linux.

> Leave Madwifi-Free and the volunteers alone, and let it succeed or
> fail on it's own.

No -- the MadWifi project is about Atheros Linux support, it includes
both upstream drivers and that driver thing called "MadWifi". Feel
free to keep on hacking on that driver thing, but don't tell me to not
use MadWifi to properly educate people on how to get their devices
supported properly with upstream drivers. Besides that "MadWifi" thing
lurking in the svn repository there is also ath5k, ath9k, ar9170, and
more to come.

> Madwifi no longer suffers from your loudest original
> complaint- that it was not FOSS. You have to keep finding new reasons to
> fault it. I wonder why you bother. If your AthX work is so wonderful, you
> are wasting your time complaining about the Neanderthals, and you should
> save your energy for more productive things.

You know if other "MadWifi" people tell me the same I'd like to hear
it so we can clarify that on the front page. We had a similar debate
before and I expressed that upstream support required MadWifi project
to be forked where we wouldn't get complaints about doing what we do
think is best. We debated this internally and it was determined
upstream support is part of the MadWifi project goal and AFAICT that
was never removed so unless I'm missing something -- quit your
bitching, do your hacking and let me do my job on educating people
properly.

>> Devices now are supported upstream both for legacy and 802.11n
>> devices. Granted it'll be a nice surprise for those using shiny new
>> distributions but it seems there is a small group still completely out
>> of touch with these developments and hence I find the need to educate
>> further on what should be used for Atheros devices. We've done as much
>> as we can at wireless.kernel.org but for some reason that's not yet
>> the first place maybe some users look at for Atheros support.
>>
>>
>
> Madwifi has done a good job of supporting users-maybe that's why folks keep
> coming here for help.

No, the user I was helping was complaining about how MadWifi sucks and
how Atheros doesn't support 802.11n devices in LInux.

Mind you the user also complained about how he was sold an 802.11n
device with no 5 GHz and how this was a violation of the spec... or
something... but the rant just got to me.

>>
>> Anyone oppose to declaring death to MadWifi now on madwifi-project.org
>> ? Or at the very least explain clearly on the front page the turn of
>> events of the last year and what is recommended and for what people
>> actually *should* look at MadWifi for?
>>
>
> As a happy Madwifi user, I see no reason why Madwifi should belly up until
> it's dead, and to paraphrase Twain, the rumors of it's demise are premature.
> Folks are still working on the driver, it's distribution and support. If it
> is to die, it shouldn't be because someone employed by Atheros wants to
> murder it.

Lets leave my employer out of this, what I express here are my own
views on upstream support. There is no one whispering in my ear about
what to do or say, I'm doing what I think is best.

  Luis

Re: [madwifi-project] Death to MadWifi!

[Frans Pop]

Luis R. Rodriguez wrote:
> Trust me, I'm not concerned at all -- reason for me sending this
> e-mail was having to hear a rant from a user about MadWifi being such
> crap and about devices not being supported properly.

Let me give at least one counter example.

I have used madwifi in the past but switched to ath5k when it came 
available for my old Trust Atheros-based PCMCIA card, mostly because of 
the ease of compilation and partly because of the better "free-ness".

There is at least one area where the madwifi driver was and probably still 
is far superior: properly blinking the leds on the card to show the 
power/connection state and during data transfer.

It's not a major issue and though I've mentioned this at least once in a 
bug report or mail, this admittedly was back when there were more 
important things to fix. I probably should submit a proper bug report 
about it now.

I'm not asking for this to be fixed with this mail. I only mention it 
because the title and tone of this mail, and the fact of its recurring 
theme, offend me somewhat.
The way FOSS normally works is that one variant of alternative software 
will gradually die a natural death once a clear superior is established. 
There should be no need for members of rival projects to try to actively 
kill eachother off (the rival project that is, not its members).

Cheers,
FJP

P.S. This is the only post I've seen from this thread as apparently it 
started somewhere other than LKML.

Re: [madwifi-project] Death to MadWifi!

[Bob Copeland]

On Tue, May 12, 2009 at 11:14:21AM +0200, Frans Pop wrote:
> The way FOSS normally works is that one variant of alternative software 
> will gradually die a natural death once a clear superior is established. 
> There should be no need for members of rival projects to try to actively 
> kill eachother off (the rival project that is, not its members).

I agree, free software is about choice, and I really don't have a problem
with people using madwifi if it works for them, especially as it is now
open source (*not* the case when ath5k was born, and it would probably
still be closed source today without the work of Luis and Atheros, and
Jiri and Nick).

Of course, it reduces our pool of testers when people use madwifi, and I 
do have a vested interest in seeing ath5k succeed.  But ath5k is still
a young driver and maturing, so for now, madwifi may work better for some.
For me, ease of compilation was enough; I switched the day it was merged
into mainline even though it didn't work at all on my hardware at the time.
Now, it is quite stable here and I don't miss madwifi at all.

For what it is worth, I find it rather counter-productive when people on
irc suggest using ath5k to solve every madwifi problem, for example, for
AP mode, which clearly isn't enabled in even 2.6.30 and has bugs with the
TIM in the wireless-testing kernel[1].  That sort of thing just leads to
pissed off users.  Then said users go rant in bugzilla, claiming the kernel
developers are in league with the illuminati to steal their working setup
and rain misery down upon them[2].

Anyway, it is shameful to criticize Luis' intentions, who has done
nothing but work to get Atheros invested in the free software community;
I say that as someone with zero ties to Atheros other than having their
hardware.  Just look at the tree now, there are three fully open source
Atheros drivers, one of which was written entirely on Atheros' dime.
 
> P.S. This is the only post I've seen from this thread as apparently it 
> started somewhere other than LKML.

Same here, well, hopefully the thread will at least result in more
ath5k bug reports.

[1] I have a patch.
[2] We aren't (officially, at least).

-- 
Bob Copeland %% www.bobcopeland.com

Re: [madwifi-project] Death to MadWifi!

[Xavier Bestel]

On Tue, 2009-05-12 at 08:38 -0400, Bob Copeland wrote:
> For what it is worth, I find it rather counter-productive when people on
> irc suggest using ath5k to solve every madwifi problem, for example, for
> AP mode, which clearly isn't enabled in even 2.6.30 and has bugs with the
> TIM in the wireless-testing kernel[1].  That sort of thing just leads to
> pissed off users.

I'm one of them. Hours dealing with hostapd and ath5k, and then
realizing that AP mode just isn't there.
It'll be a while before I'll try it again, even if madwifi has its own
ways of pissing me off.

	Xav