About madwifi-modules-* dependency on madwifi-tool

[Pascal Dormeau]

Kel Modderman a écrit :
> For a very simple mode change wlanconfig is required, unless you feel 
> modprobe -r, modprobe ath_pci autocreate=<mode> is an acceptable alternative.
> 
> That is enough to justify the choice to make madwifi-tools a hard dependency 
> for me.

I must admit I did not see it from that angle, it is indeed enough to 
justify the hard dependency.

It just happens that I make so basic usage of the modules that I never 
need one of the madwifi tools, and I made the parallel with the wireless 
tools (I sometimes use the latter, but only to check that my ap is 
reachable: iwconfig ath0 and iwlist ath0 scan).

> 
> Just because madwifi defaults to one mode of operation, and provides a 
> modprobe time option to manipualte that default is not enough to convince me 
> that wlanconfig shouldn't be shipped along with the "weird" modules _always_.
> 
OK, I see, modification brought by r1407 as to be understood as an help 
to minimize wlanconfig handlings for its most basic usage but not to get 
rid of it definitively.

> And madwifi modules _are_ weird, they use a non-standard method for such a 
> basic operation such as mode changing and totally ignore the standard that 
> already exists (which every other linux wifi module uses). Upstream support 
> and documentation refer to wlanconfig usage every time a mode change question 
> is asked, and its one of the most frequently asked support questions.
> 
In other words, the modules and the tools need to be regarded as a whole 
(they are actually shipped in the same tarball by upstream). It happens 
that they are distributed in two different packages, then only the hard 
dependency can make it appear as whole. Right?

> Thats my 2c.
> 
> Thanks, Kel.
Thank you Kel for answering so quickly. This seems pretty clear to me 
now. I won't bother you anymore with this dependency question.

Best regards,

Pascal