Re: 2.6.25 intended change: smc:kbd_backlight vs. smc::kbd_backlight ?

[Soeren Sonnenburg <kernel@nn7.de>]

On Sat, 2008-04-05 at 11:42 +0100, Richard Purdie wrote:
> On Sat, 2008-04-05 at 07:32 +0200, Soeren Sonnenburg wrote:
> > On Fri, 2008-04-04 at 15:12 +0000, Justin Mattock wrote:
> > > Hello; I noticed this as well, you loose the greatest feature; the
> > > cool leds lights on the keyboard with the macbook pro,
> > > A solution that worked for me was to compile pommed from the source
> > > and edit kbd_backlight.h
> > > regards;
> > >                Justin P. Mattock
> > 
> > Yeah, but it will not only affect pommed... and I don't see the
> > rationale of introducing :: to separate things... so I think we should
> > have another patch replacing the :: with ':' again to unbreak userspace
> > for 2.6.25.
> > 
> > Could someone more official please respond to this regression (CCing
> > Rafael to record it)
> 
> Basically the LED names were a mess, there is a documented standard
> which has been there since the beginning (Documentation/leds-class.txt)
> and a number of drivers didn't follow that. That standardisation commit
> was an attempt to deal with this problem once and for all and in that
> sense its a bugfix, not a regression. The fact function wasn't part of
> the original naming scheme was a mistake (see previous discussions on
> LKML) which we're trying to address without breaking the documented
> standard. The side effect is that some existing drivers need to be
> updated to match the documentation (and should never have been merged in
> the first place because of that) :(.

So to make it clear "devicename:colour:function" means in the case of
applesmc smc::kbd_backlight because there is no colour defined. One
could have used "devicename:function:colour" without breaking things -
no ?

So it seems we will have to file bug reports to the led dealing
userspace programs ...

Soeren