Re: Should 9aafabc7fece be reverted?

["Andrew F. Davis" <afd@ti.com>]

On 05/19/2016 04:17 AM, Pali Rohár wrote:
> On Thursday 19 May 2016 09:44:04 Ivaylo Dimitrov wrote:
>> Hi,
>>
>> On 18.05.2016 19:58, Andrew F. Davis wrote:
>>> Hello all,
>>>
>>> Why do we need this patch? I removed giving each battery a number in its
>>> name intentionally. It doesn't seem to be needed anymore, and only two
>>> old drivers out of the ~70 power supply drivers do this.
>>>
>>
>> The patch is needed because commit <703df6c09795> breaks existing userspace
>> on Nokia N900, which counts on name numbering. Also, commit <703df6c09795>
>> description says nothing about name changing, so I assumed it is an
>> unintentional side-effect of splitting the code into modules.
> 
> Right, changing name cause breaking userspace and that commit does not
> describe anything about changing name. So also from my perspective it is
> regression.
> 

Fair enough, I was unaware of the userspace breakage, the patch message
only stated it was being put back because it used to exist.

Thanks for the clarification,
Andrew

P.S. The code is still a bit strange, I'll probably go grab one of the
N900s from our test farm and make sure my future cleanups don't break
this, but are we sure the *name* of a driver is an ABI? Sound like ABI
abuse by the N900 software to me, almost like telling Linus he can't
change the kernel version number because a userspace script depends on
"uname == v2.6.x". :)