Re: [GIT PULL] RESEND: thermal for v5.9-rc1

[Daniel Lezcano]

On 07/08/2020 17:54, Linus Torvalds wrote:
> On Fri, Aug 7, 2020 at 2:40 AM Daniel Lezcano <daniel.lezcano@xxxxxxxxxx> wrote:
>>
>> It defaults to 'y' because the previous (but unused) implementation was
>> unconditionally compiled-in and because of the thermal users needs.
>>
>> Is default=y wrong given this history?
> 
> One million percent wrong.
> 
> The fact that the old implementation was never used just shows that
> it's not so important, and it shouldn't be default 'y'. Not having it
> doesn't break anything.
> 
> And the new implementation presumably isn't even compatible with the
> old format also means that it shouldn't be default 'y'. Building it in
> wouldn't help anyway.
> 
> And the fact that _some_ users might want it does not mean that it
> should be default 'y', because those users presumably _know_ they want
> it.
> 
> IOW, defaulting to 'y' is just wrong in every possible way. This is
> not some kind of "to maintain compatibility and not break existing
> users we should enable this" kind of thing.
> 
> And it's not some kind of "everybody should have it" thing either,
> since presumably nobody has the user-space support for it anyway.
> 
> It's something that a new distro would enable _if_ they actually end
> up supporting the user space. Not something the kernel should enable
> "just because".
> 
> Really: "default y" is _wrong_. Every developer thinks that _their_
> code is so magical and special that everybody should run it.
> 
> And every developer is almost always wrong. Unless you have a "not
> having this will break existing users", you don't do it.

Ok, I will send a fix.

Thanks for taking the time to clarify that.

  -- Daniel


-- 
<http://www.linaro.org/> Linaro.org │ Open source software for ARM SoCs

Follow Linaro:  <http://www.facebook.com/pages/Linaro> Facebook |
<http://twitter.com/#!/linaroorg> Twitter |
<http://www.linaro.org/linaro-blog/> Blog