Re: [PATCH v1 3/4] docs: stable-kernel-rules: call mainline by its name and change example

[Greg Kroah-Hartman]

On Thu, Apr 11, 2024 at 08:50:19AM +0200, Thorsten Leemhuis wrote:
> On 11.04.24 08:10, Greg Kroah-Hartman wrote:
> > On Thu, Apr 11, 2024 at 07:50:29AM +0200, Thorsten Leemhuis wrote:
> >> On 11.04.24 07:30, Greg Kroah-Hartman wrote:
> >>> On Thu, Apr 11, 2024 at 07:25:05AM +0200, Thorsten Leemhuis wrote:
> >>>>  
> >>>> -     Cc: <stable@xxxxxxxxxxxxxxx> # after 4 weeks in mainline
> >>>> +     Cc: <stable@xxxxxxxxxxxxxxx> # after 6 weeks in a stable mainline release
> >>>
> >>> I do not know what "stable mainline release" means here, sorry.  "after
> >>> 4 weeks in mainline" means "after in Linus's tree for 4 weeks, but
> >>> Linus's tree is not "stable mainline".
> >>
> >> I meant a proper mainline release like 6.7 or 6.8 to make it obvious
> >> that this does not mean a "pre-release".
> >>
> >> I actually had used the term "proper mainline release" earlier in a
> >> draft, but a quick search on the net showed that this is not really used
> >> out there. "stable mainline release" is not popular either, but seemed
> >> to be a better match; I also considered "final mainline release", but
> >> that felt odd.
> >>
> >> It feels like there must be some better term my mind just stumbles to
> >> come up with. Please help. :-D
> > 
> > Well, what is the goal here?  Just put it in words, I have seen stuff
> > like:
> > 	Cc: <stable@xxxxxxxxxxxxxxx> # wait until -rc3
> > 	Cc: <stable@xxxxxxxxxxxxxxx> # wait until 6.1 is released
> > 	Cc: <stable@xxxxxxxxxxxxxxx> # after -rc2
> > 
> > and so on.
> > 
> > Just pick a specific time/release might be better?  "after X weeks" is
> > assuming that we all know and remember how many weeks something
> > happened...
> 
> My reasoning was: a developer that submits a patch has no full control
> over when the patch mainlined -- and plans sometimes change, too.
> 
> So a patch that was meant to go into 6.1-rc with a tag like "# wait
> until 4 weeks after 6.1 is released" might only be mainlined for 6.2-rc1
> -- and then the tag does not express the developers intention.

I've normally seen patches end up in Linus's tree "too early" more often
(i.e. cc: stable for stuff that has never been in a stable tree yet),
but sure, I can see how changes can also take too long.

> But that might be a corner case that we could ignore. So maybe "# wait
> until 4 weeks after 6.1 is released" is the better example (from what
> I've heard something like that is what developer would like to have
> sometimes).

Yes, referencing off of a fixed point like a release is best as that's
much easier for humans to calculate.

Also because, the original "after 4 weeks", doesn't give me a reference
point to judge what the starting time is easily.  Yes, I have tools for
that, but most people don't.

So how about changing it to use the "fixed point" reference please?  The
phrasing "after -rc3" is probably what most people almost always want
anyway, given the huge churn that -rc1 is.

thanks,

greg k-h