Re: [oe-commits] [openembedded-core] branch thud updated (ad0a553 -> 748f946)

[Adrian Bunk <bunk@stusta.de>]

On Sun, Mar 17, 2019 at 08:38:20AM -0700, akuster808 wrote:
> 
> 
> On 3/17/19 6:08 AM, Adrian Bunk wrote:
> > On Sat, Mar 16, 2019 at 08:50:13AM -0700, akuster808 wrote:
> >> On 3/16/19 5:20 AM, Andreas Müller wrote:
> >> ...
> >>> 2. This was applied on Feb 6th which is not 3 month back exactly.
> >> then its worst than I thought, I can't remember what my thought process
> >> was back to Feb 6th.
> >>> [1] https://wiki.yoctoproject.org/wiki/Stable_branch_maintenance#Point_release
> >> I am glad you are bringing these things up, it helps me revisit my own
> >> processes and help me improve.
> >>
> >> We are planning on revising the maintenance guidelines soon so I hope to
> >> get your input.
> > Was the boost upgrade in thud sent to the mailing list for review?
> That series did not, the previous ones and several Sumo request have.
> 
> So you are the second person to mention the update, is it causing a
> problem?

You were requesting input.

I am not using boost on Yocto, but in Debian it is pretty normal
that several packages stop building each time boost gets updated.

"Drop signals library as upstream has removed it" in the backported
commit shows the tip of this iceberg.

What went wrong that even the removal of a library from boost
did not prevent this change from entering a stable branch?

> > My reading of the "Requesting a fix in a stable branch" section 
> > would be that this is already a mandatory part of the process.
> 
> That is not under the "Maintainers procedure" so it does not apply.
> 
> I have taken requests via IRC and a simple "please add this to stable
> branch X" emails so I have not been enforcing the letter of the law. 
>...
> Like I have mentioned already, the processes mentioned in
> "Stable_branch_maintenance" are under review.

One problem is that changes to master are getting better reviewed than 
changes to stable branches.

Upgrading boost in a stable branch wouldn't have survived a mailing
list review.

A possible improvement would be to always use thud-next, and each time 
commits are added to thud-next an email thread with all new commits gets 
sent to the mailing list (similar to the review threads for new upstream 
stable kernels, see [1] for an example).

> regards,
> Armin

cu
Adrian

[1] https://lkml.org/lkml/2019/3/12/1290

-- 

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed