Re: Missing changes in merges from emacs-25 to master

[Eli Zaretskii]

> From: John Wiegley <address@hidden>
> Cc: address@hidden (Phillip Lord),  address@hidden,  address@hidden,  
> address@hidden
> Date: Tue, 22 Mar 2016 12:21:10 -0700
> 
> >>>>> Eli Zaretskii <address@hidden> writes:
> 
> > It's not your fault. Quite a few others (myself included) produce these
> > merge-commits from time to time, this cannot be avoided. The problem is in
> > the merge procedure from emacs-25 to master -- it shouldn't skip such
> > merge-commits.
> 
> One take away from this is to be very cautious in assessing whether merge
> commits should really be skipped or not.  What is the criteria for knowing
> when they *should* be skipped?  Checking to ensure that all those commits are
> already on master?

That would be a lot of work, IMO.

I actually don't understand why we should skip _any_ merge-commits.
Can you (or someone else) describe a situation where merging such
commits to master wouldn't be TRT?