Re: Imports / inclusion of s.el into Emacs

[Dmitry Gutov]

On 04.05.2020 21:11, Eli Zaretskii wrote:

> > Of many discussion about changing default behavior that I've seen, the
> > vast majority have failed.
>
> Then you only remember the ones that led to nothing because we didn't
> do anything but talk.

Others, where some change did arrive, contained a lot of talk as well.

> > > A recent example: the :extend
> > > face attribute.
> >
> > It's an interesting counter-example. First, it did little to change the
> > actual functionality, just the looks.
>
> The reaction to this "little change" could have fooled me.

In your opinion at the time it was little, IIRC. One of the reasons it 
got in, I guess.

> > Second, there were no noticeable consensus, inside or outside the
> > core, that the new behavior is better
>
> Not true.  There was complete consensus among those who discussed the
> feature before it went into implementation.

Among the 2-3 people who participated in the highly technical 
bug-report/discussion about the display engine? Please be serious. That 
doesn't reflect the opinions even across emacs-devel, much less the 
community at large.

And once we found out the backward compatibility problem, and all the 
associated details, there was still one solution available: revert. 
That's what we usually do when we don't manage to fix a regression 
before a release, don't we?

> > (I've mostly seen dissenting feedback, but the split is probably more
> > like 50/50). But apparently you liked it well enough because it made
> > Emacs's behavior more compatible with other software _you_ were familiar
> > with, that even breaking the expectations of a lot of our users, or
> > having to force all theme authors to update their themes (until I came
> > along with a fix) wasn't price too high.
>
> So what is the lesson you suggest to take out of this example, in the
> context of "adapting"?

Maybe that backward compatibility is not as important as some people 
like to claim? Other lessons would be less kind to type out.

In any case, I don't see that particular change affecting the experience 
of new Emacs users much one way or another (as long as all the themes 
work). Thus it's not a great example of Emacs "adapting" to contemporary 
user expectations.