RE: Character folding in the pretest

[Drew Adams <drew.adams@oracle.com>]

> > Thinking first about defaults or DWIM is wrong, IMO.  We
> > should think first about how users can change the behavior,
> > including on the fly.
> 
> I don't agree. This leads to Emacs being painful to use without large
> amounts of customization. Do many Emacs devs use an empty or almost empty
> .emacs?
> Customizability is a strength, but the popularity of pre-packaged Emacs
> configurations (prelude, Emacs starter kit, Graphene, and countless .emacs.d
> repositories) says something about good defaults.

Please read what I wrote.  I do not argue that defaults are
unimportant, or that we should not choose good default behavior,
and choose it carefully.  Quite the contrary.

My point is that concentrating _first_ on the default behavior,
without considering various use cases, is a mistake.  (One reason
it is a mistake is precisely because without considering possible
use cases the default choice made is likely to not be the best one.)

I welcome the recent posts that point to different use cases.
The mere _possibility_ of char folding (treating different chars
equivalently, for some meanings of equivalence) means that there
can be, and so there will be, some very different needs and
preferences wrt which chars are to be handled as equivalent in
which contexts.  Better for us to start hearing about this at
the outset, so we have a wider vision of what this new feature
represents.

As to the popularity of starter kits:  Sure.  But the popularity of
_Emacs_ itself has a lot to do with its bendability - the fact that
different people can use it in different ways, and extend it or
customize it or change it on the fly to fit their needs.  Without
that, Emacs is not Emacs.

And in the case at hand, I feel that char folding does not yet
provide enough flexibility for users.  It provides a useful set
of foldings (equivalences) out of the box, and that's great, as a
start.  But we should make it more user-customizable.
Just one opinion.

It's not a case of one or the other: picking good defaults and
clever DWIM or providing ways for users to control the behavior.