RE: Showing all sequences bound to a prefix?

[Drew Adams]

> > The only reason `C-s C-h' does not work (show you the 
> > isearch bindings) is because RMS did not want it to work.
> > He prefers that `C-h' break out of isearch
> > (and so initiate global help).
> 
> `C-s C-h' does not work the way e.g. `C-x r C-h' works 
> because `C-s' is not a prefix key.

The discussion wasn't only about working "the way C-x r C-h works" - the goal
was to see the isearch key bindings. Which is why I explicitly said "does not
work (show you the isearch bindings)".

The "does not work" in this general sense of showing you the isearch bindings is
not "because C-s is not a prefix key". That explains why `C-s' doesn't work the
same as `C-x r', but it doesn't explain why `C-s C-h' doesn't show you isearch
help (bindings).

There is nothing about not being a prefix key that prevents having C-s C-h show
the isearch bindings. As has been said, it can easily be made to work (show the
bindings), by just binding C-h to isearch-mode-help in isearch-mode-map. It's
not a technical problem.

The *reason* that that binding has not been made in Emacs (it has been proposed
several times), and thus the reason that C-s C-h does not work (does not show
you the isearch bindings), is because Richard does not want C-h during isearch
to offer isearch help. He wants it to exit isearch and offer global, top-level
help. It is not a technical reason, but a UI choice.

> `C-s C-h' does not work the way you and I want it to work because
> `C-s' invokes a transient mode (incremental search) with its 
> own keymap, which implies that `C-h' must be bound to
> mode-specific command in that keymap,

Yes.

> and RMS prefers a different binding than we do.

Yes.

But there is no relation between the last clause and the first part of the
sentence. The first part explains why C-s C-h doesn't _automatically_ work (the
way C-x r C-h does). But all that matters for the desired result is the second
part: it wasn't implemented because of a UI design choice - the implementation
is then irrelevant.