Re: [ELPA] New package: transient

["João Távora" <joaotavora@gmail.com>]

On Sun, May 3, 2020 at 5:25 PM Dmitry Gutov <dgutov@yandex.ru> wrote:
>
> On 03.05.2020 17:45, Eli Zaretskii wrote:
> > Years of using Emacs with its superb documentation and elaborate
> > facilities to find stuff in that documentation caused me to be more
> > demanding, to expect better documentation that is easier to access
> > quickly.  I might compromise if something like that is not available
> > when I'm working with other software, but I certainly don't want to
> > give that up when I work on Emacs!  And I'm astonished to hear that
> > people don't_want_  the documentation we provide and the help commands
> > to go with it, and are willing to settle for API completion!
>
> The main reason the discussion even went that way is because instead of
> acknowledging that the user's scenario is valid and the request is
> reasonable (that code completion and describe-function's completion will
> work easier and faster if function names are more predictable), you
> responded with the recommendations to "just use manual".
>
> Whereas the manual provides a different workflow and doesn't cover all
> cases. For instance, it only covers the functions in the core. Maybe not
> even all of them. Whereas the aforementioned features work uniformly for
> all functions, no matter where defined or by whom.

Yes, but isn't a step in reconciling the two workflows that the
aforementioned features be extracted from the manual?
Instead of going the hardcore, full frontal, bike shedding,
never-gonna-happen, "rename all the symbols" route?  I
mean I like those quick references too, but they're built,
probably automatically, for languages with different ways
of organizing code, that aren't available in Elisp right now.
That doesn't mean we can't have something that approaches
their usefulness without raping Lisp in the process.

Here are a few practical suggestions:

1. Make C-h O lookup a symbol's definition in the Elisp manual.
Is this very hard to do? Seems like it would make a nice parallel
to C-h F.

2. Make a Texinfo macro or something like that (completely ignorant here),
that extracts the listing of functions and or symbols under a given node.
This is the keep-lines example of Phillipe.

3. Make every function C-h f'ed whose provenance in the Elisp
manual we can automatically recognize have a link to the manual.

4. Make elisp--xref-backend collect references in the manual
(because why not?)

5. Make a "See also" section in the *Help* buffer.  Lars'
suggestion

João