Re: Changes for emacs 28

[Ergus <spacibba@aol.com>]

On Wed, Sep 09, 2020 at 12:05:54PM -0400, Stefan Monnier wrote:
>Thanks, TEC, I found it quite useful.  Further comments and questions below.
>
>> * Org-mode, my ramp into Emacs pt.1 - vanilla
>> - I'd heard about Org-mode as a thing markdown + execution
>> - Maybe this could be better than Jupyter?
>> - However... I didn't like Spacemacs
>> - Removed Spacemacs
>> - Typed 'emacs' into a terminal
>
>So far so good.
>
>> - Oh, this looks old.
>
>Fair enough.  I don't think we (Emacs community) are in a position to
>make it look "modern" and sexy.  I know I'm not because my notion of
>"modern and sexy" is quite outmoded ;-)
>
>But "looks old" is usually not a deal breaker, just a negative
>first impression.
>
Actually spacemacs made it to look a bit more modern by just changing
some colors.

>> - Successfully opened a file
>>   + Where are the line numbers?
>
>Interesting.  It would never occur to me to expect line numbers in
>a text editor.  When and why did line numbers become fashionable?
>[ My guess is something like "ever since shortscreens became the only
>  option, creating a void in the horizontal space that needed
>  filling" ;-) ]
>
>I don't oppose enabling line numbers by default, but I do find line
>numbers to be an awful waste of valuable screen real estate.
>
I use line numbers, but instead of enabling them by default we could
make it very very accessible in the toolbar or initial config dialog.

>>   + Why aren't I given much information on the file
>
>Could you be more specific in terms of the particular information that
>you felt Emacs failed to give (and maybe how you expected it to be given)?
>
>>   + Where's the completion, the linting, etc.
>
>Do I understand you right that you expected company+eglot+flymake to be
>enabled (and configured) by default?
>
>I personally find this to be the most glaring concrete problem in
>Emacs nowadays.
>
Again we just don't need to enable them by default, but just make them
easier to config. I still find myself in the dichotomy between company
or auto-complete modes after 6 years in emacs.

flymake is good enough, but not very easy to find on the beginning, and
it still requires a lot of work. Also most of the documentation around
recommends using flycheck and actually for me it works much better in
most of the situations and languages I use.

>> - Tried to execute a command interactively (forget which command)
>>   + Typed M-x
>>   + Wait, this is just a text box
>>   + I don't know commands off by heart!
>>   + I want to be able to type key terms and see options!
>
>How much of this would be satisfied by icomplete-mode together with the
>`substring` completion-style (which would be a smaller change to
>the UI than something like helm-M-x or counsel-M-x).
>
Yes please.. If you think that icomplete/fido-mode can be improved
somehow, just make specific requests.

>> - Having an initialisation† file, well commented such that *without knowing
>>   anything about Emacs* I could have Emacs be set up such that I could
>>   actually try it with familiar tasks and not be underwhelmed, or have
>>   to deal with sudden troubleshooting
>
>Maybe we could have a "default init file" (consisting of nothing but
>commented out code snippets, accompanied by actual comments explaining
>them)?
>
This is why I have been asking for use-package integration. As a
starting point this is the easier we can provide to starters to
find/know about the different options and modes.

>>   †The important bit about this file is that it let me declare which
>>   bundles of functionality I want easily, and without having to parse
>>   much unfamiliar lisp (both Spacemacs and Vanilla fail in this regard,
>>   but in different ways).
>
>Hmm... a "default init file" would still use "unfamiliar Lisp", I'm afraid.
>
with use-packages the lisp knowledge needed to configure is
minimal. Just to understand (this detail)

>> - Having good 'discoverability enhancements' used by default
>>   - counsel for M-x
>
>IIUC this is similar to enabling icomplete-vertical and
>icomplete-show-matches-on-no-input, and maybe using a regexp
>completion style?
>
There is not icomplete-vertical yet ;).

Any way icomplete is too far to compete with counsel/ivy/swiper in
functionality and termination level.  (at the moment icomplete is Donald
Duck and counsel is James Bond).

I don't ask to add counsel to vanilla because it requires many external
optional dependencies that I would never like to see in vanilla and
retards the startup time.

>> - Used Discord for it's community, a recent chat-app which I recognised
>>   (I'm still warming up to mailing lists).
>
>Definitely a non-starter since it's proprietary.
>There are obviously acceptable alternatives.
>
>I think an important aspect is to find a communication medium that can
>be used from Emacs.  IRC and Email do satisfy this criteria.
>Whenever I have to write text outside Emacs I feel hampered.
>
>
Emacs has even a telegram client, jabber client, and IRC... so of course
any method we use to communicate must have an emacs client, a web or
desktop client and a mobile client.



>        Stefan
>
>