Published 2024-09-05
tag(s): #emacs #programming
A few days ago I posted about using default minibuffer completion in Emacs.
My hypothesis was that rather than being annoyed at the extra typing, I could adjust the way I interact with Emacs to "think & recall" before invoking the command to switch buffer, open a file, etc.
I can report that this has been quite successful, and produced the change I expected. I am now more aware not only of the naming conventions of things at work, but even for my own packages and custom code, I am revisiting the naming of some commands to make it more consistent.
Instead of TAB-ing my way around completion (or more accurately, C-i...ing?), which is why I used to do, I now type sometimes complete commands or filenames, then complete for the full path or variants of names.
This is my current minibuffer setup:
(use-package minibuffer
:demand t
:custom
(completions-format 'one-column)
(completions-max-height 20)
(completion-styles '(flex))
(read-buffer-completion-ignore-case t)
(read-file-name-completion-ignore-case t)
(completion-ignore-case t)
(completions-detailed t)
(completion-auto-help 'visible)
(completion-auto-select 'second-tab)
:bind
(:map minibuffer-mode-map
("C-n" . minibuffer-next-completion)
("C-p" . minibuffer-previous-completion))
(:map completion-in-region-mode-map
("C-n" . minibuffer-next-completion)
("C-p" . minibuffer-previous-completion))
(:map hoagie-keymap
("" . execute-extended-command)))
There's a variant of completion-style that I want to test, which
is (completion-styles '(substring partial-completion flex)) instead of just flex.
I really like this change. I don't discard going back to fido, icomplete, or something else in the future, but right now this just feel right. 😎