Protesilaos Stavrou

🏆 I provide private lessons on Emacs, Linux, and Life in general: https://protesilaos.com/coach/. Lessons continue throughout the year.

Emacs: restart MCT development

I just pushed this to the Git repository of mct.el:

commit ea1edfff5e394b2609b11a52ab2fe7659842b310
Merge: 3f15ff6 f52ee53
Author: Protesilaos Stavrou <info@protesilaos.com>
Date:   Sat Mar 25 21:02:43 2023 +0200

    Merge branch 'restart-development'

    [...]

 Makefile   |  24 ---
 README.md  |  19 ++
 README.org | 490 +++++++-------------------------------------------
 mct-avy.el | 100 -----------
 mct-tcm.el | 121 -------------
 mct.el     | 589 +++++++++++++------------------------------------------------
 6 files changed, 202 insertions(+), 1141 deletions(-)

Below is the complete commit message.

Merge branch ‘restart-development’

I am restarting the development of MCT with the sole intent to keep it in good working condition for the handful of users who like it. I do not plan to expand the scope of the package. Instead, I am removing certain features of dubious value and am cleaning up the code (which was in a good state, all things considered).

To recap:

  • MCT is a layer of interactivity on top of the default Emacs minibuffer and the Completions buffer. It treats the two as a unified space and provides commands to intuitively move between them.

  • MCT provides a facility for “live completions” which automatically update the Completions buffer given certain conditions. Users have access to options for a passlist and blocklist which further control what should be live updated.

  • On 2022-04-14 I had announced the discontinuation of the project’s development: https://protesilaos.com/codelog/2022-04-14-emacs-discontinue-mct/. The reason was that Emacs 29 was assuming certain features that MCT had. I thought that Emacs 29 would provide an MCT-equivalent experience and was making way for that eventuality. It turns out that I was mistaken: MCT is still more featureful and might show the direction for future developments on Emacs 30.

Now the other news:

  • I still think vertico by Daniel Mendler is the best User Interface for the minibuffer. Where relevant in my Emacs coaching sessions, I always recommend Vertico: https://protesilaos.com/coach. It is robust and highly configurable. What MCT does, Vertico does better. My plan for my personal config is to have two modules, one for Vertico and another for MCT, so that I can use the former by default and the latter when needed. (Again, I want to maintain MCT but will not add major new features.)

  • I have removed support for Emacs 27. This was a mistake from day one. Emacs 27 cannot show the Completions buffer in one-column view and thus lacks the primary user-facing aspect of MCT.

  • I have removed everything that has to do with in-buffer completion (completion-in-region). While it is nice to have a uniform interface for completions in the minibuffer and inside buffers, the latter was never good enough. This is not an MCT problem, but how inherently limited the Completions are in the scenario where the minibuffer is not activated. I will continue using corfu by Daniel Mendler.

  • For a brief period of time I experimented with various extensions to the core MCT functionality (e.g. integrating with avy.el. These are all removed. There now is a single ‘mct.el’ file. It works fine for what it was originally designed to do: enhance the minibuffer UI.

That was it. Further information:

Next:
Previous:
See all: Code blog entries