* feat(vertico): completion highlights a la ivy
Adds completion highlighting that works similarly to ivy/counsel's
one (which is enabled by default). It'll highlight enabled major/minor
modes and directories in a different face. On by default.
Ref: https://github.com/minad/vertico/wiki#candidate-display-transformations-custom-candidate-highlighting
* fix(vertico): major mode not being highlighted
The major mode was not being highlighted correctly;
it should work now that the buffer is set correctly in
`+vertico-highlight-enabled-mode`.
* fix(vertico): make font lock prioritize match over type
The mode and directory highlights were [overriding the match
font-lock](https://github.com/doomemacs/doomemacs/pull/7706#issuecomment-1977722188).
This should resolve that by prioritizing the match font lock using
`'append` on `add-face-text-property` instead of `propertize`.
Previously, a distinction was made only between "regular" style (assumed
to be for people who prefer to cycle directionally and commit with RET)
and +tng (assumed to be for people who prefer TAB and auto-commit). This
made composing further variations harder on the user, as they would have
to work around our bindings. Now we have many features as documentation,
and the user can compose to their liking. The global CAPF list was
pruned, suggesting to use individual keys for those functions if
required.
Currently, `SPC b k` does not allow us to kill popup buffers. If we are
say using a comint/eshell/eat buffer and we borked the state while
testing hooks or we sent EOF to the underlying process, our intention
with `SPC b k` is to kill the underlying buffer.
We should always bind to both "TAB" and "<tab>" - if they have been
mapped to different commands anywhere, Emacs will no longer treat them
as equivalent.
Relevant portion of the dependency graph:
'oc-csl → 'citeproc → 'citeproc-itemgetters → 'org → 'org-keys → 'oc
The :after linked 'oc to 'oc-csl so 'org-keys could load 'org.
`+org-init-keybinds-h' could then execute before the body of 'org-keys,
so Org default bindings could clobber Doom bindings.
If the top-level 'org started the load sequence, `org-load-hook'
functions would execute last and key bindings would be correct, but
Doom’s deferred incremental loading of Org could lead to this scenario:
1. deferred incremental loader: (require 'org-src)
2. org-src.el: (require 'org-keys)
3. org-keys.el: (require 'oc)
4. oc.el: (provide 'oc): (require 'oc-csl) [via the :after this patch
removes]
5. oc-csl.el: (require 'citeproc)
6. citeproc.el: (require 'citeproc-itemgetters)
7. citeproc-itemgetters.el: (require 'org)
8. org.el: (require 'org-keys)
9. 'org-keys from step 8 finishes loading. Keys are bound.
10. 'org finishes loading. `org-load-hook' functions are run. Doom’s
keys are bound.
11. 'oc-csl finishes loading
12. 'org-keys from step 3 finishes loading. Keys are bound again,
overwriting any keys Doom bound in step 10.
13. 'org-src finishes loading
At some point, the deferred incremental loader will (require 'org), but
it is a feature as of step 10, so it does not load and its hooks do not
run again.
Recently, org-mode made a non-standard tab-width an error state.
Unfortunately, it's way too easy for users to accidentally change
it (e.g. with editorconfig, indiscriminate setter hooks, or other
packages), and since there is zero reason to ever want a non-standard
tab-width in org-mode, I not only exclude it from dtrt-indent as a
global default (i.e. not only as part of :lang org), but I add a late
org-mode-hook hook to reset tab-width, just in case.
Previously, a distinction was made only between "regular" style (assumed
to be for people who prefer to cycle directionally and commit with RET)
and +tng (assumed to be for people who prefer TAB and auto-commit). This
made composing further variations harder on the user, as they would have
to work around our bindings. Now we many features as documentation, and
the user can compose to their liking. The global CAPF list was pruned,
suggesting to use individual keys for those functions if required.
Previously, a distinction was made only between "regular" style (assumed
to be for people who prefer to cycle directionally and commit with RET)
and +tng (assumed to be for people who prefer TAB and auto-commit). This
made composing further variations harder on the user, as they would have
to work around our bindings. Now we many features as documentation, and
the user can compose to their liking. The global CAPF list was pruned,
suggesting to use individual keys for those functions if required.
Previously, a distinction was made only between "regular" style (assumed
to be for people who prefer to cycle directionally and commit with RET)
and +tng (assumed to be for people who prefer TAB and auto-commit). This
made composing further variations harder on the user, as they would have
to work around our bindings. Now we provide a flag per feature, and the
user can compose to their liking. Some variable options were turned into
flags and the global CAPF list was pruned, suggesting to use individual
keys for those functions if required.
If we are using +tng, we can make RET always exit the minibuffer. This
has the downside of not being able to insert snippets, but no one writes
snippets for the minibuffer anyway.
We previously implemented only consult/vertico as a target for export,
now we have all of them. It was necessary to use case-by-case
conditions, unfortunately, because other UIs have subtle quirks that
prevent a single generalized approach to work.
Ivy is almost compliant, but it needs beg and end to not be markers.
Helm doesn't replace `completion-in-region-function`, it expects to go
around the default `completion--in-region`, so a small addition was made
to its module, because we weren't doing that. This was likely an
oversight due to the non-standard usage. This was fixed here because we
need it working for this feature.
Ido doesn't implement `completion-in-region` and its `completing-read`
is retricted to a list of strings as table, so it's treated the same as
absence of a framework, because it lacks the needed features.
An issue when using corfu in the minibuffer was the need for pressing
RET twice, since the first only inserts the completion. This commit
aliviates that by providing C-RET to ignore completion and conclude the
minibuffer imediately and S-RET to insert completion then conclude.
