The auto-completion layer seems not to have been designed/configured correctly.
So that, e.g., when using the julia layer, the `auto-completion/init-yasnippet`
function is called after the `yasnippet-snippets` package has been loaded.
Despite the ``yasnippet-snippets` package containing an `eval-after-load
'yasnippet`, the form within that block is, for some reason, evaluated already
before `auto-completion/init-yasnippet` has been called.
Anyway, what currently happens is that first `(yasnippet-snippets-initialize)`
is evaluated, adding `yasnippet-snippets-dir` to the `yas-snippet-dirs` list.
Then ``auto-completion/init-yasnippet` is called which contains the line `(setq
yas-snippet-dirs nil)`, setting the list to nil again.
As there is quite some logic involved (e.g. the order of evaluation is correct
when the julia layer is not being used), this commit fixes the reported issue by
simply calling `(yasnippet-snippets-initialize)` from
`auto-completion/init-yasnippet` after the yas-snippet-dirs list has been
resetted to nil.
Any cleaner solution is welcome, but I would say this is a 'harmless' very
pragmatic quick fix/solution.
* [git] Fix: Evil-collection @-dispatch keybindings for magit forge.
Load new location for forge '@' dispatch keybindings.
'Evil-collection' has moved keymaps for magit forge from an
'with-eval-after-load' block to newfile 'evil-collection-forge.el' in commit
33461aa545.
As mentioned in the commit, these changes were due to a upstream change in the
'forge' package. Source: https://github.com/emacs-evil/evil-collection/issues/543
* [git] Docs: update README to reflect changes in Forge keybinds.
See https://magit.vc/manual/forge.html#Introduction for more information.
* lsp: Bind =~SPC m g T~ to lsp-treemacs-type-hierarchy
This allows to show the type hierarchy for the type at point.
* lsp: Goto call and type hierarchy are provided by lsp-treemacs
* [compleseus] layout select current project first
* [compleseus] bind embark-export to C-c C-o
same as ivy-occur
* [compleseus] add vertico-{quick, directory}
* [compleseus] set up vertico-repeat
It's nice to have a way to use common SBT commands without having to enter a
Hydra, and without having to type them out or select from a history list. The
following should be standard and quite common:
+ `scalafmtAll`
+ `compile`
+ `test`
Added mnemonic keybindings that I think match Spacemacs conventions. They
behaviors also give better parity with the build capabilities of other major
modes, which frequently support compile, format and test functionality.
Currently most or all auto-completion, using completion-at-point (i.e. the TAB
key), is 'hijacked' by company-capf in emacs-lisp-mode, while in text mode there
is no auto-completion by default at all.
This commit adds auto-completion for text-mode and slightly improves the
situation for emacs-lisp mode.
From org version 9.5 org will not be distributed from Org ELPA but from
ELPA (see https://list.orgmode.org/87lfa7tc9v.fsf@gnu.org/t/). This commit makes
Spacemacs use/install org from ELPA.
Besides that, this commit removes some 'hack' that seemed to have no effect. As
org comes with Emacs `package-installed-p 'org` will always return t. Also
activating org via `configuration-layer//activate-package` seems to have no
effect.
Finally, the package is declared including a :min-version because appending only
`:location melpa` seems to have no effect (this can probably be considered a
bug).
I am almost certain that the information that was added when making org install
from Org ELPA is still relevant also for installing org from ELPA so I have just
adapted the version number from 0.104 to 0.300.
As the "org" archive has been removed, the `ert-deftest` for it can be removed
also.
* Add commands to compleseus+fasd layer (locate,yasnippet,search-from)
* Add metadata classifier (for embark) to fasd-layer
Embark uses metadata of the target under point to determine actions. If the
target lacks a metadata classifier then the [embark
documentation](https://github.com/oantolin/embark#acting-on-targets) recommends
to [add them using the marginalia package](https://github.com/minad/marginalia#adding-custom-annotators-or-classifiers).
This PR adds the file classifier to results for fasd searches.
The documentation of the `themes-megapack` layer states that the layer provides
all themes included in the [theme gallery](https://themegallery.robdor.com/) by
Rob Merrell. However, this theme currently is missing from the layer.
Special mode and its derived mode(s) buffers were made evilified in PRs #14995
and #15050. However, special-mode is too aggressive (as 'warned' for already in
#14995); it makes it hard to overwrite its keybindings (e.g. `evil-local-set-key`,
or using `evil-evilified-state-map` on a derived map have no effect).
Therefore it makes more sense to open the buffers in motion-state instead. This
state also does not hijack the `q` keybindings as defined by special-mode (see
https://www.gnu.org/software/emacs/manual/html_node/elisp/Basic-Major-Modes.html),
although it does still hijack its `g` keybinding, but evil navigation is
considered to have higher priority than revert-buffer anyway.
Related to #15117, this removes the helm-gitignore from the git layer because it
hasn't been updated to use the new git-modes package. The helm-gitignore
functionality isn't particularly useful anyway, it's just a utility to help
generate .gitignore files.
This change can be reverted if the upstream helm-gitignore package is fixed and
people find the functionality it provides to be useful.
Currently the python layer (without anaconda backend) provides only helm-pydoc
or pylookup-lookup-at-point functionality. Both these functions are cumbersome.
`pydoc-at-point-no-jedi` adds simpler doc functionality, therefore this commit
adds this command as a third option. It places it under `SPC m h p` because `SPC
m h h` is already taken by the anaconda doc functionality (it could be placed on
`h h` if anaconda is not available, but I did not bother to implement it)
