problems:
the delete file prompt doesn't show which file it's about to delete,
and when the deletion request is canceled, then the question remains in the
minibuffer (the text disappears when an action updates the minibuffer, like
moving the cursor, but it might be confusing to still be seeing the question
after it's been answered)
solution:
show the file name in the delete file prompt, and show a canceled message
This commit reverts ab34cc2d9d. Setting
company-minimum-prefix-length to 0 causes issues indicated in issue #8222. While
the user is free to set company-minimum-prefix-length to 0 if they please, doing
so affects editing any files that also use web-mode such as jsx files.
Additionally, new users to Spacemacs may be confused by this behavior since no
other layer sets company-minimum-prefix-length to 0.
By analogy with the "zz", "zb", "zt" vim / evil bindings, this adds an
additional key binding "z" for `recenter-top-bottom` to the following
transient states which navigate around the buffer in large jumps:
- auto-symbol-highlight
- error
- buffer
- vcs
- org-babel
This allows for repositioning of the buffer for visibility without having to
exit the transient state.
Minor updates are also made to documentation of other transient states.
c-c++-enable-rtags-support has been removed in favor of c-c++-backend. Don't mention it in README
Delete .ccls from c-c++-adopt-subprojects
as in ccls, .ccls-root is recommended for this purpose
Delete the extra parameter :levels 3 from $ccls/inheritance
Finding the immediate bases is usually more desired
The parameters of refs-* in ccls are not in :context
In json-mode tell prettier the content is always to be parsed as json.
Fixes formatting in orgmode `#+BEGIN_SRC json` blocks as prettier
otherwise expects content as javascript.
Fixes: #11343
<<Amendment 1 21/09/18>>
Added functions to facilitate keybindings in derived layers consistent
with navigation style preference.
<<Amendment 2 24/09/18>>
Incorporated feedback from yyoncho
Corrected error in peek binding extension definition
<<Amendment 3 24/09/18>>
Moved 'heirarchy' prefix/keybindings from 'SPC m h' to 'SPC m <g/p> h' for
consistency with navigation style preference selections.
Added 'members' prefix under 'SPC m <g/p> m'
Moved 'lsp-ui-imenu' from 'SPC m g m' to 'SPC m g M'
<<Amendment 4 20/10/18>>
Rebased. Updated README formatting for consistency with norms.
<<Amendment 5 25/10/18>>
Rebased. Moved 'peek' keybindings from 'SPC m p' to 'SPC m G'
<<Amendment 6 01/11/18>>
Reased. Updated for compatibility with upstream changes in lsp-ui.
<<Amendment 7 02/11/18>>
Corrected oversight -- was still binding some peek funcs under 'p'
rather than 'G'. Moved backend keybindings to 'b' (for backend)
rather than 'l' (for lsp)
<<Amendment 8 06/11/18>>
Documentation corrections -- based on feedback from sdwolfz
1. Fix `short-name` for directories. Previously it was empty string.
2. Remove redundant check for existing buffer:
1. `(get-buffer new-name)` always return `nil`, because `new-name` is a path
to file, not a file name.
2. `(find-file new-name)` handles this situation.
3. `dired-rename-file` handles this situation as well.
3. Use `dired-rename-file` to do the dirty stuff.
1. Renames files and directories.
2. All associated buffers are updated. If you rename a directory, all buffers
that were visiting files from that directory now visit files from new
directory.
4. Properly handle recentf list update for directories - all files from the
renamed directory are removed from recentf list and the same files but from
new directory are added back.
problem:
`'face 'minibuffer-prompt` and `new-name` are shown with an orange color,
and `describe-char` shows that they have the face: `font-lock-warning-face`
solution:
adding a newline before them removes the orange warning color
Following #11384
Right now treemacs-lock-width variable affects width lock only during treemacs
initialisation. Any later modifications have no effect on treemacs behaviour.
This commit changes it by moving evaluation of treemacs-lock-width to the
treemacs-mode-hook.
Document how to use the newly introduced variables to enable sayid and
clj-refactor.
Also update the documentation on adding CIDER-related dependencies. CIDER 0.10
is three years old now, so I think the case where cider middleware isn't
injected automatically is getting rare. However there is still a good use case
for running your Clojure process outside of Emacs/CIDER, meaning you don't get
to benefit from `cider-jack-in`'s magic insjections, so I have focused more on
that case, adding information for clj-refactor's and sayid's middleware, as well
as CIDER's.
When shell layer intercepts the clear commands and tries to replace it with
erase-buffer, this fails with 'Text is read-only'.
Work around this by using inhibit-read-only.
Previously org-clock-related keybindings lived under "aok" (when not
in org-mode) and "mC" (in org-mode). This changes "aok" to "aoC",
adds more keybindings under both, and makes them more consistent
overall.
After merging [this PR](https://github.com/Silex/docker.el/pull/39), the
`docker.el` package now supports integration with `docker-compose`. This
commit adds a keybinding for `docker-compose`, which allows us to
interact with compose files for the current project.
`flyspell-correct-word-generic` and `flyspell-correct-previous-word-generic` are
deprecated and replaced by `flyspell-correct-at-point` and
`flyspell-correct-wrapper` respectively.
`flyspell-correct-wrapper` has a so-called 'rapid' mode, where you can correct
multiple words in a single run.
problem:
Which-key subgroups get the default name `+prefix`. The subgroup has to be
opened to find out which keys it contains.
solution:
Add more descriptive names:
```
c treemacs-create
o treemacs-visit-node
oa treemacs-visit-node-ace
t treemacs-toggles
y treemacs-copy
C-p treemacs-projects
C-p c treemacs-projects-collapse
```
Invoking C-x C-f (spacemacs/helm-find-files) right after startup, before helm is
loaded, caused a void function error for helm-current-directory. This fixes it.
Before this commit, Spacemacs assigned an anonymous function to the
compilation-finish-function variable. This variable was deprecated
in Emacs 22.1 and removed in Emacs 27.1 in favor of the
compilation-finish-functions hook. Consequently, the assignment has
no practical effect in Emacs 27.1.
Although we could add the Spacemacs function to the new
compilation-finish-functions hook, the function serves little purpose; it
prints a generic message when compilation succeeds:
compilation ok.
and it prints the key binding for spacemacs/next-error when compilation
fails:
There were errors. Use SPC e n to visit.
However Emacs already prints messages to indicate success or failure,
respectively:
Compilation finished
or (for example)
Compilation exited abnormally with code 1
Because the Spacemacs function prints its "There were errors" message
immediately after Emacs prints its "Compilation exited abnormally" message,
the latter was obscured. Because the exit code in Emacs's message could be
useful and the added value of the Spacemacs function's message is minimal,
we can simply delete the Spacemacs function.
* layers/+spacemacs/spacemacs-defaults/funcs.el
(compilation-finish-function): Delete the assignment and anonymous
function.
Add a prefix for rake and refactor/RuboCop/robe commands. Add test and toggle
prefix for enh-ruby-mode. Use enh-ruby-toggle-block in enh-ruby-mode, instead of
ruby-toggle-block. Remove the prefix in prefixes "ruby/".
The package adds some convenient commands to run elm-test on:
- a file
- a directory
- a projects
It also provides a handy way to toggle between a test and its implementation.
The `SPC f a` keybinding was already used by the `fasd` layer so a
workaround had to be found for the newly introduced one.
Refs: c2e377c902
Refs: #11224
Thanks @agzam for pointing it out and investigating potential solutions.
`m g a` typically used to switch between .c and .h files.
Rebound `goto-viewport-symbol` under `m g k`
Added missing bindings for lsp-ui-find-implementation under `m g l`
Incorporated some other feedback from @yyoncho and @sdwolfz
(See PR discussion for detail)
Fix issue #11207.
Only try to disable eyebrowse's advice on rename-buffer if we can find that
advice. If eyebrowse is older than version 0.7.6 (2017-11-22), the advice
will not exist, and attempting to delete it will cause an error:
ad-disable-advice: rename-buffer is not advised
* layers/+spacemacs/spacemacs-layouts/funcs.el
(spacemacs//activate-persp-mode): Only disable the
eyebrowse-fixup-window-configs advice for rename-buffer if the advice
exists.
See README.org for details
<<amendment 1>>
Updated some keybindings based on CONVENTIONS doc
Corrected file headers
Incorporated some immediate feedback from MaskRay
<<amendment 2>>
Corrected keybindings in README.org
<<amendment 3>>
Eliminated stray org-mode tag at table foot in README.org
Eliminated new 'l' prefix and moved bindings under 'g'
<<amendment 4>>
Updated defaults in config.el based on feedback from sebastiencs (lsp/lsp-ui dev)
- lsp-ui-sideline enabled by default
- lsp-ui-peek-expand-by-default disabled
<<amendment 5 09/04/18>>
Removed 'spacemacs/' prefix from lsp-format-buffer binding
<<amendment 6 09/04/18>>
Moved lsp-ui-peek bindings under j (jump)
Added goto bindings for new lsp-mode functions goto type definition and goto implementation
<<amendment 7 31/05/18>>
Corrected layer title in file headers
Rebased on dev tip (390462e)
<<amendment 8 03/07/18>>
Added keybindings for lsp-describe-thing-at-point,
lsp-workspace-restart, lsp-execute-code-action suggested by Yyoncho (LSP
Java)
Added avy keyboard navigation function provided by MaskRay
Reverted lsp-ui-peek to expand by default after an upstream change that
restricts expansion to current document, addressing the previous
performance issue.
<<amendment 9 04/07/18>>
Corrected keybinding for lsp-describe-thing-at-point
<<amendment 10 19/07/18>>
Rebound lsp-restart-workspace under mlq
Declared 'lsp' prefix (myrgy)
Added evil-set-command-property fix suggested by Yyoncho
Moved lsp-c-c++ layer from private branch to this PR after spending too
many hours of my life rebasing after circle CI picks up a formatting
error :)
<<amendment 11 25/07/18>>
Rebased
Bound cquery-freshen-index under lf
Bound cquery-preprocess-file under lp
<<amendment 12 01/08/18>>
Rebased
(c-c++ layer) moved semantic refactor refactor-at-point binding from mr
to mrp to prevent key binding error when semantic layer enabled
<<amendment 13 17/08/18>>
Added option to select ccls or cquery backend based on work by myrgy
Rebased on current upstream develop
<<amendment 14 20/08/18>>
Incorporated feedback from myrgy and maskray.
Corrected some duplication/inconsistencies.
Rebased.
<<amendment 15 21/08/18>>
Reduced duplication in backend config
<<amendment 16 22/08/18>>
Removed lsp-c-c++ layer example -- to be merged with c-c++ layer once
this PR is sorted
<<amendment 17 23/08/18>>
Added CHANGELOG.develop entry as per updated contribution guidelines.
<<amendment 18 24/08/18>>
Moved some keybindings as per feedback from sdwolfz
Before this commit, spacemacs-layouts/init-eyebrowse advised rename-buffer
with spacemacs//fixup-window-config. However, this advice causes problems
if persp-mode is not activated when a function tries to rename a buffer.
In particular, Spacemacs's CI uses Emacs's batch mode to run
spacemacs/publish-doc, which uses rename-buffer. In batch mode, persp-mode
is not activated, and consequently spacemacs/publish-doc was failing:
Publishing file /home/travis/.emacs.d/news/news01.org using ‘org-html-publish-to-html’
Skipping check for new version (reason: dotfile)
Wrong type argument: hash-table-p, nil
Warning (emacs): recentf mode: Wrong type argument: hash-table-p, nil
spacemacs/publish-doc failed
In order to avoid problems, this commit moves the advising of rename-buffer
from spacemacs-layouts/init-eyebrowse to spacemacs//activate-persp-mode.
* layers/+spacemacs/spacemacs-layouts/packages.el
(spacemacs-layouts/init-eyebrowse): Move advice from here...
* layers/+spacemacs/spacemacs-layouts/funcs.el
(spacemacs//activate-persp-mode): ...to here.
The problem was not with the existing bindings, but rather:
1. C-x C-e (spacemacs//counsel-edit -- from an ivy completion session) would
land you in an ivy-occur buffer with wgrep enabled but in motion state. This
meant the spacemacs key bindings for wgrep were not available. In particular,
the usual routine is:
* (enter ivy completion session, e.g. with SPC s a P)
* C-c C-e
* edit the buffer as needed
* , , (wgrep-finish-edit)
* , s wgrep-save-all-buffers
2. That last binding (, s) was not there.
One issue is that , s is available before wgrep-finish-edit is called, which is
confusing because I don't think it has any useful function at that point in
time.
Eyebrowse stores window configurations for workspaces in frame properties.
These window configurations reference buffers by name, which means that
these references must be updated when a buffer is renamed. To this end,
eyebrowse advises rename-buffer.
Spacemacs integrates eyebrowse with persp-mode by saving the eyebrowse
workspaces for a perspective as a parameter for that perspective so that
Spacemacs can restore a perspective's workspaces when switching to that
perspective. However, eyebrowse's advice for rename-buffer fails to update
inactive workspaces, for which the window configurations are stored in
perspective parameters and not frame properties.
This commit disables eyebrowse's advice and adds perspective-aware advice
to update buffer references in workspace window configurations.
* layers/+spacemacs/spacemacs-layouts/funcs.el
(spacemacs//fixup-window-configs): New.
* layers/+spacemacs/spacemacs-layouts/packages.el
(spacemacs-layouts/init-eyebrowse): Remove eyebrowse-fixup-window-configs
advice for rename-buffer and advise with spacemacs//fixup-window-configs
instead.
Define an alias and use it internally so that (1) the name of the command will
be better to understand what it does than the original name and (2) Helm or Ivy
interface is invoked instead of Ido.
These are suggested in PR #11115.
Try again to fix issue #11176.
* layers/+source-control/github/packages.el (github/init-magithub): Disable
the GitHub project status, pull requests, and issues sections by default.
* layers/+source-control/github/README.org: Document that these sections
are all disabled by default, and how to enable them.
Syntax-checking and auto-complete layers should more
directly communicate that their function depends on
the installation of individual language layers to avoid
misunderstandings with new users who only install
syntax-checking or auto-complete and install the packages
manually.
This fixes issue #5776.
* layers/+spacemacs/spacemacs-layouts/packages.el
(spacemacs-layouts/init-persp-mode):
Set persp-add-buffer-on-after-change-major-mode to 'free.
Escape the filename before calling python on the filename in the shell in the
function `spacemacs/python-execute-file` by calling `shell-quote-argument`.
Fix issue #9477: spacemacs/smart-closing-parenthesis sometimes adds an
extra closing parenthesis.
spacemacs/smart-closing-parenthesis may call sp-up-sexp twice: first to
find the position of an existing closing delimiter or the position at which
one would be inserted, and second to insert a delimiter if necessary or
else move point after the existing closing delimiter.
The problem is that each call to sp-up-sexp may insert a delimiter, but
spacemacs/smart-closing-parenthesis is supposed to insert at most one.
The solution is to undo any edit performed by the first call to sp-up-sexp.
* layers/+spacemacs/spacemacs-editing/funcs.el: Undo any edit performed by
the first call to sp-up-sexp.
Fix SPC m c c (spacemacs/python-execute-file) and SPC m c C
(spacemacs/python-execute-file-focus), which were causing
the following errors:
save-current-buffer: Symbol’s value as variable is void: python-shell--interpreter
error in process filter: python-shell-comint-end-of-output-p: Wrong type argument: arrayp, nil
error in process filter: Wrong type argument: arrayp, nil
This commit also fixes issue #9070: spacemacs/python-execute-file-focus
does not change focus.
The problem is that spacemacs/python-execute-file-focus invokes
spacemacs/python-execute-file, spacemacs/python-execute-file invokes
inferior-python-mode, and inferior-python-mode uses the variables
python-shell--interpreter and python-shell--interpreter-args, which
the python package declares but does not initialize.
To resolve this problem, this commit initializes both variables to nil.
* layers/+lang/python/config.el: Initialize python-shell--interpreter and
python-shell--interpreter-args to nil.
It seems like the package is missing. We waited a while thinking it will be fixed but
people are starting to get annoyed by this. To be re-enabled once the upstream issue
is resolved.
Refs: https://github.com/melpa/melpa/issues/5657
Fix issue #10760: Magithub sections are not added to Magit.
The problem is that the Spacemacs init function's use-package form for
Magithub specifies both :after magit and :defer t. It should specify only
the former so that Magithub will be loaded when Magit is loaded.
* layers/+source-control/github/packages.el (github/init-magithub): Delete
:defer argument.
Add aliases for commands that ido-mode remaps that we want to bind to
Spacemacs keys, and use these aliases in the Spacemacs key bindings so that
these bindings invoke the original commands rather than the ido commands.
See <https://github.com/syl20bnr/spacemacs/issues/10237#issuecomment-359861570>.