Emacs Quick Reference
posted on 25 Jan 2023 under category ref
<keys> C-h press <keys> and ask for available bindings begin with that
C-h
?
C-h help help
---------------------------------------------------------
a <topic> apropos, match command names to word list, org regexp
with (u), also show noninteractive functions
m desc this modes
k cmd name and doc for key
x <cmd> docs for cmd
f <fn> doc on lisp func
v <var> doc on lisp variable
b desc bindings
w <cmd> where-is, show keys for cmd
c echo cmd name for keys
d <topic> cmds or vars whose doc matches topic
o <sym> doc of lisp symbol
r manual
e show echo area msg (*Messages* buffer)
---------------------------------------------------------
C <coding> desc coding system
C-d help debugging emacs
C-f faq
h "HELLO"
i info browser
l show last 200 keystrokes and cmds (lossage)
n what's new (prefix to specify version)
C-q "quick help" cheat sheet
s show current "syntax table"
t tutorial
---------------------------------------------------------
p list package
P desc package
---------------------------------------------------------
F <cmd> enter info, goto node for cmd
K ..keys
S <sym> ..symbol
I <method> desc input method
. special text area
Cmd
apropos-*
user-option
[local-]variable
value
Variable
system-configuration-options
Examples
C-h a <topic>
C-h i d m emacs <ret> i <topic> <ret> search by topic in info
C-h i d m emacs <ret> s <topic> <ret> search in text in info
s show source
I lookup topic in elisp manual
c customize
C-c C-c help for symbol under cursor
b/e node begin/end
<ret>/f follow
<tab>/S-<tab> next/prev cross-ref or menu
m menu by name
g node by name
d directory node
L show history
n/p forward/back in pages
r/l forward/back in history
</> top/final
[/] back/forward, considering all nodes flat
^ up
s/S search info by regex case in/sensitively
T TOC
C-s, CM-s Isearch multiple info
i jump by index topic
, next i match
M-n clone to new window
C-x prefix, typical do with windows, files, buffers, etc.
M-x prefix to exec extended cmd
C-g get out current cmd, or long running process (frozen)
esc-esc-esc "all purpose" get out
C-x C-c exit
To start Emacs as server:
M-x server-start(server-start) to .emacs.emacs --daemonemacsclient --alternate-editor=""systemctl --user enable emacsEach “major mode” make few cmd behave differently, they are also the name of a extended cmd, which can be used to switch to that mode.
Mode are auto chosen when first visit the file, you can explicitly select a new mode with M-x <mode>. You can then return to the emacs chosen mode by M-x normal-mode.
To change default mode, use Customization, or by adding line like:
(setq-default major-mode 'text-mode)
Every major mode (except Fundamental) defines a “mode hook”, commonly used to enable minor modes:
(add-hook 'text-mode-hook 'flyspell-mode)
Every buffer can have many “minor modes”, some minor modes are global, affects entire editor.
“X (eXtend) Command” comes in two flavors: C-x for “Char eXtend”, followed by one char; M-x for “named cmd eXtend”, followed by long name.
M-x <cmd> run cmd
M-X run cmd relevant to buffer
C-x z repeat last cmd
C-x <esc> <esc> repeat last minibuffer cmd
Each buffer has a “default dir”,
pwd
cd
Everything you type is saved in “minibuffer history list”. Different type of argument(filename, cmd name, etc.) is saved inside separate list.
You can ignore the prompted default path, just start typing / or ~, like:
# ignore local path
Find file: /u2/emacs/src//etc/termcap # with /
Find file: /u2/emacs/src/~/.emacs # with ~
# ignore remote path
Find file: /remote/dir///etc/termcap # with // to ignore remote
You can be in “Recursive editing” if you issue new commands while the previous is still running, the mode line will show ‘[…]’ for each editing level. Remember to exit when you no longer need it.
Inside Minibuffer:
M-n/p,down/up next/prev item in history
M-r/s regex search history
C-o expand minibuffer with newlines
---------------------------------------------------------
CM-c resume to outer editing level
C-] abort this editing level
---------------------------------------------------------
TAB complete, works also when point is not at the end
SPC complete up to one word
? list completions
M-up/down select prev/next item
M-v focus completion window, See "Completion Mode"
Cmd
top-level abort all level
Var
completion-style
completion-auto-help
completion-auto-select
read-file-name-completion-ignore-case
----------------------------------------------------------------
icomplete-mode
fido-mode
---------------------------------------
history-delete-duplicates
Completion
“Icomplete Mode” and “Fido (Fake Ido) Mode” show completion entries in a single line.
Completion Mode
q quit window, back to minibuffer
z quit window and minibuffer
n/p,<tab>/S-<tab> next/prev item
ICompletion Mode
C-j complete first and exit
RET nocompletion exit
C-./, rotate list
M-<tab> select first, don't exit (convient for descending a dir)
CM-i force complete
Fido Mode
C-s/r rotate list
C-k kill buffer/file
Command may use it as repeat count, change default behaviors.
C-u <num> 4* times
M-<num>
CM-<num>
<num>
The reason why these all do the same, is to NOT break your TEMPO.
“pages” are delimited by formfeed (displayed as ^L).
“defun” means top level function definition.
“paren” also mean brac, bracket or other delimiters to match pairs.
“sexp” (borrowed from lisp) means any balanced expresesions in a language.
C-b/f back/forward by char
M-b/f ...word
CM-b/f ...sexp
C-p/n ...line
CM-p/n ...paren
C-a/e beign/end of line
M-a/e ..sentence
M-{/} ..paragraph
C-x [/] ..page
M-</> ..buf
CM-a/e ..defun
C/M-v down/up by screen
C-x </> left/right screen
M-m first non-space char of this line
CM-<up> back up one level of indent
CM-u/d up/down in paren level
---------------------------------------------------------
C-xn set goal column (affect where C-n/p move to)
---------------------------------------------------------
[N] M-g goto
g line
c char pos in buffer
<tab> column
i defun ("imenu")
---------------------------------------------------------
C-l recenter
M-r reposition cursor
CM-l smart recenter (try to show comments and definition)
Mode
which-function-mode show func name now inside
subword-mode CamelCapsAsSeperateWordsWhenMoving
glasses-mode display extra _ for CamelCaps
Var
imenu-auto-rescan
imenu-sort-function
Cmd
what-page
C-x i insert from other file
C-x x i insert from other buffer
C-q <num> quoted insert, next insert will treat "literally", useful
for inserting constrol chars
-- key translation -------------------------------------------------------
C-x 8
<num> common unicode char
< «
> »
+ ±
C ©
R ®
P ¶
S §
T ™
Y ¥
[ ‘
] ’
{ “
} ”
x ×
< «
> »
~= ≈
_<num> ₁
_< ≤
_> ≥
^<num> ¹
// ÷
/= ≠
Mode
electric-pair-mode auto insert pairing parens
Cmd
insert-file-literally
write-region inverset of C-x i, copy region to other file
C-x 8e
+/-/0 zoom in/out/reset
d describe
e/i insert
l list
r recent
s search
<del> in emacs means <backspace>.
“Killed” text can be reinserted (“yank”), “deleted” cannot (but can be undo). typical cmds that remove just one char or blank lines or spaces do deletion, like DEL or C-d (but given argument, they kill).
By default, emacs expect “sentence” to be ended with two spaces, but can be changed with sentence-end-double-space.
<del>
C-d del next char
M-d ..to word end
M-<del> ..to word begining
C-k kill to line end
M-k ..to sentence end
C-x <del> ..to sentence begin
C-x C-o ..blank lines around, leaving one
M-\ ..spaces around
M-<spc> ..spaces around, leaving one
CM-k ..sexp
----------------------------------------------------------------
M-^ join up
----------------------------------------------------------------
C-w region ("cut")
M-w save region to kill ring ("copy")
C/M-y yank last/prev kill
M-z zap (kill) to char
----------------------------------------------------------------
yank-media
[-N] M-l/u/c lower/upper/capitalize [prev]next word
C-x C-l/u lower/upper region
[u] M-; dwim: insert or realgin line (or region), or go to start
[u] C-x C-; toggle line
[u] C-x ; set column
CM-j,M-j <ret> and insert and align comment
C-c C-c add comment delimiter to lines in region
Cmd
uncomment-region
Var
comment-*
padding
start
end
multi-line
indent-function
C-o insert newline after point but leave point in front
C-j insert newline without auto-indentation
C-t char
M-t words
CM-t sexp
C-x C-t line
---------------------
transpose-sentences
transpose-paragraphs
transpose-regions
<tab> indent (insert whitespace, indent whole line or region) by mode
M-i indent to next stop
C-u <tab> indent in paren group rigidly
CM-o split at pos, indent next
CM-q reindent lines in sexp
C-cq reindent all defun
CM-\ indent all lines in region
C-x <tab> indent rigidly, interactively
C-c . set indent "style" (like "k&r", "gnu" in C)
Cmd
indent-relative
edit-tab-stops
tabify space to tab
untabify tab to space
align align region by "alignment rules"
[u] align-regexp align by [complex] regexp
align-highlight-rule
align-unhighlight-rule
electric-indent[-local]-mode
Var
indent-tabs-mode
tab-width
tab-always-indent
When using edit-tab-stops, the resulting stops will be extended forever by repeating.
CM-i,M-<tab> complete
Cmd
completion-help-at-point list available completions
“Filling” mean lines breaked (physically) by EOF char. “Wrapping” (in “Visual Line Mode”) only means for display, when not wrapped, then lines are “truncating”.
The “fill prefix’ feature allows paragraphs to be filled so that each line starts with a special string of characters.
In “Auto Fill Mode”, lines break automatically when you type SPC or RET, inserting “adaptive fill prefix”, which will be deduced from the first and/or second line of current paragraph. Note that it does not refill entire paragrah; it breaks lines but not merge lines, hence editing in middle of paragraph can result not being correctly filled, we can call fill commands to fix this.
C-x . set prefix (affects M-q, M-^, and C-o)
C-x x t toggle truncation lines
C-x f set fill column
M-q fill paragraph with prefix
<num> M-q "justify" (insert spaces between words) and re-fill
Cmd
display-fill-column-indicator-mode
auto-fill-mode
fill-region
fill-region-as-paragraph
center-line/paragraph/region
Var
display-fill-column-indicator-column/charactor
fill-column-indicator
There is no redo, but undoes are undoable, too.
If you need to get back to a previous buffer state, simply move the cursor, and press C-/ until you find what you want, notice the Mode Bar, where there is no ‘*’, then the buffer is same as last SAVE.
With “Selective undo”, you can undo only in region.
C-/ undo
C-_
C-x u
---------------------------------------------------------
M-% query replace
sort-*
lines/paragraphs/pages
fields/numeric-fields
columns
reverse-region
Insertion in hexl mode always overwrite.
hexl-find-file open as hex
hexl-mode view this in hex
----------------------------------------------------------------
CM-d/o/x insert byte with dec/oct/hex
CM-a/e move to begin/end of 512-byte page
CM-[/] ..1k-byte..
M-g/j goto address in hex/dec
C-cc leave hexl, go back to normal
By default, searching:
But replacement commands do NOT use lax space matching by default.
Some of those can also be used to toggle mode when already in search session, Hilight-related cmds can be used when exiting search.
C-s/r [<ret>] [non]isearch forward/reverse
CM-s/r [<ret>] regex [non]isearch
M-% query(one by one) replace
CM-% regex query replace
---------------------------------------------------------
M-s. isearch this symbol
M-s _ isearch given symbol
M-s w isearch given word
[N] M-s o occur [show N context line]
M-s h hilight:
./r/l/p symbol at point/regex/line/phrase
u unhilight (done)
f/w lock find/write ??
---------------------------------------------------------
<ret> exit, leaving pos there
C-g abort, return starting pos
When Searching
Motion:
M-s </> goto first/last match
Search String Editing:
M-n/p next/prev in history
M-e focus search string for editing
C-w yank next char
CM-w yank next symbol
CM-z yank until given char
M-s C-e yank whole line
CM-d del next char
CM-y append next char
M-<tab> complete
Toggle Search Mode:
M-s <spc> lax space
M-s c, M-c case sentitivity
M-s i text visiblility
M-s r, M-r regex
M-s w word mode
M-s ' char folding
RET when search string empty launches “nonincremental search”, it’s necessary only if want type a char next.
When editing search string, other editing cmds works, too. Like yank, undo, etc. To search for newline, use C-j.
Cmd
replace-string unconditional replace
replace-regexp
---------------------------------------------------------
multi-isearch-buffers[-files][-regexp]
occur M-s o
multi-occur[-in-matching-buffers] occur in many buffers [regex]
---------------------------------------------------------
how-many search and only show match number
flush-lines search and delete matched lines
keep-lines ..nonmatched lines..
kill/copy-matching-lines
Var
search-highlight[-submatches]
search-default-mode
search-upper-case
case-fold-search
isearch-*
allow-scroll scroll won't exit isearch
allow-motion motion ...
yank-on-move extend search string with motion
hide-immediately
---------------------------------------------------------
replace-*
[regexp-]lax-whitespace
“Occur Mode” is an interactive search mode, it bring all matches in a buffer with context lines around and let you press e and modify matchs one by one.
RET goto occurrence
o same, in other window
C-o show this occurrence
e enter "occur edit mode"
g revert buffer, refresh result
n/p next/prev
l recenter
---------------------------------------------------------
M-n/p goto next/prev (works globally)
Changes to the *Occur* buffer are applied to origin buffer.
C-c exit editing and apply changes (occur-cease-edit)
NOTE: in elisp, you also have to escape the escape character.
Search
NOTE there is no \d, you have to use [0-9] or [:digit:].
.
*
+
?
*?,+?,?? non-greedy
[…], [^…] alternative chars
[:ascii:]
[:alnum:]
[:alpha:]
[:blank:]
[:cntrl:]
[:digit:]
[:graph:]
[:lower:]
[:multibyte:]
[:nonascii:]
[:print:]
[:punct:]
[:space:]
[:unibyte:]
[:upper:]
[:word:]
[:xdigit:]
^
$
\ escape, also introduce special constructs
-----------------------------------------------------------------------
\| alternative expr
\(…\) grouping
\(?:…\) shy grouping (dont capture)
\D back ref #d
\{m\} repetition
\{m,n\}
\` start of string or buffer
\' end of...
\= point
\b, \B word,non-word boundry
\<, \> start,end of word
\_<, \_> start,end of symbol (\w or _)
\w,\W any word,non-word
\scode, \Scode char whose syntax code is,not code. code can be:
- whitespace
w word(lower,upper,digit,unicode char)
_ symbol(word,!,_)
. punct(.,;)
(,) open,close of pairs ((),[],{})
" string block mark(',")
<,> comment boundary
\ccate, \Ccate ..catagory is,not cate
Replace
\& entire match
\D Dth match
\# count (from 0) of replements already made
\? prompt for manual input
\, Lisp expr start, inside it, can use:
\& same as above
\D same, but D can be multiple digits
\# same as \D
\#& reference match as a number
\#D reference match as a number
Examples
Add consecutive numbered strings like ‘ABC00042’ to column 73-80:
M-x replace-regexp <RET> ^.\{0,72\}$ <RET>
\,(format "%-72sABC%05d" \& \#) <RET>
“M-x re-builder” let your construct regex interactively. Best to change reb-re-syntax to string to avoid to much of backslashes, if you use lots of lisp, then better changed it to rx.
C-c
C-b change target buffer
C-c toggle case
C-e enter subexp mode
<tab> change syntax
C-q quit
C-r go prev match
C-s go next match
C-u force update
C-w copy
Cmd
reb-*
toggle-case
change-syntax
grep grep regex
lgrep grep glob pattern
grep-find/find-grep find&grep regex
rgrep find&grep glob, recursively
zrgrep/rzgrep find&grep glob, recursively, in gzip file
kill-grep
(An alternative to xref-find-reference and tags-search.)
When greping, use ‘grep’-style regexp (usually single-quoted). You can chain grep together, too.
When giving filename, $ can be used to subbstitue env variable, use $$ for literal ‘$’.
If permissioin don’t allow reading the file, we can use ‘su’ or ‘sudo’ methods:
/su::/path/to/file
To prevent special characters in filename, use /: to escape:
/:/foo:/bar file bar in dir /foo:
/ssh:baz:/:/foo:/bar file bar in dir /foo: on host baz
quoting with /: is also a way to enter in minibuffer a filename that contains ‘$’.
The visited file’s buffer named are typicalled the same as file name but with dir name omitted. If specified a dir name, emacs enter “Dired” mode. Emacs can auto uncompress compressed files when visiting, then recompress them when saving. “Tar Mode” is like Dired when visiting archive file like *.tar.
“Revert” a buffer mean reload its content from disk file, but for non-file buffers (such as Dired), reverting means recalculate their contents.
Using “fileset”, we can op on group of files.
“Backup file” are named by appending ‘~’ to filename, there can be many backup files for one file, like eval.c.~1~, eval.c.~2~.
“Auto-save files” are named by wrapping ‘#’ to filename. if auto-save-visited-mode is on, then auto-saving is identical to explicit saving. Upon system crash, you can use M-x recover-file or M-x recover-session to recover from auto-save files.
“Shadow” is a copies of certain files in more than one place – possibly on different machines. “Shadow cluster” is a group of hosts that share directories, so that copying to or from one of them is sufficient to update the file on all of them.
Prefix,keys desc
-----------------------------------------------------------------------
C-x
[4/5] C-f find("visiting") a file [in other window/frame]
NOTE: can also be used to switch buffers
C-r readonly
C-v kill this and visit alternate
C-s save this, with numarg, also save backup file
s save all buffer
C-q make writable
C-w write file to given path(save as)
d list dir (Dired)
x g revert
-----------------------------------------------------------------------
[C-u] M-~ mark as not modified, prevent from accidentally saving
C-xs save buffer
To save buffer to a different file without visiting that file, use C-x h then M-x write-region.
Cmd
find/insert-file-literally as ASCII without encoding
find-sibling-file
ff-find-related-file
set-file-modes
copy-file/directory
delete-file/directory
rename[-visited]-file
add-name-to-file make hard link
make-symbolic-link
move-file-to-trash
set-visited-file-name
---------------------
filesets-add/remove-buffer
filesets-edit
filesets-open/close
filesets-run-cmd
Mode
[global-]auto-revert-[tail]-mode useful for "tailing" a log file
Var
save-some-buffers-default-predicate
delete-by-moving-to-trash
require-final-newline
Emacs uses its own multibyte char encoding, a superset of the Unicode standard.
C-x RET
f choose for save or revist
c choose for next cmd
r revist this file with
Cmd
recode-region
“TRAMP”: Transparent Remote Access, Multiple Proto.
Remote path syntax:
/method:host:/path/to/file
/method:user@host:/path/to/file
/method:user@host#port:/path/to/file
ANY text in emacs window is always part of some buffer.
C-x
C-b listing
[4/5] b switch/create [in other window/frame]
<left>/<right> switch prev/next
C-q toggle readonly/writable
x r change buffer name
x u rename uniquely by appending a number
x i insert text from another buffer
k kill
4 c create indirect buffer in other window
C-u M-g g <num> switch and goto line in other window
Cmd
buffer-menu[-other-window] edit buffer list
ibuffer list, Dired like
bs-show list, customizable
bs-customize
view-buffer view in View Mode
clean-buffer-list kill all unmodified
kill-some-buffers ask each for kill
kill-matching-buffers
----------------------------------------------------------------
make-indirect-buffer
clone-indirect-buffer
Var
global-auto-revert-non-file-buffers
uniquify-buffer-name-style
In buffer listing like:
CRM Buffer Size Mode File
. * .emacs 3294 Emacs-Lisp !/.emacs
% *Help* 101 Help
’.’ means this is current buffer, “%” means readonly, ‘*’ means modified.
A “indirect buffer” shares text from its “base buffer”, like symlink between files:
One way to use it is to view different views of an outline.
Run clean-buffer-list or other funcs in ‘midnight-hook’ once a day.
“Frames” is collection of windows, together with its menus, scroll bars, etc.
When buffer shown in many windows, they can have different regions, but same value for marks.
“Tab line” is per window for showing buffers in window; “Tab Bar” is per frame for switching between window configs.
-- Window ---------------------
C-x
0 delete this
1 keep this only
2 split down
* with positive numarg, set origin window height
* with negative numarg, set new window height
3 split right
^/_ taller/shorter
{/} narrower/wider
+ all equal
o focus on other
4 0 del this, kill bf in it
w 0 del window who showing given buffer
4 other window:
b/f/r/d/m/. switch to or create buf/file/readonly/dir/mail/definition
4 general prefix affect next cmd
l general prefix affect next buffer cmd
CM[S]-v scroll other up[down]
CMS-l recenter other
-- Frame ---------------------
C-x
5 frame:
0 delete
1 keep only this
u undelete
2/c create default/as current
b/f/d/m/./r other frame, see "other window"
o select other and raise it
C-z min("iconify") current frame
M-f10 max
f11 fullscreen
“Speedbar” is a special frame, op on it will act on its associated frame, for managing directories and tags.
Q delete frame
g refresh
n/p next/prev entry
CM-n/p next/prev list
M-n/p next/prev restricted
q close frame
t toggle updates
C-x b switch buffer attached frame
Winner Mode
Set winner-dont-bind-my-keys to prevent the following key bindings.
C-c
left/right undo last/redo window configuration
Windmove Mode
S-up/down/left/right select by direction (disable shift selection!)
Cmd
compare-windows
windmove-*
swap-states* swap contents with another
speedbar
winner-mode
follow-mode scroll buf together when it's in many windows
scroll-all-mode scroll all when scroll one
----------------------------------------------------------------------------
windmove-[delete-]default-keybindings
Var
window-resize-pixelwise
----------------------------------------------------------------------------
winner-boring-buffers[-regexp] ignore when restore
C-x C-+/-/= incr/decr/reset font size
[N] C-x $ fold ("selective display") [at indent level N]
C-x $ unfold all
M-= show region's lines,sentences,words,char counts
[C-u] C-x = show char info [describe char]
M-s h
r <regex> hilight word with face
u <regex> unhilight
l <regex> hilight line
p <phrase>
. hilight sym at pos
w write hilighting regex/face to this point
f extrace from w writen contents
C-x w
h <regex>
r <regex>
l <regex>
p <phrase>
.
b
i
Mode
scroll-bar-mode
fringe-mode
---------------------
line-number-mode
column-number-mode
hl-line-mode
size-indication-mode
Cmd
what-line
display-line-numbers-mode
“Outline Mode” consider lines begin with * a “heading line”, and other lines “body line”. A heading line with all its following body line is an “entry”, and with all even deeper body an “subtree”.
A common trick is to use indirect buffer to display one outline and one expanded buffer of one file.
The major mode prefix at C-c, minor at C-c @.
<tab> cycle sub-header visibility
S-<tab> cycle entire buffer visibility
C-c
C-c/e hide/show (expose) entry
C-d/s hide/show subtree (not including header)
C-l/k hide/show leaves (including subentries)
C-i show immediate subheading
C-t hide all body
C-a show all
C-q hide all except top N level of heading lines
C-o hide all other
----------------------------------------------------------------
[N] C-c
C-n/p goto next/prev visible header
C-f/b same, but at the same level
C-u up a level
“Narrowing” restrict view to certain region.
C-x nn to region
C-x nw widen (visible again)
C-x np to page
C-x nd to defun
“Foldout Mode” extends Outline Mode, let you zoom in (show then narrow) and out of subtrees. Since narrowed, global editing actions will only affect text under zoomed-in heading.
[n] C-cz zoom in this this header's body and [n level of]child
subheadings
[-] C-cz ..body
[0] C-cz ..whole subtree
----------------------------------------------------------------
[u] C-cx exit zoom (unfold)
To use this package:
(with-eval-after-load "outline"
(require 'foldout))
Cmd
reveal-mode
Var
search-invisible
outline-regexp
outline-default-state
For spell check to work, one of Hunspell, Aspell, Ispell or Enchant must be installed. Spellchecker will look up in standard dictionary and your personal dictionary.
[N] M-$ [continue last] correct
CM-i complete word
C-. same, only in flyspell mode
When there is “near-misses”, here are valid action:
digit replace this with option N
SPC skip
r replace this with given
R replace this and do a query-replace
a accept, treat it as correct for this session
A same, but only for this buffer
i insert to private dict file
m i, also specify complete info
u i, with lowercase
l look in dict for other then select one to replace
x quit, move point back
q quit, kill checker subprocess
C-r recursive edit
Cmd
ispell-*
-buffer/region/message check and correct
-comments-and-strings/comment-or-string-at-point
-change-dictionary
-kill-ispell kill checker subprocess
flyspell-mode hilight all misspell (NOTE might slow
down cursor movement)
flyspell-prog-mode same, but only for comments and string
“Mark ring” contains marks per buffer, each buffer has only one mark pos, per buffer marks also are in “global mark ring”. When buffer’s mark is active, we say also the region is active. If “Transient Mark Mode” is enabled (default), region will be hilighted.
Long distance jump cmds(M-<, M->, etc.) will auto set mark.
Region always extends between point and the mark, or pressing “SHIFT” (“shift selection”) when moving.
Many cmds only operate on active region if there is one, like ‘M-%’.
C-<spc> mark point
C-xx swap point and mark (return to prev pos)
----------------------------------------------------------------
C-<spc> C-<spc> mark without activating
C-u<spc> pop and go last mark in buffer
C-x<spc> ..in global
----------------------------------------------------------------
M-@ end of next word
CM-@/<spc> end of next sexp
----------------------------------------------------------------
M-h paragraph
CM-h defun
C-xp from page begin to end
C-x h all
Var
highlight-nonselected-windows
Useful for editing multicol formate text. Given combination of point and mark can be treated as region or rectangle, depending on the cmd.
killed rectangle is not saved in kill ring, but in special place.
CM-mouse1 drag to create
C-x <spc> shrink/grow
C-x C-x cycle point between four corners
C-x r
k kill text
M-w copy text
d delete text
y yank last killed text
o fill space
N insert lineno
c clear
t <str> replace with str on each line
Cmd
delete-whitespace-rectangle
string-insert-rectangle
“Registers” are used to save text, rectangles, positions, etc. for later repeating use. Register has a name made of single char (letter or number).
“Bookmarks” are similar, it record files and pos for jumping. But have long names and persist between sessions.
C-x r
<spc> R save pos
w R save window in R
f R save frameset in R
j R jump to R (can be file/buffer/pos/macro/window/frame)
----------------------------------------------------------------
s R copy region to R
+ append to R
i R insert text from R
----------------------------------------------------------------
r R copy rectangle to R
r i R insert rectangle from R
----------------------------------------------------------------
n store 0
+ incr register value by 1
----------------------------------------------------------------
m set pos as bookmark (prompt for name)
m BMark set bookmark named BMark
M BMark ..., don't overwrite
b BMark jump to BMark
l list all
C-u C-x r s R move region to R (delete from buffer)
C-u C-x r j R restore window/frame (deleting current)
C-u <num> C-x r n R save num in R
C-u <step> C-x r + R incr num in R by step
C-x C-k x R save last macro in R
Cmd
view/append/prepend-to-register R
---------------------
bookmark-save/load/write/delete/insert-location/insert
Var
register-separator
Fn
set-register
Macro differ from commands in that they are written in the Emacs command language rather than in Lisp. All macro are saved in a global “macro ring”.
When defining, “current counter (from 0)” can be used to insert a number to buffer that depend on the number of times the macro has been called. There is also a “previous counter” which is the value of the current counter had had last time it was incremented or set.
f3 start or insert counter (can abort by C-g)
f4 end or exec
C-g quit definition
C-x e end and exec
C-u f3 re-exec last, then append new keys
C-u C-u f3 append new keys to last without re-exec
[N] C-xk r exec last on each line in region[or N lines]
[N] C-u C-xka recounter to last given [N]
C-u C-xka reset counter to last value
---------------------------------------------------------------------------
C-xkk call macro at ring head
C-xkn/p cycle to next(eailer)/prev and show
C-xk d force redisplay while executing
---------------------------------------------------------------------------
C-xk n name save last, become command for M-x or key binding
C-xk b bind last with key
---------------------------------------------------------------------------
C-xke edit last defined macro
C-xk e edit given named macro
C-xk l edit last 300 keystrock lossage as macro
C-xk <spc> step by step edit while executing
Keybinding C-xk 0 to C-xk 9 and C-xk A to C-xk Z are specically reserved for our macro, so we can use like C-xk b 4 to directly bind to C-xk 4.
To save macros for later session:
insert-kbd-macro and save it to file.In later session, load the file with load-file.
insert-kbd-macro, it will also generate key binding code, so the keybinding will be available, too.When Defining Macros:
C-xki insert counter to buffer
(same as f3, but can be used outside definition)
[N] C-xka incr counter by given [N]
C-u C-xka reset counter to last value
C-xkf set format for inserting counter (default is "%d"),
if used outside definition, affect all later definitions
C-x q query(confirm) at this point when executing
C-u C-x q enter recursive editing (used both defining and executing)
Examples. Build a number list:
F3 C-a F3 . SPC
Cmd
insert-kbd-macro insert into buffer a macro definition as lisp code
Var
kmacro-ring-max
C-x d enter Dired
h desc mode
Normal dired cmds never consider fiels in hidden subdirectory.
You can select a set of files for display more flexibly by using ‘find’ util.
Dir Listing
( hide details
g/l refresh whole/subdir listing
i insert this subdir listing
[u]k kill [n] marked lines
s toggle alpha/time ordering
$ toggle hide/show this subdir
C-x q make editable (enter "WDired Mode")
-----------------------------------------------------------------------
find-
dired find by given 'find' conditioin
name-dired match filename with glob (no quote needed)
grep-dired match file content with grep regex
locate
locate-with-filter
Navigate, Search
n/p,C-n/p next/prev line
j jump to line
^ parent dir
-------------------------------------------------------------------------
>/< next/prev dir
CM-n/p next/prev dir, ignore level
CM-u/d up/down
M-G promt and go
-------------------------------------------------------------------------
M-s f C-s isearch only filename
M-s a C-s isearch in this or marked files
M-s a CS-s regex isearch
Visit
<ret>/e/f visit
o visit in other window and focus
C-o ..no focus
v view in View Mode
Flagging (for deletion)
d/u flag/unflag
<del> move up and unflag
---------------------------------------------------------------------------
#/~/. all auto-save(#)/backup(~)/numeric backup files
% & matche "dired-garbage-files-regexp"
% d filename matche regex
---------------------------------------------------------------------------
x eXpunge (delete) files marked with D
Marking. (deletion can also be done with marking)
* m, m mark
* * executables
* @ symlinks
* / dirs (except . and ..)
* s same subdir level files
* <del>, <del> move up and unmark
* !, U unmark all
* ?, M-<del> unmark by MARKCHAR
* t, t toggle all
* c change mark char
% m, * % with regex, match filename
% g with regex, match file's content
----------------------------------------------------------------------------
* n/p, M-{/} go next/prev marked
* N show total size
Operation (if prefix arg given, op on next N files; else on marked or single)
[N] w copy [relative] filepaths (separated by space)
C copy
D delete
R rename
H link
S symlink
Y relative symlink
M/G/O/T chmod/grp/owner/touch
[0] % [consider absolute path] prompt and do regex matching and...
u/l/R upper/lower/rename
C copy
H/S/Y hardlink/symlink/relative symlink
Z zip/unzip
c zip to single archive
P print
---------------------------------------------------------
A regex search in marked files (recur to subdir)
Q ..search-replace..
----------------------------------------------------------------------------
+ create dir
= diff
----------------------------------------------------------------------------
:e encrypt
:d decrypt
:s sign
:v verify
L load as elisp
B byte compile
I info
N man
Shell Commands
![&]/X[&] run command [async]
Cmd
dired-compare-directories
Var
dired-listing-switches option when listing
dired-isearch-filenames
dired-kill-when-opening-new-dired-buffer
C-t
C-t/. show thumb inline
./d show thumb in a "thumbnail buffer" (See Image Dired Thumbnail)
j jump to thumb buffer
a append this thumb
c create/edit comment and tag
d display thumb
e edit comment and tag
f mark tagged
i show image
r remove tag
t tag
x open external
D set desc
L/R rotate left/right
S slideshow
m/U mark this/unmark all
W set wallpaper
c comment
w copy filename
x delete flagged
----------------------------------------------------------------------------
tt tag
tr tag remove
----------------------------------------------------------------------------
gf
gg
gi
Provided “eXtra” Dired feature, like omitting files, Virutal Dired, Find file at point, etc.
Omitting
C-x M-o omitting boring files
* O mark omit files
Virtual Dired
Let you put any buffer with Dired-like contents (e.g the output of ls -lR) in Dired mode.
…
Shell mode is a derivative of “Comint mode”, a general-purpose mode for communicating with interactive subprocess.
[u] M-! run sh cmd [output to this buffer]
[u] M-| run sh cmd with region as input
M-& run sh cmd async
Cmd
shell subshell
term subshell with full terminal emulation
eshell run eshell (implemented entirely in Emacs)
“Abbrev Mode” will auto expands an abbrev, after your typed a punctuation. It preserve case: like ‘foo’ to ‘find out otter’, ‘Foo’ to ‘Find out otter’, and ‘FOO’ to ‘Find out otter’.
You can manually disable expand by using C-q.
Beside normal abbrev, which need defition to expand, “dynamic abbrevs (Dabbrev)” can be determined from buffer content, but expansion happens only when you request it.
C-x a
[N] g generate new abbr definition for last [N] word (or region)
[N] l same, but for major mode
[N] i g same, but inversed
[N] i l i, but for major mode
e expand manually (even when Abbrev Mode not enabled)
M-' separate prefix from following abbrev to be expaned, like:
're|cont' expanded by 'cont' abbrev to 'reconstruction'
(the | represent the point you type M-')
----------------------------------------------------------------------------
M-/ expand, "dynamically"
CM-/ complete as dynamic abbrevs
New definition will overwrite (update) old one. To remove, use negative arg.
Another expansion tools is “Hippie Expansion”, which provid variety of completion and expansions. Typically you would bind hippie-expand to M-/, replacing the adbbrev-expand.
Cmd
define-global-abbrev
define-mode-abbrev
kill-all-abbrevs
unexpand-abbrev undo last expansion
expand-region-abbrevs
list-abbrevs
edit-abbrevs
write-abbrev-file
read-abbrev-file
define-abbrevs
insert-abbrevs
hippie-expand
Var
abbrev-all-caps
abbrev-suggest-hint-threshold
abbrev-suggest-show-report
dabbrev-casee-fold-search
dabbrev-case-replace
dabbrev-abbrev-char-regexp
dabbrev-abbrev-skip-leading-regexp
Skeleton commands take num arg, negative one make it do backward. It can be used with Abbrevs, rather than key binding for every skeleton command. Like:
(define-abbrev c-mode-abbrev-table "ifst" "" 'c-if)
Auto-pairing can be done by binding the first key in pair, like " ( [, to function skeleton-pair-insert-maybe.
Fn
skeleton-pair
skeleton-pair-on-word
AutoInsert is used to auto insert text every time a empty file is visited, by putting (auto-insert-mode t) to init file, and they are defined with Skeleton Language.
Cmd
auto-insert
copyright
copyright-update
copyright-current-year
executable-insert
executable-query
time-stamp
time-stamp-toggle-active
Var
auto-insert-alist
auto-insert-query
auto-insert-prompt
time-stamp-active
time-stamp-format
time-stamp-time-zone
time-stamp-line-limit
time-stamp-start
time-stamp-end
time-stamp-count
time-stamp-inserts-lines
The Tempo package provides simple way to define templates, a template is defined as a list of items to be inserted in the buffer at point.
See doc for ‘tempo-define-template’ and comments in ‘tempo.el’.
Cmd
tempo-backward-mark
tempo-forward-mark
All those commands can be prefixed by C-x v ! to edit the shell command line before running.
[u] C-x vv [prompt for] next logical action
[u] C-x v= diff last [given] version
[u] C-x vD same, but diff entire working tree
C-x
vi register under VC
v~ visit given version in another buffer
vg show latest version of each line ("annotated")
----------------------------------------------------------------
[u/N] C-x
vl logs of this fileset [starting from revision/show N entries]
vh ..of this region
vL ..of entire repo
vbl ..of other branch
vI ..of what incoming will "pull" receive
vO ..of what outcoming will "push" sent
----------------------------------------------------------------
vu revert (undo) to last revision
vG ignore file
vd enter VC Dir Mode
----------------------------------------------------------------
vbs switch branch
vP push
v+ pull
vm merge from another branch
----------------------------------------------------------------
va visit 'ChangeLog' file
----------------------------------------------------------------
vs create tag
vr retrieve tag
Mode Line Indicator:
@@ registered but not commit ("added")
- unmodified
: modified
! conflict or removed from VC
? VCed but missing from working tree
Log Edit Mode (*vc-log* buffer)
To abort a commit, just don’t type C-cc in vc-log buffer, you can just switch buffers and do other editing.
C-c
C-f view fileset
C-d diff
C-w generate ChangeLog entreis
C-a read relevant entreis from ChangeLog files
C-c
M-n/p/s/r browser commit message history
Git Log View Mode
n/p next/prev line
a annotate this lines's revision
e edit comment (NOT all VC support this)
f visit this line's revision
d diff revision at point and next earlier revision
D changeset diff..
<ret> toggle full log
Anotate Mode
n/p annotate next/prev revision
j annotate revision indicated by this line
f same, but show in buffer
a annotate prev revision indicated by this line
d diff this line's revision with its prev revision
D same, but for all files in the changeset
l show log of this line's revision
v toggle annotation visibility
VC Dir Mode
* indicate the file has been marked as current VC fileset.
n, <spc>/p next/prev line
<tab>/S-<tab> next/prev dir line
<ret>,f visit file
o visit in other window
q quit
x hide up-to-date or ignored
S search on fileset
Q regexp query replace on fileset
M-s a C-s isearch on fileset
M-s a CM-s regexp isearch on fileset
d delete marked (NOT from VC, mostly useful for unregistered)
----------------------------------------------------------------
m mark in VC fileset
M mark all with same status
%m, *% mark by regexp
*r mark registered
u unmark
U unmark all with same status
G ignore
----------------------------------------------------------------
bc new branch
bl show logs for given branch
bs switch to branch
Cmd
vc-ediff like v=, but using ediff
vc-log-search search log history by matching message with pattern
vc-create-branch
vc-delete/rename-file
Var
[vc-[BACKEND-]]diff-switches
[vc-BACKEND-]annotate-switches
vc-annotate-background-mode
vc-revert-show-diff
vc-directory-exclusion-list
Cmd
diff
diff-backup
diff-buffer-with-file
ediff
ediff-region-linewise/wordwise
[N] compare-windows compare two windows starts from each's point
Var
diff-switches options for `diff` util
Diff Mode
n/p,N/P next/prev hunk,file start
M-k/K kill hunk/file part
C-c
a apply
b break down to see refined changes
c go to source
e start ediff
n narrow view to this hunk
r reverse direction
s splite hunk to two
d convert buffer to "context diff format"
u to "unified diff format"
l refresh hunk
w refresh hunk, disregard whitespace
C-x 4 A generate ChangeLog entry
During emerge session, you must NOT try to edit the A and B buffers yourself. You can have any number of merges going at once – just don’t use any one buffer as inputu to more than one merge at once.
emerge-files/buffers[-with-ancestor]
Submodes of Emerge
A directory is a “project” if it’s a VSC repo, or EDE (UNMAINTAINED). this list can be extended in future). The “current project” is determined by ‘default-directory’.
To create new projects, C-x pp then ... (Choose a dir).
C-x p p switch project
C-x p in project:
f visit a file
g find files whose content match regex
r regexp replace
----------------------------------------------------------------
b goto buffer
C-b list buffer
k kill all buffers
----------------------------------------------------------------
v run vc-dir
d find dir, into Dired
D Dired to project root
----------------------------------------------------------------
s run shell
e run eshell
c run compile
! run a shell cmd
& run a async shell cmd
Cmd
project-search same as C-x pg, but stop at first match
Var
project-kill-buffer-*
conditions which to kill when C-x p k
display-buffer-list display before killing
project-list-file where to save project info
project-forget-project
Offers find major definitions in file, also useful in text modes where it treas chapter, section, etc., as definition.
M-g i
Var
imenu-*
auto-rescan[-maxout]
max-index-time
sort-function
Which function mode display current function name in mode bar.
Xref is a unified interface for tag/symbol finding, it needs a backend to work with:
“Tags table” can be produced by etags util, Emacs use the tags table as one of the backends for xref.
List of language and what’s considered as tags in their construction:
lang tags
---------------------------------------------------------------------------
C function, typedef, struct, #define...
Lisp defun, defvar, defconst...
Go package, function, type
HTML title, headers, anchors, id
PHP function, class, define, var
Python def, class
Rust fn, enum, struct, macro_rules!
Definition, Reference
M-. find definition
C-x 4/5 . ..in other window/frame
M-? find reference
CM-. find identifiers by pattern
----------------------------------------------------------------
M-, go back
CM-, go forward
----------------------------------------------------------------
CM-i perform completion
In XREF Mode (*ref* buffer):
n/. goto next and display in other window
p/, goto prev..
N/P goto next/prev group
C-o show pos in other window ??
----------------------------------------------------------------
r query replace this identifier
g refresh
M-, quit, jump to prev xref stack location
CM-, go forward again (useful if M-, go back too far)
q quit
Cmd
xref-*
find-definitions-at-mouse
etags-mode
find-references-and-replace
visit-tags-table
list-tags
tags-* in all file
search
tags-query-replace
next-file
fileloop-continue restart tags search/replace cmd at this point
Var
xref-*
auto-jump-to-first-definition
tags-*
table-list
case-fold-search case sensitivity for find-tag
apropos-additional-actions
The default compile command is make -k, if you’ve done M-x compile before, the command that you specified is auto stored in compile-command. A file can also specify a file-local value.
You can control the env value passed to compilatioin command with compilation-environmentt.
To run two compilations at once, rename the first *compilation* buffer.
C-x ` next error
M-gn/p next/prev error
NOTE that M-gn, M-gp don’t just act on compilation error, it works for like error produced by other command, like occur.
Compilation Mode (compilation buffer)
[u] g [edit and] recompile
[n] M-n/p next/prev error, without find source line
[n] n/p ..find and hilight source, stay focused
[n] M-{/} next/prev file
C-cc goto error
C-cf enable "next errorr follow" mode
C-ck kill compilation
Cmd
compile run async a compiler
recompile run last compile command
kill-compilation kill the running subproc
Var
compilation-*
scroll-output
always-kill
skip-threshold skip when nexting-error
hidden-output hide too verbose output
error-regexp-alist what's "error" and where's "locus"
On-the-fly syntax checking. It’s only useful when a backend provide the buffer-checking service – this is done via flymake-diagnostic-functions. When LSP is enabled (via Eglot or lsp-mode), flymake will use that as backend.
Flymake Mode
Auto check when buffer saved or more than 0.5 seconds after buffer changes.
Mode Line Status
Wait backends haven't respond for last request
! all backends has disabled
? no applicable backends
Extending
By place a func in hook flymake-diagnostic-functions.
Cmd
flymake-*
mode enable mode
start manually invoke checking
goto-next/prev-error
show-buffer/project-diagnostics show errors in one place
----------------------------------------------------------------
reporting/running/disabled-backends
switch-to-log-buffer
Var
flymake-*
mode-line-lighter/[counter-]format
no-changes-timeout
start-on-flymake-mode
start-on-save-buffer
fringe-indicator-position
wrap-around
Feature
Other LSP Clients
Setting Up
eglot-server-programs value, you just need to install its lsp server.M-x eglot.For eglot, emacs’s project is what “workspace” for lsp server, and consdiers a buffer’s file who turned eglot on as belonging to a project (even a project of only itself).
In some case, additional info of project layout is needed for lsp server when starting it, eglot-workspace-configuration is for that.
For each buffer in which eglot is active, eglot is “managing” the file visited by it, and communicate with lsp server about its content updates. Variable eglot-managed-p tells whether current buffer is being managed or not.
Cmd
eglot figure out lsp server and start it
eglot-*
show-workspace-configuration
reconnect
[u] shutdown [don't kill buffers used for talking to server]
[u] shutdown-all same, but all server
----------------------------------------------------------------
rename rename symbol
format reformat region by server's rules
format-buffer
code-actions
code-action-orgnize-imports
code-action-quickfix
code-action-extract
code-action-inline
code-action-rewrite
----------------------------------------------------------------
inlay-hints-mode
----------------------------------------------------------------
events-buffer
stderr-buffer
forget-pending-continuations
signal-didChangConfiguration update server conf
clear-status clear JSONRPC error of this buffer
Var
eglot-*
autoreconnect
connect-timeout
sync-connect
events-buffer-size
autoshutdown
confirm-server-initiated-edits
ignored-server-capabilities
extend-to-xref jump definition even outside (liek /lib)
mode-map
----------------------------------------------------------------
server-programs
strict-mode
server-initialized-hook
connect-hook
managed-mode-hook
stay-out-of
reporot-progress
Config Exmpale
(define-key eglot-mode-map (kbd "C-c r") 'eglot-rename)
(add-hook 'foo-mode-hook 'eglot-ensure)
(with-eval-after-load 'eglot
(add-to-list 'eglot-server-programs
'(foo-mode . ("fools" "--stdio"))))
In LSP spec, project-specific settings is called “workspace configuration”, to do this, use the director-local variable eglot-workspace-configuration in ‘.dir-locals.el’ file. The value of it is a plist:
(:SERVER1 PLIST1 :SERVER2 PLIST2 ...)
The settings take effect when establishing connection, or eglot-signal-didChangeConfiguration.
Here is an example, demo using two lsp server pylsp and gopls in one project, notice how alists are used for associating Emacs mode names with alists associating variable names with variable values. Then notice how plists are used inside the value of ‘eglot-workspace-configuration’:
((nil
. ((eglot-workspace-configuration
. (:pylsp (:plugins (:jedi_completion (:include_params t
:fuzzy t)
:pylint (:enabled :json-false)))
:gopls (:usePlaceholders t)))))
(python-base-mode . ((indent-tabs-mode . nil)))
(go-mode . ((indent-tabs-mode . t))))
Apply to all porject the server is for. One way to pass cmdline options is to use the eglot-server-programs:
(with-eval-after-load 'eglot
(add-to-list 'eglot-server-programs
`(c++-mode . ("clangd" "--compile-commands-dir=/tmp"))))
Another way is to give all your projects a cerntain default config:
(setq-default eglot-workspace-configuration
'(:pylsp (:plugins (:jedi_completion (:include_params t
:fuzzy t)
:pylint (:enabled :json-false)))
:gopls (:usePlaceholders t)))
JSONRPC In ELisp
JSON are represented by the following elisp construct:
JSON ELisp
----------------------------------------------------------------
true t
false :json-false
null nil
{} eglot-{}
array vector
For example, the plist:
(:pylsp (:plugins (:jedi_completion (:include_params t
:fuzzy t
:cache_for ["pandas" "numpy"])
:pylint (:enabled :json-false)))
:gopls (:usePlaceholders t))
is serialized by Eglot to JSON:
{
"pylsp": {
"plugins": {
"jedi_completion": {
"include_params": true,
"fuzzy": true,
"cache_for": [ "pandas", "numpy" ]
},
"pylint": {
"enabled": false
}
}
},
"gopls": {
"usePlaceholders": true
}
}
Try first using eglot-events/stderr-buffer.
If eglot is working but slow, trying customize eglot-events-buffer-size to 0; else you might try advanced server
C-x
C-ab set breakpoint
C-c
C-l list source
C-s step line
C-n next
C-i stepi
C-p print
C-r continue execution
C-d del breakpoint
C-t temp breakpoint
C-< stack up
C-> stack down
C-u run until this line
C-f run until frame return
C-aj jump
Cmd
gdb
gud-gdb
perldb
jdb
pdb
guiler
dbx
xdb
~/.emacs
~/.emacs.el
~/.emacs.d/init.el
~/.config/emacs/init.el
Emacs will set user-emacs-directory to the directory it decides to use.
To know what font is under cursor, use “describe-face”.
To eval expr: either with M-:; or inside the “scratch buffer” (C-j to run).
To save customization in seperate files:
;; in init file
(setq custom-file "~/.config/custom.el")
(load custom-file)
Custom Mode
C-cc "Set for Current Session"
C-xs "Save for Future Sessions"
Cmd
customize[-*]
option
face
icon
----------------------------------------------------------------
rouge all user variables modified outside customize
group
apropos
browser
changed
saved
unsaved
Var
custom-
search-field show search box?
buffer-done-kill kill custom buf when done?
Installed pkgs are put in package-user-dir (default is ~/.emacs.d/elph/), in addititon, Emacs looks for installed pkgs inside package-directory-list, these dirs are meant to make pkg system-wide.
Package Status
available
avail-obso obsolote (has new version), hidden by default
built-in
dependency
disabled by 'package-load-list'
external not from dir given by 'package-user-dir'
held
incompat cannot be installed (like depend on uninstallable pkgs)
new newly available on the pkg archive after last listing
obsolete newer version installed
Package Menu Mode
i/d mark for install/delete
~ mark all obsolete for delete
w open pkg's website
u unmark all
U mark all upgradable
g, r refresh list
H hide
( toggle showing old version pkg
/ filter by:
a/d/k/N archive/desc/keyword/name or desc/
n/s/u/v/m/u name/status/version/marked/upgradable/
// clear filter
x execute
Cmd
list-packages enter "Package Menu Mode"
package-*
install by default, don't consider builtin pkgs
install-file manually install from file
upgrade[-all] never consider builtin
recompile[-all]
Var
package-*
archives "repo"
archive-priorities
menu-hide-low-priority
import-keyring
load-list
pinned-packages
install-upgrade-built-in
A macro to setup pkg customization in init file in a declarative way, and keep emacs startup fast. It’s NOT a pkg manager, and NOT intended to replace the customize cmd, but to work together with them.
(use-package color-moccur ;; load color-moccur
;; add dir to load-path for manual installation
:load-path "site-lisp/org/lisp/"
;; eval before loading
:init
(setq foo-variable t)
;; autoloads triggered by these two commands
;; (when one of these cmds are used,
;; color-moccur will be loaded.
:commands (isearch-moccur isearch-all)
:bind (("M-s O" . moccur) ;; bind keys globally (before :map)
:map isearch-mode-map ;; and locally (using :map)
("M-o" . isearch-moccur)
("M-O" . isearch-moccur-all))
:config ;; eval after loaded
(use-package moccur-edit))
Package Loading
By default, use-package load packages when evaluating the expression in your init files. You should try to set things up so that packages are only loaded when they are actually needed (a.k.a. “autoloading”).
When installing packages manually, you must make sure its libraries are available on your load-path.
:defer [N] DONT load [until N seconds] (rarely used alone). redundant
if there already using autoloading keywords, should use in
cases when use-package isn't creating any autoloads for you.
:demand force load now (but :defer is higher).
---------------------------------------------------------------------------
:if arg if arg evaluated to non-nil, then load
:when arg alias to :if
:unless arg :if (not arg)
---------------------------------------------------------------------------
:after (:all foo bar) after both foo and bar
:after (:any foo bar) after any of foo or bar
:after (:all (:any foo bar) (:any baz quux)) mixing
---------------------------------------------------------------------------
:requires (foo bar) dont load if missing dependencies
---------------------------------------------------------------------------
:load-path add dir to load path
:commands autoload an interactive command
:autoload autoload a non-interactive function
Config Using Lisp
When possible, avoid the below three, prefer autoloading keywords such as ‘:bind’, ‘:hook’, ‘:mode’.
:preface eval before anything, except :disabled and :ensure
:init eval just before package is loaded
:config eval after package loaded
Key Binding
:bind globally. take either signle cons or list of conses. each
con has the form '(KEY . CMD)'.
:map bind within local keymap (only exists after pkg loaded)
:bind-keymap same as bind, but must be keymaps defined in pkg, not
interactive functions.
:repeat-map bind to repeat-mode map
Hooks, Modes, Interpreters, Magic Handler
:hook HOOKS add functions to one or more hooks
:mode deferred binding within `auto-mode-alist`
:interpreter ... `interpreter-mode-alist`
:magic
:magic-fallback
User Options
:custom
:custom-faces
Hiding Minor Mode
:diminish
:delight
Examples
;; Rebind fill-paragraph(M-q) to unfill-toggle.
(use-package unfill
:bind ([remap fill-paragraph] . unfill-toggle))
;; These two are equivalent, the first one save us some typing.
;; 1.
(use-package ace-jump-mode
:bind ("C-." . ace-jump-mode))
;; 2.
(use-package ace-jump-mode
:defer t
:init
(autoload 'ace-jump-mode "ace-jump-mode" nil t)
(bind-key "C-." 'ace-jump-mode))
;; :map binding to local.
(use-package term
:bind (("C-c t" . term) ;; any binding before the first :map are global
:map term-mode-map
("M-p" . term-send-up)
("M-n" . term-send-down)
:map term-raw-map
("M-o" . other-window)
("M-p" . term-send-up)
("M-n" . term-send-down))
)
;; :bind-keymap
(use-package foo
:bind-keymap ("C-c p" . foo-command-map))
;; :custom
(use-package comint
:defer t
:custom
(comint-buffer-maximum-size 20000 "Increase comint buffer size.")
(comint-prompt-read-only t "Make the prompt read only."))
;; :hook
;; NOTE that '-hook' is not needed when using :hook.
;; The following examples are all equivalent.
(use-package company
:hook (prog-mode text-mode))
(use-package company
:hook ((prog-mode text-mode) . company-mode))
(use-package company
:hook ((prog-mode . company-mode)
(text-mode . company-mode)))
(use-package company
:commands company-mode
:init
(add-hook 'prog-mode-hook #'company-mode)
(add-hook 'text-mode-hook #'company-mode))
Note that the values customized using :custom are not saved in the standard Emacs custom-file. You should therefore set each user option using either the :custom keyword or M-x customize-option command; the latter will save customized values in the Emacs custom-file. Do not use both for the same variable, as this risks having conflicting values in your use-package declaration and your custom-file, which can lead to problems that are both tricky and tedious to debug.
Cmd
describe-personal-keybindings
Var
use-package-always-defer
Local, Global
“Customizable variable” are also called “user options”, and can be set with set-variable cmd, and it can set only customizable var locally. To set any variable, use setq. To set customizable var in init file, use setopt.
To make customization save into other file other than your init files:
(setq custom-file "~/.config/emacs-custom.el")
(load custom-file)
“Local” variables only has effect to a buffer, some variable cannot be local, they are always local to each display instead. To set the global value of a variable, use setq-default. Lisp can use default-value to see a variable’s default value.
File Variable
File and directory can also specify value when editing the file, file variable will override directory level variable, here is how to set value in file’s first line:
-*- mode: MODENAME; VAR: VALUE; ... -*-
Or by “Local Variables:” to give a variable list:
/* Local Variables: */
/* mode: c */
/* comment-column: 0 */
/* End: */
The following keywords are not really variables, setting them in other context has no meaning:
mode enable major mode
eval eval Lisp expr (can be dangerous!)
coding use coding system
unibyte load or compile a Lisp file in unibyte mode (if 't')
Per-Directory Variable
Once way is in file .dir-locals.el or .dir-locals-2.el (useful if the former is VCed anad we still want personal customization). These file should hold a list, which maps major mode names to alists:
((nil . ((indent-tabs-mode . t) ;; nil: for any file in dir
(fill-column . 80)
(mode . auto-fill)))
(c-mode . ((c-file-style . "BSD")
(subdirs . nil))) ;; only apply to current, not subdir
("src/imported"
. ((nil . ((change-log-default-name
. "ChangeLog.local"))))))
Another way is to combine the usage of dir-locals-set-class-variables and dir-locals-set-directory-class, these function calls typically go in your init file. This way is useful if you cannot write to directory.
(dir-locals-set-class-variables 'unwritable-directory
'((nil . ((some-useful-setting . value)))))
(dir-locals-set-directory-class
"/usr/include/" 'unwritable-directory)
Per-Connection Variable
Cmd
make-local-variable read and make a var local
kill-local-variable
make-variable-buffer-local mark a var to be local when set
----------------------------------------------------------------
add/delete-file-local-variable-prop-line
copy-dir-locals-to-file-locals-prop-line
add/delete-file-local-variable to/from list
copy-dir-locals-to-file-locals
add/delete-dir-local-variable
copy-file-locals-to-dir-locals
----------------------------------------------------------------
enable-local-variables/eval
safe-local-eval-forms
Menu bar and mouse button events are also represented as fictitious “prefix key”.
Bindings between keys and command functions are stored in “keymaps”, the “global” keymap is always in effect. Some prefix keymaps are stored in variables:
ctl-x-map for chars that follow C-x
help-map ..C-h
esc-map ..<esc>
ctl-x-4-map ..C-x 4
mode-specific-map ..C-c
project-prefix-map ..C-x p
C-c prefix and F5~F9 is reserved for user key bindings. Major modes providing their own key bindings in “local keymaps”. Minor modes do so, too; and override major mode’s when activated. A local keymap can redefine a key as a prefix key by defining it as a prefix keymap. The minibuffer has its own set of local keymaps.
Keys Rebinding
You can rebind keys with commands:
keymap-global/local-set RET KEY CMD
keymap-globa/local-unset RET KEY
If you want to retract changes, you need to redefine it to standard key. To find that, go to Fundamental mode in a fresh Emacs and use C-h c.
Another way is to put the rebinding in init files:
;; the most easy way:
(keymap-global-set "C-z" 'shell)
;; lower level functions:
(global-set-key (kbd "C-z") 'shell)
If you want to define your own prefix key as new keymap:
;; change to specific keymap:
(define-prefix-command 'my-prefix-map)
(global-set-key (kbd "`") 'my-prefix-map)
(keymap-set 'my-refix-map (kbd "C-o o") 'my-coo-cmd)
;; to remove, use keymap-unset
Since mode’s keymaps are not constructed until it has been loaded, you must delay binding by putting in mode hook:
(add-hook 'texinfo-mode-hook
(lambda ()
(keymap-set texinfo-mode-map "C-c p"
'backward-paragraph)
(keymap-set texinfo-mode-map "C-c n"
'forward-paragraph)
(keymap-set texinfo-mode-map "C-c C-x x" nil)))
“Hooks” is a lisp var which holds list of functions, to be called on some occasion:
If the hook variable is buffer-local, the buffer-local variable will be used instead of the global variable. However, if the bufefr-local variable contains the element t, the global hook variable will be run as well.
It’s better to use function when add-hook, and design your hook functions so that the order in which they are executed does NOT matter.
Call remove-hook to remove your functions from hook.
(setq my-c-style
'((c-comment-only-line-offset . 4)
(c-cleanup-list . (scope-operator
empty-defun-braces
defun-close-semi))))
(add-hook 'c-mode-common-hook
(lambda () (c-add-style "my-style" my-c-style t)))
“Diary” records events asoociated a date, stored in dairy-file. Every events begin with a date. Then can be viewed in calendar.
Diary files can contain directives to include the contents of other files with diary-list-entries-hook.
[u] calendar calendar mode
Calendar Mode
[n] C-f/b,n/p,{/},[/] move by day,week
M-{/} ..month
C-x [/] ..year
C-a/e move to start/end of week
M-a/e,</> ..month,year
g d/D/w goto date/day of year/week
o ..month
. ..today
---------------------------------------------------------------------------
</> scroll one month
C-v/M-v ..three..
---------------------------------------------------------------------------
M-= count days of region (marks keys are available)
p d show day in year
C-cl redraw
---------------------------------------------------------------------------
h show this day's holidy
x/u mark/unmark holidays
a list all holidays
S/M sunrise/moon phase
---------------------------------------------------------------------------
p/g ... show/from this day in calendar system of...
o others
c ISO commercial
j Julian
a Astronomical
h Hebrew
i Islamic
f French
b Baha'i
C Chinese
k Coptic
e Ethiopic
p Persian
m Maya
---------------------------------------------------------------------------
H m/y export one-month/year to html file
t ... export ... to latex buffer
m/M/d
w 1/2/3/4/W
f w/W
y/Y
fy
Diary
d/s show this day/all diary
m/u mark/unmark all days with diary
---------------------------------------------------------------------------
i ... insert entry ...
d/w/m/y for day/dow/dom/doy
a/b/c as anniversary/block (of region)/cyclic
Clocking In Out
timeclock-*
in/out
change
workday-remaining
when-to-leave
<f2>/C-x 6
2 enter 2C mode
s split current buffer to 2C
b 2C with other buffer
<ret> insert newline in both side
1 merge (from right to left) and done
d dissociate
Cmd
table-*
insert
[un]recognize[-region/table/cell] [de]activate as table
forward/backward-cell
span/split-cell[-vertically/horizontally]
heighten/shorten/widen/narrow-cell
justify
insert/delete-row/column
capture/release convert from/to text
query-dimension
insert-sequence
generate-source for html/latex/wiki...
Set Inserting Direction
C-c
< > ^ . left right up down
` ' / \ nw ne sw se
(u) C-c
` ' / \ wnw ene wsw ese
Move Cursor
C-
n/p
e/f
b
C-c
C-f/b
-- tab stops -------------------------------------------------------
CM-i
<tab>
C-c <tab>
Deletion
C-d
C-c C-d
<del>
C-k
C-o
Rectangle
C-c
C-k/w/y/x/r
C-x rr copy
C-x u undo
Drawing with mouse.
mouse2 popup menu
[S] mouse1 draw [straight or normalized]
[S] mouse3 erase char[rect]
Features
Key Bindings
C-x *c start
Calculator Mode
? help, repeat it for more
+ - * / ^ %
^ reciprocal
n change sign
Cmd
calc
print-buffer/region print with page headings containing filename
lpr-buffer/region ..without...
---------------------------------------------------------------------------
ps-print-buffer/region[-with-faces] convert to PostScript
ps-spoll-buffer/region[-with-faces] generate and spool a PS image
ps-despoll send spooled PS to printer
handwrite
---------------------------------------------------------------------------
htmlfontify
Var (lpr-*)
With “printing” package:
(require 'printing)
you can also preview PostScript files before printing, with pr-interface.
s search new word
m search by matching pattern
n/p next/prev link
D select default dictionary
M select default strategy
C-c
C-c toggle display
C-x toggle hex display
sw set fit window
sp set fit percent
ss set scale
s0 reset to initial
so reset to original
n/p next/prev file
m/u mark/unmark in dired buffer
w copy file name
---------------------------------------
i
+/- incr/decr image size
r/h/v rotate, horizon/vert flip
o save to file
c crop
x cut rectangle
---------------------------------------
<RET> toggle animate
f/b next/prev frame
F goto frame
a+/a-/ar/a0 incr/decr/reverse/reset speed
Var
image-*
auto-resize[-on-window-resize]
(Superceded by org-mode’s Capture.)
By default, notes are captured in ~/.emacs.d/notes. You can save notes to variety of backends, by setting remember-handler-functions, like text file, diary file, mailbox, or integrate with org-mode capture.