356 lines
11 KiB
Org Mode
356 lines
11 KiB
Org Mode
#+TITLE: Emacs init
|
|
#+STYLE: <link href="http://ryuslash.ninth.su/test2.css" rel="stylesheet">
|
|
#+OPTIONS: author:nil
|
|
|
|
* Startup
|
|
|
|
Startup requires a bit of customization to handle all my
|
|
customizations.
|
|
|
|
** Load paths
|
|
|
|
I have two versions of Emacs installed on my main computer. I have
|
|
a daily build of Emacs's ~trunk~ (or currently ~emacs-24~) branch from
|
|
bazaar and I have the official ~emacs~ package from archlinux
|
|
installed. I keep that second one around so that the occasional
|
|
emacs package that I install using ~pacman~ will recognize it as a
|
|
dependency and so that if there has been some horrible mistake in
|
|
the ~trunk~ branch I still have a stable version to fall back
|
|
on[fn:1].
|
|
|
|
*** Package initialization
|
|
|
|
In order for packages installed through ELPA to be included
|
|
without having to mess with load paths and such, use:
|
|
|
|
#+NAME: package-initialize
|
|
#+BEGIN_SRC emacs-lisp
|
|
(package-initialize)
|
|
#+END_SRC
|
|
|
|
*** site-lisp
|
|
|
|
Because of the setup I wrote about I need to have both the
|
|
self-built ~site-lisp~ directory in my load path *and* the "official"
|
|
one.
|
|
|
|
#+NAME: load-site-lisps
|
|
#+BEGIN_SRC emacs-lisp
|
|
(mapc #'oni:add-all-to-load-path
|
|
'("/usr/share/emacs/site-lisp"
|
|
"/usr/local/emacs/share/emacs/site-lisp"))
|
|
#+END_SRC
|
|
|
|
The =oni:add-all-to-load-path= function just binds =default-directory=
|
|
to the given directory and calls
|
|
=normal-top-level-add-subdirs-to-load-path= to add it and all its
|
|
subdirectories to the load path.
|
|
|
|
#+NAME: add-all
|
|
#+BEGIN_SRC emacs-lisp
|
|
(defun oni:add-all-to-load-path (dir)
|
|
(add-to-list 'load-path dir)
|
|
(let ((default-directory dir))
|
|
(normal-top-level-add-subdirs-to-load-path)))
|
|
#+END_SRC
|
|
|
|
**** TODO Add load paths in different order depending on version
|
|
|
|
The officially installed version should load
|
|
=/usr/share/emacs/site-lisp= before
|
|
=/usr/local/emacs/share/emacs/sit-lisp= and the bzr version should
|
|
do the reversed.
|
|
|
|
*** Projects
|
|
|
|
Then there are some projects I'm working on, which I use on a
|
|
daily basis, these should also be added so I don't have to
|
|
constantly remove and re-install them through ~package.el~ when
|
|
working on them. And there is the ~load-defs.el~ in my personal
|
|
~site-lisp~ directory of course.
|
|
|
|
#+NAME: load-projects
|
|
#+BEGIN_SRC emacs-lisp
|
|
(mapc #'oni:add-to-load-path-maybe-load-defs
|
|
'("~/projects/emacs/dispass.el" "~/var/src/emacs/mode-icons"
|
|
"~/.emacs.d/site-lisp"))
|
|
#+END_SRC
|
|
|
|
The =oni:add-to-load-path-maybe-load-defs= function just adds the
|
|
given directory to the load path and then looks for a file named
|
|
~loaddefs.el~ within that directory, if it exists it loads it.
|
|
|
|
#+NAME: add-with-loaddefs
|
|
#+BEGIN_SRC emacs-lisp
|
|
(defun oni:add-to-load-path-maybe-load-defs (dir)
|
|
(add-to-list 'load-path dir)
|
|
(let ((loaddefs (concat dir "/loaddefs.el")))
|
|
(when (file-exists-p loaddefs)
|
|
(load loaddefs))))
|
|
#+END_SRC
|
|
|
|
*** Themes
|
|
|
|
And, finally, I'm also working on a color theme for emacs, this
|
|
should be added to ~custom-theme-load-path~. By using =mapc= here as
|
|
well I'm keeping in mind that this isn't the first theme I've
|
|
worked on and it might not be the last either.
|
|
|
|
#+NAME: load-themes
|
|
#+BEGIN_SRC emacs-lisp
|
|
(mapc #'oni:add-to-custom-theme-load-path
|
|
'("~/projects/emacs/yoshi-theme"))
|
|
#+END_SRC
|
|
|
|
The =oni:add-to-custom-theme-load-path= just adds the given
|
|
directory to the cutom theme load path.
|
|
|
|
#+NAME: add-themes
|
|
#+BEGIN_SRC emacs-lisp
|
|
(defun oni:add-to-custom-theme-load-path (dir)
|
|
(add-to-list 'custom-theme-load-path dir))
|
|
#+END_SRC
|
|
|
|
*** Evaluation
|
|
|
|
Because during byte-compilation certain parts loaded so far might
|
|
also be required I put it in an =eval-and-compile= form, so that all
|
|
components are loaded with ~emacs -Q~ as well. Without this
|
|
compilation might fail at certain points.
|
|
|
|
#+BEGIN_SRC emacs-lisp :tangle init2.el :noweb yes
|
|
(eval-and-compile
|
|
<<add-all>>
|
|
|
|
<<add-with-loaddefs>>
|
|
|
|
<<add-themes>>
|
|
|
|
<<package-initialize>>
|
|
<<load-site-lisps>>
|
|
<<load-projects>>
|
|
<<load-themes>>)
|
|
#+END_SRC
|
|
|
|
** Modules
|
|
|
|
While I try to use =eval-after-load= and =autoload= as much as
|
|
possible, some things require direct =require='ing to be of use.
|
|
|
|
*** Require
|
|
|
|
- ~auto-complete-config~ :: This sets up some default settings to
|
|
make ~auto-complete~ work for most[fn:2] of the modes it
|
|
supports.
|
|
|
|
#+BEGIN_SRC emacs-lisp :tangle init2.el
|
|
(require 'auto-complete-config)
|
|
#+END_SRC
|
|
|
|
- ~geiser-install~ :: Sets up geiser autoloads and such.
|
|
|
|
#+BEGIN_SRC emacs-lisp :tangle init2.el
|
|
(require 'geiser-install)
|
|
#+END_SRC
|
|
|
|
- ~uniquify~ :: Provides more helpful buffer name uniquification.
|
|
The default of using ~buffer-name<2>~ is boring and
|
|
uninformative, ~uniquify~ fixes this.
|
|
|
|
#+BEGIN_SRC emacs-lisp :tangle init2.el
|
|
(require 'uniquify)
|
|
#+END_SRC
|
|
|
|
- ~ext~ :: Functions from external sources.
|
|
- ~oni~ :: Functions written personally.
|
|
|
|
#+BEGIN_SRC emacs-lisp :tangle init2.el
|
|
(require 'ext)
|
|
(require 'oni)
|
|
#+END_SRC
|
|
|
|
*** Autoload
|
|
|
|
These might not be used at all in a session, so they should only
|
|
be loaded when necessary.
|
|
|
|
=define-slime-contrib= was used by some module that didn't autoload
|
|
or require it[fn:3]. But since I don't use whichever model it
|
|
was every day, it is of no use to load it every single time.
|
|
|
|
#+BEGIN_SRC emacs-lisp :tangle init2.el
|
|
(autoload 'define-slime-contrib "slime")
|
|
#+END_SRC
|
|
|
|
I installed ~global~ with ~pacman~, but this doesn't add anything to
|
|
any =loaddefs.el=, so doesn't create any autoloads.
|
|
|
|
#+BEGIN_SRC emacs-lisp :tangle init2.el
|
|
(autoload 'gtags-mode "gtags" nil t)
|
|
#+END_SRC
|
|
|
|
~jabber.el~ does create ~jabber-autoloads.el~, but I only ever start
|
|
using it through ~jabber-connect~, so anything else isn't really
|
|
necessary.
|
|
|
|
#+BEGIN_SRC emacs-lisp :tangle init2.el
|
|
(autoload 'jabber-connect "jabber" nil t)
|
|
#+END_SRC
|
|
|
|
I used to work a bit on ~php-mode~, but that was a while ago, so
|
|
it's still in my ~site-lisp~ directory.
|
|
|
|
#+BEGIN_SRC emacs-lisp :tangle init2.el
|
|
(autoload 'php-mode "php-mode" nil t)
|
|
#+END_SRC
|
|
|
|
The same that goes for ~gtags.el~ also goes for ~po-mode.el~.
|
|
|
|
#+BEGIN_SRC emacs-lisp :tangle init2.el
|
|
(autoload 'po-mode "po-mode" nil t)
|
|
#+END_SRC
|
|
|
|
And the same that went for ~php-mode~ also goes for ~pony-mode~,
|
|
except I'm still working on it and I was too lazy to put it with
|
|
my other projects. I should still do that.
|
|
|
|
#+BEGIN_SRC emacs-lisp :tangle init2.el
|
|
(autoload 'pony-mode "pony-mode" nil t)
|
|
#+END_SRC
|
|
|
|
~sawfish.el~ has the same problem that ~gtags.el~ and ~po-mode.el~ have.
|
|
|
|
#+BEGIN_SRC emacs-lisp :tangle init2.el
|
|
(autoload 'sawfish-mode "sawfish" nil t)
|
|
#+END_SRC
|
|
|
|
I use =server-running-p= to check whether or not I should start a
|
|
new server, but this function isn't autoloaded by default.
|
|
|
|
#+BEGIN_SRC emacs-lisp :tangle init2.el
|
|
(autoload 'server-running-p "server")
|
|
#+END_SRC
|
|
|
|
I was starting to try ~slime-js~ to make JavaScript programming
|
|
more interesting, but I haven't gotten around to trying it out
|
|
fully, yet. It shares issues with ~gtags.el~, ~po-mode.el~ and
|
|
~sawfish.el~.
|
|
|
|
#+BEGIN_SRC emacs-lisp :tangle init2.el
|
|
(autoload 'slime-js-minor-mode "slime-js" nil t)
|
|
#+END_SRC
|
|
|
|
I found ~xmodmap-mode~ on the [[http://emacswiki.org][EmacsWiki]] some time ago, it was
|
|
simple and a good example of how to use ~define-generic-mode~, but
|
|
since it's not really my project and it's really small it just
|
|
sits in my ~site-lisp~ directory.
|
|
|
|
#+BEGIN_SRC emacs-lisp :tangle init2.el
|
|
(autoload 'xmodmap-mode "xmodmap-mode" nil t)
|
|
#+END_SRC
|
|
|
|
~w3m~ also has a setup module like ~geiser-install~, but since I only
|
|
ever use these two functions to start it, there is no real need
|
|
for anything else.
|
|
|
|
#+BEGIN_SRC emacs-lisp :tangle init2.el
|
|
(autoload 'w3m-bookmark-view "w3m" nil t)
|
|
(autoload 'w3m-goto-url "w3m" nil t)
|
|
#+END_SRC
|
|
|
|
* Aliases
|
|
|
|
I've never had any trouble with accidentally pressing ~y~ while being
|
|
asked a question, so I've never had any reason to prefer ~yes-or-no-p~
|
|
over ~y-or-n-p~.
|
|
|
|
#+begin_src emacs-lisp :tangle init2.el
|
|
(defalias 'yes-or-no-p 'y-or-n-p)
|
|
#+end_src
|
|
|
|
~ibuffer~ is a drop-in replacement for ~list-buffers~, but with more
|
|
features.
|
|
|
|
#+begin_src emacs-lisp :tangle init2.el
|
|
(defalias 'list-buffers 'ibuffer)
|
|
#+end_src
|
|
|
|
I don't know if replacing ~dabbrev-expand~ with ~hippie-expand~, but at
|
|
least ~hippie-expand~ doesn't use ~dabbrev-expand~, and I haven't
|
|
noticed anything wrong so far, and ~hippie-expand~ does so much more
|
|
than ~dabbrev-expand~.
|
|
|
|
#+BEGIN_SRC emacs-lisp :tangle init2.el
|
|
(defalias 'dabbrev-expand 'hippie-expand)
|
|
#+END_SRC
|
|
|
|
* Faces
|
|
|
|
Define a face to how the summary line of git commits as the first
|
|
headline of an ~org-mode~ file.
|
|
|
|
#+begin_src emacs-lisp :tangle init2.el
|
|
(defface git-commit-summary-face
|
|
'((t (:inherit org-level-1)))
|
|
"Face for the git title line."
|
|
:group 'local)
|
|
#+end_src
|
|
|
|
Define a face to show characters that have been placed beyond the
|
|
maximum length of a summary line.
|
|
|
|
#+begin_src emacs-lisp :tangle init2.el
|
|
(defface git-commit-overlong-summary-face
|
|
'((t (:background "#873732")))
|
|
"Face for commit titles that are too long."
|
|
:group 'local)
|
|
#+end_src
|
|
|
|
Define a face to show characters that have been placed on the second
|
|
line of a git commit. Those should always remain empty.
|
|
|
|
#+BEGIN_SRC emacs-lisp :tangle init2.el
|
|
(defface git-commit-nonempty-second-line-face
|
|
'((t (:inherit git-commit-overlong-summary-face)))
|
|
"Face for the supposedly empty line in commit messages."
|
|
:group 'local)
|
|
#+END_SRC
|
|
|
|
* Keys
|
|
|
|
Since the ~C-l~ combination is so much easier than ~C-j~ when using the
|
|
[[http://colemak.com][colemak]] keyboard layout and I use ~C-j~ much more, switch them.
|
|
|
|
#+BEGIN_SRC emacs-lisp tangle init2.el
|
|
(define-key key-translation-map (kbd "C-j") (kbd "C-l"))
|
|
(define-key key-translation-map (kbd "C-l") (kbd "C-j"))
|
|
#+END_SRC
|
|
|
|
* eldoc
|
|
|
|
Diminish ~eldoc~'s lighter to nothing after it loads to keep the
|
|
mode-line clean.
|
|
|
|
#+begin_src emacs-lisp :tangle init2.el
|
|
(eval-after-load "eldoc"
|
|
'(diminish 'eldoc-mode))
|
|
#+end_src
|
|
|
|
* eshell
|
|
|
|
After ~em-term.el~ loads add ~unison~ to the ~eshell-visual-commands~ to
|
|
make sure it gets unbuffered input.
|
|
|
|
#+begin_src emacs-lisp :tangle init2.el
|
|
(eval-after-load "em-term"
|
|
'(add-to-list 'eshell-visual-commands "unison"))
|
|
#+end_src
|
|
|
|
* Footnotes
|
|
|
|
[fn:1] Though it doesn't happen often that ~trunk~ is so messed up that
|
|
I can't use it.
|
|
|
|
[fn:2] Or perhaps all.
|
|
|
|
[fn:3] I think it was ~slime-js-minor-mode~, but I'm not sure.
|