Having a general script folder makes little sense if the scripts are
targeted to specific applications. This commit moved every script that
solely, or mainly (like ueberzug), targets a single application into
that respective stow module.
Enables easier vcs integration of vim spelllang dictionaries. Will cause
them to be reloaded on external editing, add their binary parts to
ignore files, and set their addition files to unison merge mode.
Plugins are managed directly through pacman packages now. Since I am
only using six zsh plugins and they are staying relatively constant and
not changing much, pacman seems a much more sane way to manage them
without cluttering up my zsh initialization routine. All zgen management
can go and just needs 6 `source` lines in order to load the plugin
entrypoints.
Anything which needs to be (at least partly) touched by stow should stay
in the top-level normal directory name that they are. Anything that
*only* concerns the repository, not the dotfiles, should be moved into a
hidden folder, preceded by a dot.
These directories should still contain a `.stow-local-ignore` file
patterning everything which should not show up in stow (i.e. `^/.*`,
everything) since it is not guaranteed that stow will ignore the
.dotfile directories. The installation script does ignore them on its
own, for now, but is not guaranteed to in the future either.
When moving to commit, this hook will automatically fire and check the
current system's installed packages against those explicitly committed
to the repository. If they mismatch it will inform the user.
It will not prevent the commit, but simply add a comment at the top of
the commit messages to remind the user that something is unbalanced
between both. It would be recommended to either check the additional
package into source control, remove it from the current system, or
explicitly add it to ignored packages.
Updated package list in preparation for unified package management. For
the moment still uses basic bash script to manage package installation.
Packages are collected in a big list, with some packages being ignored.
This is in preparation for the possibility to manage future package
'groups' directly in pacman meta packages. They will then be added /
removed to the individual PKGBUILD files as dependencies and will
automatically be kept up-to-date, added and removed on all systems which
use the corresponding meta group.
It also opens the path to have commit-time checking of irregularities
between the packages stipulated in the dotfile repository and the
current running system. This can be done by adding all packages of the
different package group lists, unifying them and removing potential
ignored packages.
It allows e.g. printing a warning when committing changes to the
repository while the system packages and the repository packages differ,
to make sure no package dependencies were introduced without knowing it
and which would introduce potential breaking changes into the dotfiles.
Added custom term spell dictionary for neovim. Will track any words that
should be needed on all my machines.
Added spellmake plugin which watches spellfiles and rebuilds the binary
whenever a new word has been added through external edits like git.
The plugin also automatically adds a gitignore file for the .spl
file, and a unison merge strategy through a gitattributes file to any
spell directory it finds. Ignoring is due to .spl being automatically
generated binary files. The merge strategy ensures that no merge errors
will pop up even when words get added in different orders. All it cares
about is that the content of the files is the same, in whatever order.
Fixes tmux xdg-compliance (and, more importantly, Tmux Plugin Manager's)
by setting the environment variable TMUX_PLUGIN_MANAGER_PATH to follow
xdg specifications. Tmux, due to not being xdg-compliant, needs to be
aliased to start with the `-f` option pointing into the configuration
directory.
Fixes tmux vim nagigator's controls being overwritten by other control
schemes in tmux.
Added general usage information. Made sure gitignore api is never called
without any arguments.
Made script fzf aware -- if it finds fzf it can be invoked without
arguments, if it does not it will display usage information instead.
Removed the `autostow.sh` scipt. Its use was to call stow for every
folder in base directory and ignore certain folders. Both those
functions can be handled by stow on its own.
Stow allows defining per-directory ignore patterns with
`.stow-loca-ignore` files, which can be set to `.*` to completely ignore
a folder, just as before. And Stow can be called with a glob pattern to
automatically call it for every directory in the repository.
`.stowrc` additionally makes sure that all operations take place
targeting the home directory of the current user, since that is where
the dotfiles will (generally) be stored. Of course, this can be
overridden with the stow command-line options (see option precedence in
stow manual).
Finally, the bootstrap stow module adds an alias `dotlink` to the shell,
which allows fast (re-)stowing of all directories in the dotfile
repository. It uses a hard-coded location for the .dotfiles base
directory, so if the dotfiles are cloned anywhere else this has to be
customized.
Removed superfluous styler `--theme`-style options, instead relying on
the commands provided to be the primary way of interaction.
Documented the new standardized ways with the `help` command and added
improved documentation for all avaliable options.
When creating a new link within notes, using the wiki.vim openlink
shortcut (return by default), it will prepend a Zettelkasten-like unique
id based on current time in front of the link.
It will also lowercase it and substitute spaces for dashes.
Pandoc citations added through fzf (with the :CiteRef command) were
added in place, with an i. The usual case however is to write a
sentence, or part of a sentence, going out of insertion mode and wanting
the citation to appear *after* what was written.
Same story for insert mode citation (automatically called by typing
`@@`); also simplified its calling function somewhat.
Added simple maps for fzf finding note files from anywhere, searching
through wiki tags.
Vim experimental modules should be short-lived and really used for
experimentation, not be always in git and messing up diffs.
An initial foldlevel of 2 has been added to vim, this is experimental
and if it doesn't suit me can be easily reverted.
wiki.vim overwrites omnifunc for any buffer it sees as its own. Pandoc
uses g:pandoc#competion#Complete as its omnifunc to generate citations.
So we use deoplete to collect the citations and display them on being
invoked by `@`. When typing more, it filters the list accordingly.
Have not gotten preview window working again, nor searching of fields
other than bibkey.
Setting everything during PlugLoad function caused some plugins to
misbehave. Plugins are now loaded by Plug and their settings can be
either set with a file in the plugin directory (this mimicks the old way
of setting plugin up during load) or in after/ directory, which sets
options *after* everything has loaded.
Added tmux session chooser. Aliased to `tm`, calling `tmux_attach_start`
(the original tm script).
When called without arguments displays a fzf list of currently running
tmux sessions, with a preview to their respective open panes. A session
can be chosen in fzf which tmux will attach itself to. When creating a
query in fzf which does not have a valid target and confirming, tmux
will automatically create that session and attach itself to it.
When called with an argument, tmux will attach itself or create a
session of the same name. If called with the name of a session file, as
before, tmux will automatically execute that session file and attach
itself to it.
Added `$XDG_CONFIG_HOME/zsh/completions` to fpath for zsh to source
completions from. Any stow modules creating scripts and functions which
should be auto-completable can add their own completion scripts to this
directory for zsh to automatically pick them up.
Add zsh-users/zsh-completions plugin to provide completions for a wide
array of applications (e.g. glances, nvm).
The only file left in $HOME is .zshenv, which sets up zsh to source everything from XDG_CONFIG_HOME/zsh.
Shell files are split into sh and zsh directories, for global assignments (which should be posix compliant, work on any posix shell) like environemnt variables, xdg vars, and global aliases. zsh contains zsh specific customization (prompt customization, plugin loading, zsh completions).
Zsh initialization will pull from sh directory first, loading the respective mirror to its startup file (`.zprofile` loads `sh/profile` and `profile.d/*`, `.zshenv` loads `sh/env` and `sh/env.d/*` and `zsh/env.d/*`, `.zshrc` loads `sh/alias`, `sh/alias.d/*` and `zsh/alias.d/*`)
Once all is done, it will have loaded both global variables, aliases and settings, and zsh-only specifications. Other stow modules, if they want to add shell functionality, can include their aliases and functions in one of the above directories to automatically be picked up by zsh.
Added global git config file, following XDG specifications (putting it
under `XDG_CONFIG_HOME/git/config`).
Adding a small improvement to default dot session for tmux, so that its
log watcher now shows all refs and remote refs.
Makes polybar use Xresources color values for its colors. Resets rofi
values to their default after being overwritten by the base16 themes.
Makes qutebrowser include the colorscheme in its configuration file.
Ignore dynamically generated colorschemes in .gitignore file.
The theme function provides a quick but temporary switch to a specific
theme.
Themes can be applied with `styler theme [themename]`, they will be
instantly applied to all enabled applications. They will be lost on
restarts of the application or machine however.
Themes can be applied permanently with `styler set [themename]`, they
will be written into the application's settings. This *will* change
things in your local filesystem, so be wary. Generally, processors
should take the least invasive approach and use inclusion to append the
theme to the application's other settings.
But be *careful* with this option, and when in doubt double-check what
the processors you have installed do before losing your settings.
Can take input in the form user/repo and pulls packages from github into
its package directory. Documented use of new function.
Can download both processors and packages. Styler decides between
both with their naming schemes, searching for `/base16-` or `baseproc16-`
respectively. For example:
username/base16-programname - package
username/baseproc16-programname - processor
If it can not determine it based on name alone, it will assume to be
dealing with a processor but spit out a warning for the user.
Styler processors use same layout as packages, such that package and
processor directories are mirrors of each other. So:
`chriskempson/base16-vim` lies in the packages directory,
`marty-oehme/vim-processor` mirrors it.
This opens up the way for custom processors.
Can list available themes, packages, and processors. Invoked by `styler
list` and then the intended target. When invoked without any valid
target will remind the user to supply one.
`list`, `--list`, `-l` are aliases and perform the same function.
Examples:
`styler list themes`
`styler --list packages`
`styler -l processors`
Added up-to-date install instructions and a simple image to the README.
Added an assets directory to house these things and some notes. Removed
`.gitlab-ci.yml` from being linked to the home directory, it does not
belong there.
Renamed bootstrap directory to _bootstrap to conform to default ignore
pattern of autostow.sh, and make it visually clear that this is a
special directory, not one used for the dotfiles itself.
Marty Oehme