Since I often use git for prose and textual writing, wordwise
diffs make a lot of sense to have quick access to. This commit
sets up a git-internal alias `git diffword` as well as an
even shorter `gdd` (diff) and `gdds` (diff staged) which
mimic the other diff aliases `gd`/`gds` already existing.
Since sioyek does not support including into its configuration
file, we switch it to dotter templating and use the in-place
flavours insertion on theme selection for theming.
Any theming will not back filter into the repository.
Added strings which signify to flavours that it should insert
its theme color template in-between. Since the file is not
symlinked but *moved* to the config dir (through being
designated as 'template' in dotter), any future changes
should not reflect back into the repository.
Fixed an accidentally templated (papistui.yml) file, due to its
use of mustache-like double-braces.
Explicitly target another file as templated so we can change it
in the configuration directory (mako getting flavours colors
since it STILL lacks includes).
Fix the warnings given by dotter for files that are not
utf-8 encoded to explicitly be identified as symlinks.
Should ultimately probably be either moved into data
directories or moved out of repo altogether, but for
now it works well.
Since we now use dotter we can simplify the dir structure for
qutebrowser a lot. Everything dot-filed earlier can now reside
in simple directories called config (for ~/.config/qutebrowser),
data (for ~/.local/share/qutebrowser), and scripts
(for ~/.local/bin) files.
Everything that is in the repo should be listed in the global.toml
file. The local.toml file then picks the stuff to actually
write. For systemwide packages, a systemwide.toml (local) file
is used instead.
HACK Since there is some weirdness with the whoosh backend for papis not
being able to either index or search for integer values in custom
fields, this commit sets the priorities across the library (as well as
the tui picker) to make use of textual reprentations.
1 -> high
3 -> medium
5 -> low
This means we can't do simple arithmetic evaluations on priority anymore
but at least we can search it correctly in papis. Should probably put in
a ticket to see if it can be fixed.
Since we switched to viu, we should also have qr
make use of it. Renamed the function to
qrurl to make its purpose clearer (sending
a simple file location/string/url via qr code)
and to distinguish it from programs like qrcp
which actually start a server and send a file
through qr.
We now use chafa to preview images. We use its iterm mode
since that seems to be the best integrated into wezterm
currently.
We have to use uncached previews so it is a little slower,
but at least it functions perfectly after the first image
is displayed.
For anything markdown or that gets displayed as markdown
(e.g. docx files) in the vifm preview, we now use glow
or bat to display a nicely styled and colored version.
For glow I added a script which tries to detect the
current terminal background between dark/light to
correctly set the color scheme.
Added markmap plugin to make mindmaps from markdown. Uses
headings for leaves, and works pretty automatically.
Testing its functionality longer-term for now.
Removed notification on activating a python venv.
It should just work, transparent to the user and we do
not need a big ol' notification each time we enter
a file.
Can edit (`<ll>ce`), jump-to (`]c`, `[c`), insert (`<ll>co`,
`<ll>cO`) and run (`<ll>cc`) code cells in markdown files.
They will not be evaluated as part of an overall repl but
only stand-alone!
The `papis tags` command is extended with the functionality to not
just display all tags used in current query, sorted alphabetically,
but also to display how often they appear.
Use `papis tags -c` to show a space-separated count of how often
each tag is used next to the tag name, in descending order.
Do not allow html5 canvas reading in qutebrowser. This shuts off one of the main
avenues of browser fingerprinting (aside from cookies) still currently
undertaken. It will possibly stop some obscure streaming sites from working but
I do not care, especially at the cost of fingerprinting information.
When not showing tabs (only displaying them on actually
switchting between tabs), so far the duration that the
tabs would be displayed was just 800ms, way too short
to actually provide legible info.
This commit changes it to 2000ms, which may still be
too short but is at least more digestible.
Added simple configuration for the imv image viewer which is
wayland-compatible. Nothing too fancy but some simple
vim-like mappings are set, as well as some styling applied.
In addition to the existing `glg` and `glga` aliases for showing
one-line git logs for current or all branches, this commit adds
`glgd` and `glgad` for showing one-line git logs for current or
all branches *with date information*.
The date info comes from authoring date not committing date.
Kanshi used to be restarted every time river would be reloaded.
This is not desired however, as restarting river would also
kill the kanshi daemon, and subsequently reload it, reseting
any custom kanshi output layouts or options set.
With this commit kanshi only gets started if it is not already
running through river (i.e. on first boot most of the time).
Use F10 to enable gaps (default) or Shift+F10 to disable gaps.
TODO Would love to make it a toggle on pressing F10 alone, but
I am not sure we can get the current gap width from the
layout in any way.
Replaced or extended all default references to zathura with
sioyek so it will automatically take on any tasks meant
for pdf reading. With the current configuration, hardly a
change should be noticed.
fasd is unmaintained and slower than zoxide. The transferral
was painless. I imported my old database and can continue as
before. It does not care about files but that is completely
fine for me. Same `z` invocation as before. Has the
'interactive' mode on `zi` which is also completely fine.
Automatically set formatting mode to soft for markdown, text and asciidoc files.
Also automatically format on startup (no lazy-loading), and thus have
nicely word-ending linebreaks (a word will not just be cut off wherever the
line is over but will be fully moved to the next line instead).
Add a hash sign in front of 'project' tags received from
taskwarrior (i.e. where it is `project:something` it will
end up as #something in timewarrior.
Similarly, add a plus sign in front of tags: `+mytag`.
Only descriptions will be left as-is in translating from
taskwarrior to timewarrior.
This should make finding tags and projects specifically
much easier in timewarrior (e.g. for counting up the total
time spent on a single project).
Added neotest with some mappings to run tests and view
outputs, as well as neotest-python for now to make it
work under python.
Added registering with which-key if it exists.
Mapped to `[D` / `]D`, reflecting the capitalization of going to the
next general diagnostic (`[d`/`]d`).
Now it is similar to spelling mistakes and spelling errors
(`]s[s`/`]S[S` respectively).
