I just checked on an open shell and noticed that LESS contained what I
meant to put in thrice. Don't know if that has something to do with
tmux, but apparently the zprofile is sourced multiple times? - good to
know.
I found this in feh's .desktop file and liked the idea:
With `--start-at` feh will load all files from the directory of the
given file and start the filelist at the given one.
This of course breaks easily if I want to pass more flags before the
file argument, but is easily fixed by first specifying the argument to
`--start-at` and continuing with the rest. I also rarely use any flag
besides the one already defined in the alias.
The `${(*)...}` parameter expansion flag was only introduced in
zsh-5.8.1.2-test, but EXTENDED_GLOB is needed for the `(#m)` globbing
flag. To support older zsh versions too, solve this by using a subshell
in which EXTENDED_GLOB is temporarily enabled.
It's really annoying if it tries to correct me when I create a new
directory that is named similar to an existing one.
Group together aliases that add a precommand modifier.
Since `mkdir` receives the `-p` flag already via `add_flags` the `md`
alias can directly alias to `mkdir` instead.
I am still unsure if I want the author or committer name. For the date I
think it makes sense to have the committer date since it reflects better
when the branch last changed.
When setting up the remotes the fetch url can use https since the fork
will be public anyways. This delays a prompt for the ssh key until it is
really necessary.
I often get annoyed when I try to copy the command I typed but the
timestamp is also copied (e.g. with tmux line selection). Move the time
segment to the very right of the first line of RPROMPT which empties the
second line completely.
In the flat/non-graph view merge commits are a bit distracting. Still
unsure if I will forget about this and one day wonder why I don't see
merge commits.
Replace custom awk solution with uniq, by first flipping filename and
filesize so that uniq's `-f` flag can be utilized (as there is no
inverse of it, i.e. "only look at field n").
This increases performance by quite a bit.
Use core.commentchar to identify commented lines and use the cut line
instead of just deleting from the first comment on, as this would break
for example in git generated messages (e.g. squashes).
