Oct 2, 2020
A Better Zsh History, pt. 2
Configuration options
Previous: Part 1 — Setup
In part one we did some basic setup to get the history mechanism working in Zsh. In this article we will explore a few options to customize the Zsh environment using setopt in the .zshrc. The full list of options can be found in the documentation.
Persistence
Default behavior is to overwrite the existing history file upon shell exit. The following are all methods of appending rather than rewriting the file — taking advantage of that potentially massive history size we set up in part one.
APPEND_HISTORYINC_APPEND_HISTORYSHARE_HISTORY
The difference being that APPEND_HISTORY appends lines in the order in which each session is closed while INC_APPEND_HISTORY adds lines to the history in the order in which they are executed. SHARE_HISTORY is the equivalent of INC_APPEND_HISTORY but also imports new commands, allowing lines to be shared across multiple open sessions.
So what does any of this mean? It means that someone like me who works with a terminal multiplexer like TMUX can configure Zsh so that every shell command persists in the history beyond open/close and are immediately available across all open windows, making the terminal feel much more integrated.
In the .zshrc I’ve added:
setopt INC_APPEND_HISTORYDuplicates
Now let’s take a look at how we can handle duplicate lines. For example, if we’ve just run exit on three open TMUX windows and would like to cycle back to a history item just before, we would have to traverse the duplicates. We can do better.
HIST_IGNORE_DUPSHIST_IGNORE_ALL_DUPSHIST_EXPIRE_DUPS_FIRSTHIST_SAVE_NO_DUPSHIST_FIND_NO_DUPS
HIST_IGNORE_DUPS prevents storing only repeated lines that are adjacent to one another, HIST_IGNORE_ALL_DUPS removes old duplicates of the line and keeps the new one even if lines are not adjacent, while HIST_EXPIRE_DUPS_FIRST removes old duplicates only when the history is filled up.HIST_SAVE_NO_DUPS is the most aggressive of these options since, as the name suggests, it saves no more than one duplicate.
And lastly, HIST_FIND_NO_DUPS means that even if duplicates are saved in the history, they won’t be shown more than once while cycling back.
My current duplicates handling is on the heavy handed side and along with everything else we’ve set up to this point looks like this:
# ~/.zshrcexport HISTFILE="$HOME/.cache/zsh/zsh_history"
export HISTSIZE=1000
export SAVEHIST=$HISTSIZEsetopt INC_APPEND_HISTORY
setopt HIST_IGNORE_ALL_DUPS
setopt HIST_SAVE_NO_DUPS
setopt HIST_FIND_NO_DUPS
A few more (one for now) options
- A well explained use case for
HIST_IGNORE_SPACE
Conclusion
Your configuration ultimately depends on the way you prefer to use the Zsh history mechanism. I’ve personally found that I am more apt to use features that fit my workflow; what can be more fitting than one that I have complete control over?
Don’t be afraid to configure, try it out for a while, ask “what if…” and then reconfigure. For the amount of time we spend with our tools, we deserve to enjoy using them.
Previous: Part 1 — Setup
