I have a Project directory for all of my coding projects. For smaller projects, I created another directory sandbox. However, I soon realized that there were many smaller mini-projects. I often wanted to explore a topic for about a day. So I always created a directory /tmp/shit and stored everything there. In the next boot, they were deleted, and I could focus on another topic in /tmp/shit.

I noticed this was convenient to use. The path to this directory was fixed. I could locate /tmp/shit in my browser very fast, compared to lots of subdirectories inside sandbox. Also, it reinforced the idea of “delete and rewrite”. I started my big projects as one or two codes inside /tmp/shit. The next boot, I had to rewrite that code. This new code didn’t take as long as the first one, yet it was already simpler and cleaner.

The bad part was that I lost too many lines of code. Sometimes I learned something new about a subject I wrote in a /tmp/shit, and I would have wanted to see that old code again. To make my shits permanent, I created a new command mkshit:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14

mkshit() {
    local shit name template when where
    shit="~/shit"
    name="$1"
    template="$name-XXX"
    if [[ -z "$name" ]]; then
        template="XXX"
    fi
    when="$(date +'%Y-%m-%d')"
    where="$(mktemp -d "/path/to/shit/$when-$template")"
    rm "$shit"
    ln -sf "$where" "$shit"
    cd "$shit"
}

The directory /path/to/shit is where shits are going to be stored. The name of each shit has a format similar to 2024-11-04-sql-49t. The first part is the creation date, the second is the name of shit (optional), and the third is a random three-letter string. The directory ~/shit is linked to the last shit created.

If you want mkshit, copy and paste the above code into your .bashrc.