I wrote myself a python cli tool that manages the stack by updating a sqlite database, and one of the commands creates a text file associated with the task (if it doesn't already exist) and opens it in Emacs. The text files are stored in a hard-coded directory and an anacron job does a git commit once a day.

I can tell what I worked on each day by querying the git commit history, and I can grep the entire directory for keywords.

It's a little janky but it works pretty well for me.

Ahh, I was expecting "journal" to refer to a physical book for some reason!

This makes a lot of sense. Why not simply store the task with the sqlite database? I'm assuming ease of editing + the ability to manage the stack separately from the log of text entries, which presumably need no maintenance nor will ever be deleted?

Pretty much what you said, yeah. I thought that a git repo of text files was easier to work with then storing them as blobs inside sqlite.

I do insert completed tasks into a "completed task" append-only table when I pop them off the stack, so I do have a record of completed tasks in sqlite. (I find that useful for remembering what I did recently for standups and 1:1's)