I was asked on the Fediverse1 to describe my $dayjob daily workflow. Blogging about it seemed like a good opportunity to take stock of my current setup.

The core of my approach is to maintain a daily log, rather like an engineering Lab Book2. My number one tip for anyone starting a career in Software Engineering (or for that matter systems administration) is to keep a daily log of what you intended to do, what you actually did, what you changed, what changed as a consequence, etcetera. I've already written about the tools I use for that: vimwiki.

They key drawback of that system is managing TODO lists. In Vimwiki, tasks look like this:

  * [ ] buy milk (still todo)
  * [X] clean the car (done)

These get dispersed or duplicated around different days/wiki pages. It's hard to get a current view of open (or otherwise relevant) tasks, and impractical to update many copies of the same task when its state changes (such as when it's done).

The solution I've adopted for now is another Vim plugin, taskwiki, which synchronises tasks with Taskwarrior3, an external task-management tool.

If I mark a task as "done", Taskwiki updates all references to that task to reflect the new state. I can also construct queries to list all tasks matching some criteria. I have a special Vimwiki page named "Backlog" which runs the query "all tasks tagged 'redhat' in state 'pending'" (Linked from the boilerplate at the top of every page I write, for quick access):

= Backlog | +redhat status:pending =
* [ ] buy milk (still todo)
...

Much like the base Vimwiki plugin, Taskwiki is very opinionated, and I've had to tame it by disabling several of its features. I've also hit a couple of mildly frustrating bugs (#368, #425). I might one day have a go at writing an alternative, simpler plugin in Lua (Neovim's native scripting language), but for now it works well enough and I don't have the time.

There's very little in this current workflow for managing scheduling tasks, and that's probably where the focus should be for my next iterative improvement efforts. I think Taskwarrior, the underlying tool, has some good support for that. I'd particularly like some more visual approaches for managing the backlog, such as something Kanban-style.


  1. sorry I can't find who/the toot any more
  2. Analogous to an engineering lab book, but the analogy is not exact. I recently read a little bit about what Engineers are actually taught to do in a lab book, and it's quite a bit more narrowly-scoped than I realised. This slide-deck is interesting: https://www.training.nih.gov/assets/Lab_Notebook_508_(new).pdf
  3. I rarely use Taskwarrior directly yet, and so I haven't written about it independently of Taskwiki..