Project and task management with Obsidian and Dataview

This post is about how I manage my tasks and projects using Obsidian with the Dataview plugin.

This is going to sound like a lot. A complex setup that seems overkill. It’s not completely the case. It has evolved naturally from what my setup used to be with Things, and by adding features I wanted Things to have. At the end I will explain how it all evolved to show the historical perspective.

Overall, it is roughly based on a very loose interpretation of GTD (Getting Things Done). If you don’t know what GTD is, here’s a quick reference. Otherwise, get the book. It’s a bit outdated but still worth reading.

Projects

Projects are the main container for tasks. Except for periodical tasks (like, watering plants or recharging headphones) all my tasks are part of exactly one project. For periodical tasks, I use logic in templates crafted for each day of the week, as detailed in this post.

Creating projects is very cheap on my setup: create a new note and insert the “project” template, which looks like this:

---
status:
type: project
prio:
org:
reviewed:
---

---

- [ ]

---

- !

Here’s an annotated version for clarity:

---
status: should be open or closed
type: project
prio: an integer
org: anything (pybcn, work, empty)
reviewed: a date in my preferred format, YYYYMMDD
---

Space for support material. Usually I make this a list
of links to Notion pages, URLs, other notes.

---

- [ ] This would be a task. More later

---

- !
- [[20220814]]@18:48 Space to enter a log of events related to the project

I keep the format fast and loose, except for the task section which I keep always as in here. Sometimes reference material is part of the log, and sometimes the support material is so long it spins out a separate page.

Project review

One of the fundamental practices of GTD is regularly reviewing your projects. OmniFocus has facilities for reviewing projects on schedules, but Things had nothing: either I reviewed everything or nothing. I created some crazy automation to get something close to reviews.

But, using Dataview I can build my own review schedule. I have several project views, main ones being Personal (anything except work) and Work (only work, duh). The Dataview query for reviewing projects in Personal looks like this:

I know, WTF? But this is how it looks for 2 projects. These two are reviewed.

Clicking on the file name takes me there. I can then go back to the review list with the go-back command, cmd+[.

The header of the project view needs some explaining:

• File: quite obvious. This is the same as the project name.
• status: This is the view for open, only. I have a separate view for closed, and one for fuck, why did I put something else there?
• ◼️: How many incomplete tasks there are in this file.
• 🎫: If there are tasks with due or start dates in the file (that’s the t.annotated lambda in the query). More on this in the Tasks section.
• 🏷: Shorthand for priority to save on column width.
• org: Most of my personal ones have this empty, but PyBCN would be here too.
• 📅: If this project has been reviewed or is due for review. More details below. This is used in sorting: false will be first.

To show these green check/red cross (for 📅 and 🎫) in that HTML block in the query, I have a custom CSS snippet:

.true {
  width: 24px;
  height: 20px;
  font-size: 15px;
  text-align: center;
  padding: 0px 0px 21px 0px;
}
.true:before {
  content: "✅";
}
.false {
  width: 24px;
  height: 20px;
  font-size: 15px;
  text-align: center;
  padding: 0px 0px 21px 0px;
}
.false:before {
  content: "❌";
}

The most important part of this query in a way is the reviewed block. How does it work?

• Projects have a reviewed field with a date in the header;
• Optionally, they have a period field with a duration (default is otherwise 7d);
• The query checks whether reviewed > today - period.

By customising period in each project file header, I can have projects I only review once every 2 weeks, or once every month. In general, projects I am actively working on I want the default, but some things I want to just be reminded every 2 or 4 weeks. Visiting each unreviewed project, I have mapped alt+. to expand to the template reviewed mentioned in the beginning, so review has a very low overhead aside from having to check each project needing review. Which should be few!

New/modified note review

I hijack the reviewed field to use it to review any notes I have created recently. This would include projects, of course. The Dataview query is the following:

TABLE
FROM "/"
  WHERE file.mtime > (date(now) - dur(14 days))
  AND !regexmatch("\d+", file.name)
  AND file.name != this.file.name
  AND file.folder != "templates"
  AND file.name != "Inbox"
  AND file.name != "Pending reading"
  AND file.name != "Pending watching"
  AND !reviewed
SORT file.ctime asc

It excludes certain files that I don’t want to see appearing here or don’t need the header. I have recently switched from file.ctime (creation time) to file.mtime (modified time), to include notes I have touched, not only notes I have created recently. I’m not sure what I prefer yet. If I keep mtime I may need to add more exclusion rules.

Tasks

Tasks are where the rubber meets the road, the shit hits the fan, etc. Tasks that I plan to do are tagged (and untagged and tagged tasks appear in the project review query shown before), and a fully tagged task would look like this:

- [ ] Conquer the world [⏏️:: YYYY-MM-DD], [▶️:: YYYY-MM-DD], [🏷:: 42],  [🧷::YYYY-MM-DD],

Each of these tags means:

• ⏏️: due date. I want this task done on this day.
• ▶️: start date. I plan on starting this task on this day. I don’t use this tag much, and may remove it. It is still on all my queries though. One of due or start is needed for a task to show on a specific day.
• 🏷: priority for the task, an integer.
• 🧷: reviewed. It means I’m still happy with this task as being due on this date, or having this priority. More on this below.

In general, I only fill due date when I create a task, if I know I’ll be doing it on that day or around that week. Otherwise, I’ll choose the date when I review the project the task is in.

Task review

Reviewing tasks is more akin to reviewing the week ahead. This is something I do on Sundays, as part of my Sunday recurring template, sometimes on Monday morning. For one, I want to know what random crap out of my projects I have coming up, but I also want to be sure I’ve had a look and I don’t have 25 things to do on Monday and 3 the rest of the week. For that, I have a special note called Upcoming 7 days. It has two queries:

• One for tasks that are due in the next 7 days and where the reviewed date is in the past, or tasks that are overdue and were reviewed in the past. This lets me be sure I’m still convinced I want to do this and lets me adjust dates accordingly.
• One for tasks that have been reviewed, are due in the next 7 days, grouped by day and project. This gives me an overview of the 7 days ahead.

The first query, wrapped in an Admonition block to make it look better:

```ad-warning
title: To review
```dataview
TASK
FROM "/"
  WHERE (
    (▶️ < (date(today) + dur(7 days)) AND ▶️)
    OR
    (⏏️ < (date(today) + dur(7 days)) AND date(⏏️))
    AND !completed
    AND (🧷 <= (date(today) - dur(7 days)) OR !🧷)
  )
GROUP BY "<code style='font-size: 80%'>" + default(▶️, "") + dateformat(default(⏏️, date("2001-01-01")), "dd LLL ccc")+ "</code> " + "<em>" + file.name + "</em>"

This is how this block looks:

Clicking on the task takes me to the file with the task, so I can mark it as reviewed, for example. Then I can go back with cmd+[

The second block, is defined by

```ad-info
title: upcoming
```dataview
TASK
FROM "/"
  WHERE 🧷 AND (🧷 >= (date(today) - dur(7 days))) AND
  (
    ((▶️ < (date(today) + dur(7 days)) AND ▶️)
	OR
	(⏏️ < (date(today) + dur(7 days)) AND date(⏏️)))
    AND !completed
  )
GROUP BY "<code style='font-size: 80%'>" + default(▶️, "") + dateformat(default(⏏️, ""), "dd LLL ccc")+ "</code> " + "<em>" + file.name + "</em>"

And it looks like this:

The daily view

This is the main driver of my work and process. I have a very complex daily template that fills up certain things, including a Dataview query of tasks due today (where today depends on the filename, to play well with the Journal and Calendar plugins). I have 4 Dataview blocks:

• urgent: priority at least 42;
• top 3: except for 42s, top 3 ordered by priority;
• all due: all tasks with priority less than 42;
• done today: tasks marked as done today.

You can find more details about the templating method and plugins I use in this post: Templates for quotes and recurrent tasks in Obsidian. For more information, here’s a gist with these blocks before automated interpolation. This needs the Templater plugin with Javascript (auto)expansion enabled.

The (interpolated for exactly today Sunday, so has all projects except work) Dataview blocks, wrapped in Admonitions, look as follows:

```ad-danger
title: urgent
icon: fire
```dataview
TASK from "/" WHERE
    type = "project" AND
    ((▶️ < date(2022-08-14)
AND ▶️ AND !completed)
    OR (▶️ = date(2022-08-14) AND ▶️ AND !completed)
    OR (⏏️ = date(2022-08-14) AND ⏏️ AND !completed)
    OR (⏏️ < date(2022-08-14) AND ⏏️ AND !completed))
    AND 🏷 >= 42
	SORT completed asc, 🏷 DESC
	GROUP by file.name
```ad-warning
title: top 3
icon: arrow-up
```dataview
TASK from "/" WHERE
    🏷 < 42 AND
    type = "project" AND org != "work" AND
    ((▶️ < date(2022-08-14)
AND ▶️ AND !completed)
    OR (▶️ = date(2022-08-14) AND ▶️ AND !completed)
    OR (⏏️ = date(2022-08-14) AND ⏏️ AND !completed)
    OR (⏏️ < date(2022-08-14) AND ⏏️ AND !completed))
	SORT completed asc, 🏷 DESC LIMIT 3
```ad-info
title: tasks
icon: list
```dataview
TASK from "/" WHERE
    type = "project"
    AND (org != "work") AND
    🏷 < 42 AND
    ((▶️ < date(2022-08-14)
AND ▶️ AND !completed)
    OR (▶️ = date(2022-08-14) AND ▶️ AND !completed)
    OR (⏏️ = date(2022-08-14) AND ⏏️ AND !completed)
    OR (⏏️ < date(2022-08-14) AND ⏏️ AND !completed))
    SORT completed asc, 🏷 DESC, prio desc, org desc
    GROUP by file.name
```ad-success
title: done
icon: checkmark
```dataview
TASK from "/" WHERE
    type = "project" AND
    ✔️ = date(2022-08-14)

They would look like this, when empty of tasks (with tasks, they’d look similar to those above):

The queries are crazy, I know. They involved a lot of trial and error to play well with Dataview, and in my template are mixed with Javascript to interpolate the current date, too. Also, having start and due f-ks up a lot the queries, when I remove start I can simplify them a great deal.

How this process has evolved

I started by just having tasks on days.

First, recurrent tasks that I had to do repeatedly (see this post), then, project tasks that I would manually copy from one day to the next.

Eventually, I realised I needed projects… And started copying tasks from projects to the day I wanted to do them.

Then, using Dataviews for tasks from projects. I kept adding to the query to make it better, until I had the 4 blocks above.

I stumbled upon somebody checking recently created files, and I wanted that too.

From there, I realised I could review projects more effectively by listing projects that needed review instead of checking each project, which was slow and very repetitive.

The most recent addition is removing work-related tasks after work hours.

Some ideas I’m exploring, are to automated “holiday checks” so my templates don’t expand for work on days I’m not working because I’m off.

• Productivity
• GTD
• Obsidian