her.esy.fun/src/drafts/XXXX-org-mode-intro/index.org

How I use org-mode

• Overview
  • daily routine
  • document writing
  • note taking
  • journaling
• Task Management
  • Workflow 1; planned tasks : org-agenda + clock
  • Workflow 2: Tracking; org-capture
  • Workflow 3: Add new tasks; org-capture / org-refile
• Footnotes

In this article I'll try to give an overview of my current use of org mode. I use org mode for:

• tasks management & time tracking
• writing documents (articles, book, etc…)
• note taking ; which I consider slightly different from just writing documents

In this article I would like to share a tool that was a real life changer to me: org mode.

I use it for a few years now. And for me emacs is worth learning just for org-mode. This is by far the best solution I ever used to manage my tasks. More than that, the solution is so versatile that the power is decupled by all the plugins, tools, libs that work around org-mode. Also the tool will adapt to your very specific needs.

Mainly the major problem I see with most tasks management applications is the wrong level of complexity facing the user. This is a very hard problem to tackle.

If your system is too simple, the users will not be able to manage the tasks how they would like. If your system is too complex, the user will be faced with too much details. So most successful systems have a way to adapt their apparent complexity to the need of their users. And org mode is exactly like that. Not only starting with org mode can be extremely simple but also there are mostly no complexity limit. Whatever your need most of the time there is already an emacs package that take cares of that. Even if this is not the case, it is easy to write your own.

Here is the result of a few years of improving my use of org mode. Today I can say that org mode is part of my day to day life. I still invest a bit of time to improve minor details of my workflow time to time. But now my daily workflow is mostly stable. So I think I can share it.

Overview

daily routine

The first thing I do in the morning is to open org-agenda view for today. It shows me the tasks planned for today. What are urgent tasks, deadlines, tasks that have deadlines in a few days, etc… I also have a glimpse of my habits, tasks that I should start at some hour in the day, etc…

I then start to track (clock) the tasks I'm currently working on.

Often during the day, I need to create new tasks. Most of the time I create a task and I add either a deadline or a schedule date.

Sometime I also need to deal with interruptions. In that case, I capture the interruption that will also create a new task being clocked.

At the end of the day, every tasks I worked on are saved in a tracker.org file. That file look like a date tree. And I generally generate a report that tell me how much hours I worked today. Some tasks are tagged work. The report filter only on the work tagged tasks.

Also I have some repeating tasks like review memory cards using a spaced repetition plugin. I start it, and it shows me a few cards with questions that I review. So mainly those cards contain info I want to keep in my mind and not only in my notes.

document writing

Writing documents with org mode and in particular technical document is just incredible. So at first writing document with org mode is a lot like writing markdown. Neat, there are a few minor differences. The org mode format is not only intended to be exported to HTML only so there are a few differences.

Where org mode shine is with the use of org-babel.

Mainly you can write code that will make real HTTP call, execute real functions in the language of your choice.

I wrote most part of an OAuth2 provider in Clojure. And to generate a documentation for our users it is very nice to provide the full HTTP request along the response.

But I also often need to play a few tricks in the doc and directly use our Clojure code to generate JWT for example. The great part is the ability to use those JWT generated from Clojure code in the following code block making HTTP call.

That plus the natural ability to fold/unfold the tree structure of the org mode file is great.

note taking

Time to time, I need to really take the time write note on a technical subject or sometime about articles I read about anything. For that I use org roam. I only started to use it a few months ago. But this is a great addition to my previous workflow that used deft (that I still use). But I must say, this is pretty perfect as a note taking app.

Mainly you capture notes quite easily and put links about the subject, but also tags. In the end that generate a graph of notes that you could use later to dig into your own notes.

journaling

Along with note taking. I also try to write a journal note everyday. For that I use org-journal (another org mode related package). I have a default template which take care of a few metrics I want to focus on. And I guess it is different for anyone of us.

Task Management

Workflow 1; planned tasks : org-agenda + clock

1. look at the current tasks planned for today
2. select a task, clock it
3. work on the task
4. back to the task and clock it out.

I work most of my using emacs¹. Generally the first thing I do in the morning is opening `org-calendar`. It looks like this:

Pretty brutalist interface which is a great thing to me. Distraction free interface going to the essential.

With this view, I see what I planned to do today. I also see a few "Due Soon" tasks in case I have the time to handle those.

When I start working on a task I start a clock on it (I simply type I when my cursor is on the TODO line). When I finished some task I change its status from TODO to something else. Mainly I'm prompted when doing so:

{ [t] TODO   [p] IN-PROGRESS   [h] HOLD   [w] WAITING
  [d] DONE   [c] CANCELLED     [l] HANDLED }

And that's it. The time spent on the task as been clocked I can work on another task.

Looking at the agenda view you could notice habits. They start to become green when you are doing them correctly.

But generally, I don't use much direct clocking from the agenda. Most of the time I prefer the capture mechanism. Which bring us to "Workflow 2".

Workflow 2: Tracking; org-capture

Most of the tasks I perform on the day are not planned. I have a generic routine + some prepared events and tasks to performs. But during the day you have multiple interruptions, and part of my job is to write code reviews too. I cannot plan those.

In that case I use org-capture along org-refile. Mainly org-capture helps you create a new TODO entry. And org-refile will help you move that TODO entry to the correct place.

So let say I get a direct message in the chat asking me to do something. I generally start org capture (for me it's SPC X). I am presented with the following choice:

Select a capture template
=========================

[t] todo
[c] chat
[e] email
[m] meeting
[p] pause
[r] review
[w] work
[i] interruption
[f] chore
---------------------------------------------------------------------------
[q] Abort

In my example it was a chat interruption. So I type i that presents me with this

  **** IN-PROGRESS |  :interruption:
  :LOGBOOK:
  [2020-09-23 Wed 08:01]
  ref :: [link-to-where-I-was-in-emacs-when-captured]

My cursor placed where the | is displayed. Here I add the tag chat and a small description, "dm from John about X" for example. Then I type C-c C-c and the TODO is placed in a tracker.org file under a date tree that looks like this:

  * 2020
  ** 2020-W39
  *** 2020-09-21 Monday
  *** 2020-09-22 Tuesday
  *** 2020-09-23 Wednesday
  **** IN-PROGRESS Chat with John about X                          :interruption:chat:
  :LOGBOOK:
  :END:
  [2020-09-23 Wed 17:58]
  ref ::
  ...

So the clock for this task started at the moment at made the capture. In my workflow, I prefer to finish the capture and stop clock later. So after I finished the capture, the clock is still running while the task is put in my tracker file.

Once I finished with that task. I can:

1. Jump to the tasks with SPC n o (org-clock-goto), and stop the clock SPC m c o (clock-out).
2. Jump to the task and change its status to DONE which will stop the clock.
3. Capture another tasks which will stop the clock on the current task and will start on the new one.

By the end of the day, my tracker file will contain a date tree with all the tasks I done in the day. All tasks nicely clocked. I generally create a clock report that look like this:

  #+BEGIN: clocktable :scope subtree :maxlevel 4 :timestamp t
  #+CAPTION: Clock summary at [2020-09-23 Wed 08:20]
  | Timestamp              | Headline                                       | Time   |   |      |      |
  |------------------------+------------------------------------------------+--------+---+------+------|
  |                        | *Total time*                                   | *6:40* |   |      |      |
  |------------------------+------------------------------------------------+--------+---+------+------|
  |                        | \_    2020-09-21 Monday                        |        |   | 7:40 |      |
  | [2020-09-21 Mon 08:54] | \_      check chat                             |        |   |      | 0:36 |
  | [2020-09-21 Mon 09:30] | \_      check reviews                          |        |   |      | 0:41 |
  | [2020-09-21 Mon 10:11] | \_      check emails                           |        |   |      | 0:07 |
  | [2020-09-21 Mon 10:37] | \_      review PR about xxx                    |        |   |      | 0:44 |
  | [2020-09-21 Mon 11:21] | \_      update my PR from feedbacks            |        |   |      | 0:36 |
  | [2020-09-21 Mon 12:08] | \_      review John's PR about Foo             |        |   |      | 0:12 |
  | [2020-09-21 Mon 13:41] | \_      review M's PR about Bar                |        |   |      | 0:11 |
  | [2020-09-21 Mon 13:53] | \_      another thing                          |        |   |      | 0:16 |
  | [2020-09-21 Mon 14:09] | \_      review PR                              |        |   |      | 0:51 |
  | [2020-09-21 Mon 15:00] | \_      work on PR                             |        |   |      | 1:30 |
  | [2020-09-21 Mon 16:49] | \_      check another PR                       |        |   |      | 0:33 |
  | [2020-09-21 Mon 17:03] | \_      answer email                           |        |   |      | 0:55 |
  | [2020-09-21 Mon 17:58] | \_      Chat John about X                      |        |   |      | 0:28 |

And that's mostly it for TODOs and tasks handling.

Workflow 3: Add new tasks; org-capture / org-refile

Another thing I do quite often. I need to add new task to be done. Be it for today or another day.

In that case, I generally use org-capture again. This time I choose t for TODO and I generally detail the task to be done. I add either a SCHEDULE (when I plan to start) or a DEADLINE (when this must be finished) and I refile it.

So refile will start a fuzzy search to put this task under some subtree. So instead of going to my tracker.org file, this goes to my inbox.org file.

And it will appear in my agenda.

Footnotes

---

¹

Short digression: Historically, I coded using different IDEs. Then I worked for a company that forced me to use terrible keyboards and after just a few weeks I started to have serious wrist issues. So to minimize that pain I switched to vim. And it was awesome. Once you're use to the power of vim keybinding forever your soul will bound to them. So learning vim is a bit like learning a new music instrument. You need to construct some muscle memory and integrate one after one new tricks. Once learned your personal editing power start to become overwhelming.

After a few years of vim, I wanted to try to explore new editor tooling. So I switched to emacs using the spacemacs distribution. So mainly it's vim but with even better keybindgs, helpers and within emacs. The main reason for the switch was that vimscript is a really bad language to configure your editor. Emacs use emacs-LISP. For editor customization a LISP looked perfect to me. LISP is still one of the most powerful and easy to use programming language to date.

And recently, as my personal configuration started to grow so much I switched to doom-emacs. I was quite hesitant to do the switch but so far its been a pleasure. IMHO using doom-emacs is a lot better than using my own personal configuration from scratch because I wouldn't be able to end up with so much configuration quality.