Introduction
My work-life revolves around my TODO list. It’s how I structure my day, it gives me short-term goals to keep me motivated, and provides an easy way to recall what I did the day before for the daily standup.
The problem I’ve run into is that todo list apps universally suck. They’re slow, mouse-centric, low information density, and most importantly completely inflexible. As a tool you use all day, your todo list needs to completely match how you work (and think!) or you’ll end feeling miserable and eventually abandon it (see Jeff Atwood’s polemic Todon’t).
Organized people may use todo lists, but todo lists won’t make you organized.
The only solution I’ve found to work long term is to abandon the idea that a single tool can solve this problem, and instead focus on creating a workflow that best matches how you really get-things-done.
The rest of this post details my personal workflow which, even if it doesn’t work for you, hopefully inspires you to take another look at how you stay organized.
Design
The workflow I’ve adopted wasn’t really designed, but evolved over the years from trying a variety of approaches, finally settling on a relatively stable set of repeatable steps. Looking back, though, the things I’ve found important have been:
-
FREEFORM: The TODO list should be a single text file with minimal structure. This means that if you have raw data, tracebacks, or anything else, just paste it directly into the TODO list. Almost always, this data will be much more useful to you than the short-description that todo-list apps require.
-
DAY-CENTRIC: The list should be broken up into days. This allows you to see what you did yesterday at a glance (useful for standups). This also allows you to, upfront, set a personal goal for what you expect to get done that day. I find this motivating, not guilt-inducing.
-
REVERSE CHRON: New entries should be at the top, older entries at the bottom. This means when you open the file up, you’re immediately ready to add a new entry.
-
KEEP IT SIMPLE: The temptation will arise to try to automate everything. Don’t! The goal should be to have a process that stays out of your way, not one that minimizes the number of keystrokes. Trust me, a few extra keystrokes paired with a simple, repeatable workflow, will make you much happier than a workflow built around incantations and black-magic.
-
BACKUP: Once you’ve developed a decent TODO list, you’re going need a way to back it up. You could use TimeMachine or the like, but, I find
gitprovides the extra benefit of giving you a change history for your TODO list. Personally, I use my own homefiles tool for this purpose (along with backing up all my other documents).
My Workflow
Probably the best way to describe this workflow is just to walk through some
of the basic steps and, where appropriate, describe the few customizations
I’ve done to vim to speed things up.
Pull Up The List
This is something you’re going to do many, many times a day, so this needs to be quick. To accomplish this, I’ve mapped a key sequence to horizontally split the screen and open the TODO list file in the new pane:
nmap ,td :new ~/Documents/TODO<CR>
Add Today’s Entry
Each day gets a date heading and a section devoted to PENDING and DONE tasks and looks like:
2012-11-29
PENDING
- Not started yet
DONE
- Finished this one
Rather than having to type this out each day, I’ve automated this part as well:
nmap ,nd :r! date +"\%Y-\%m-\%d"<CR>$"="\nPENDING\n\n\n\nDONE\n\n"<CR>pjjj"="- "<CR>pA
This creates the entry and drops you into insert mode on the first pending item for the day.
Add a Pending Task
To add a task, just create a new line under the PENDING section. Remember, this document is completely freeform, so put anything you want here:
PENDING
- Fix whatever's causing this:
2012-11-29 21:47:59 TRACE nova File "/opt/stack/nova/nova/compute/manager.py", line 3135, in update_available_resource
2012-11-29 21:47:59 TRACE nova nodenames = self.driver.get_available_nodes()
2012-11-29 21:47:59 TRACE nova File "/opt/stack/nova/nova/virt/driver.py", line 768, in get_available_nodes
2012-11-29 21:47:59 TRACE nova return [s['hypervisor_hostname'] for s in stats]
2012-11-29 21:47:59 TRACE nova KeyError: 'hypervisor_hostname'
2012-11-29 21:47:59 TRACE nova
Mark a Task Done
Marking a task done is accomplished by cutting and pasting the item from the PENDING into the DONE section for the day.
DONE
- Fix whatever's causing this:
2012-11-29 21:47:59 TRACE nova File "/opt/stack/nova/nova/compute/manager.py", line 3135, in update_available_resource
2012-11-29 21:47:59 TRACE nova nodenames = self.driver.get_available_nodes()
2012-11-29 21:47:59 TRACE nova File "/opt/stack/nova/nova/virt/driver.py", line 768, in get_available_nodes
2012-11-29 21:47:59 TRACE nova return [s['hypervisor_hostname'] for s in stats]
2012-11-29 21:47:59 TRACE nova KeyError: 'hypervisor_hostname'
2012-11-29 21:47:59 TRACE nova
- Something else I did earlier
Trust me, ddjjjp will eventually feel way more satifying than clicking a checkbox.
Moving Unfinished Work
Invariably you’re going to have tasks that span multiple days. In this case, you just push forward any unfinished tasks by cutting them from the previous day’s PENDING section and pasting them into today’s.
This means, ideally, the PENDING sections for previous days should be empty. It’s a good idea to periodically scan down the list to make sure that condition is holding.
Handling Sub-Tasks
Sub-tasks are added by indenting them underneath the parent task, like so:
PENDING
- Main task
- Subtask1
- Subtask2
The key difference between sub-tasks and top-level tasks is that sub-tasks
aren’t moved into the DONE section when completed. Instead, a [DONE] tag is
added to the end of the line. When all of the sub-tasks are tagged as done,
I then cut-and-paste the top-level-task and its associated sub-tasks into the
DONE section to mark it completed:
DONE
- Main task
- Subtask1 [DONE]
- Subtask2 [DONE]
At the risk of over-automation, I also have a keybinding for this:
nmap ,dn A [DONE]<ESC>
That’s It
That’s all there is to it. With these simple operations, I’m able to keep track of what I have left to do, and see what I’ve done without having to futz with a separate tool.
This workflow is simple, flexible, and most importantly, stays out of my way. I find myself spending very little time actively thinking about my TODO list, and instead, rely on habits I’ve built up from using this on a day-to-day basis to move quickly from task to task.
There’s only enough structure to be useful and not a bit more.
It’s with this in mind that I challenge you to think about the TODO list app that you use and ask yourself how well it matches up with how you really work. And if you find it wanting, hopefully, this blog post serves as an inspiration for you to create a process that 100% works for you.
Postscript
Here are some entries from my TODO file that show how it (and this process) have evolved over time:
First Entry
2010/10/13
==========
#info {yesterday} Verified #T2164 for dabo, Fixed Python 2.6 compat bug in
glance, inter-review call
#info {yesterday} Troubleshot Iback issue w/ hinch, Drafted email explaining Iback enabling in Backstage
#info {today} grabbing and working a blueprint for glance, review mergeprops
etc
done
First Entry w/ PENDING and DONE Sections
2011-06-02
PENDING
DONE
- Writing blueprint for GlanceZones (pending feedback)
- Work on jk0's bug - decided to punt that to next sprint
- Fix 2b per B Waldons comments
- Demo Prep
- Looking at glance package issues with murkk, looks like ubuntu package is
built incorrectly, call `glance manage db sync` instead `db_sync`
- Got comments from Jaypipes and Dubs on Glance Zones
- Reviewed JayPipes S3 Patch
First Entry w/ Sub-Tasks Tagged with DONE
2012-07-09
PENDING
DONE
- Fix migration
- Refactor migration [DONE]
- Make migration use import_vhd [DONE]
Appendix
Commands in one place for pasting into your .vimrc:
nmap ,td :new ~/Documents/TODO<CR>
nmap ,nd :r! date +"\%Y-\%m-\%d"<CR>$"="\nPENDING\n\n\n\nDONE\n\n"<CR>pjjj"="- "<CR>pA
nmap ,dn A [DONE]<ESC>