OrgMode

Switching to Multiple-File Setup

Note: I was supposed to put this post up about two weeks ago… Well, you know how it is. Life got in the way. As such, this is not fully updated with what I’ve discovered in the last two weeks. But, that’s why I have a blog for, right?


I used to write all my tasks, personal and work, into one tasks.org file.

On Sunday night, this was good. I had 5 tasks on my list, and I was ready to start my work week. But it didn’t take long (two days actually) for tasks.org to become monster.org. It didn’t happen because of the number of tasks, which I kept (more or less) under control. It happened because of the size of the projects I was working on. Setting up computers, encryption, and even elementary personal stuff like paying my bills; each task naturally grew to sub-tasks, and those, in turn, had their own notes and lists.

My initial solution to that was to create a “Details” heading for the projects. It contained time rangers I worked on a project (entered manually) and links to other sources I needed. This kept extra information out of the way when I didn’t need it. This caused two issues. First, I now had “Details” showing on my agenda, since my time-range was directly under that header. Second, things quickly got out of control on my Android phone with Orgzly. That’s because Orgzly does not fold secondary headings. I had to deal with walls of texts which I had to scroll through before I got to the most recent ToDo items.

As I was scratching my head at this, work and life continued. My tasks list grew each day – No, each couple of hours. Interruptions kept coming in, obscure urgent new projects popped up while older ones from previous weeks resurfaced. My list was quickly overtaken by work stuff, while personal projects remained in the background, often pushed down the list and out of view.

Realizing that I can’t handle just viewing my tasks.org on my phone anymore (toward the end of the week, I could barely do it even in Emacs) I started using the tool I should have used more from the beginning: The agenda.

The Agenda

The agenda view changed everything. Opened from everywhere with a quick key combo, it enabled me to see everything that I need to do. This is thanks to one thing that survived through the mess was my method of scheduling assignments I intended to work on that day or in the next couple of hours. I picked this habit from one of the old Org tutorials floating around, which I cannot find right now. Scheduling means I still had a wall of ToDo items in tasks.org, but I only scheduled up to 5 things I intended to work on. I am only human, after all. Scheduling showed me what I wanted to do, and ToDo keywords showed me assignments that I haven’t yet scheduled, but need to at some point. I could view both comfortably from the agenda view with C-c a n.

This proved to be effective not only in Emacs but also in Orgzly, on my phone. Thanks to the “Scheduled” search, I now also have a widget on my homescreen, an effective todo list. There’s even a check button to check off Items on this widget, which marks them as “done.” Orgzly also does a good job at creating customized searches, so that I can specifically see what is scheduled for today vs what is scheduled for the next 3 days, or week. I can have another filter showing me my unscheduled ToDos as well, in case I want to start working on them. What a wonderful thing.

Orgzly in action on Android.

Working with the agenda, I realized I’m faster than before. At work, the first thing I do after I launch Emacs is to get into agenda view. From there, it is much faster to “tab in” to whatever task I need. This replaced my need for C-x C-r, recentf, since now I was not only in the file I need but also in the section I need. Even better: C-x n s can be used then to “zoom in” to the task at hand, blocking out the long list of other items.

In agenda, I was also able to quickly see tags and categories, edit properties, and most importantly, quickly schedule ToDos. As the agenda became my bread and butter, another small issue surfaced: the category property. Since I still used one file, I used the category property to differentiate between personal and work tasks. On the agenda, each scheduled task was placed in a “Tasks” category by default, since that was the file I was using. Even when I did assign a category from the agenda, I still had to do so for a child header (I thought these are supposed to be inherited, but this didn’t seem to work). This may sound like aesthetics, but being able to filter out all personal/work tasks can be very handy.

I was also thinking of my issue of having multiple values for one property. This was a problem I was trying to solve for the last several weeks. As an example, consider a task of setting up several computers, where extra information such as serial numbers and models is needed. Up to that point, I had a header nested inside my task.org file, like “set up 4 computers.” I used the custom property :Serials: followed by the serial numbers for set-up tasks: :Serials: 1111 2222 3333 4444 etc. This didn’t’ work well, since Org considered the whole thing, including spaces, as the value of the property “Serials.” I could still search, using Swiper (Ivy is one of the first things I install), for the serial as text, but anything that has to do with properties as functions did not work. I asked about this on Reddit and in IRC several times, but could not find a satisfactory solution to breaking down properties that way. Someone, at one point, offered a rather complicated function – but
I kept feeling this was too common of an issue to be overlooked like that. Something more fundamental was off in the way I was working with Org, but I didn’t know what it was. So I decided to “go back” and reflect again on how Org was meant to work originally.

After re-reading some sections of the manual and watching Carsten Dominik’s presentation back from 2008, I was reminded of Org-mode’s original built-in tools. In the lecture, Carsten emphasized Column View in Org. It is something I saw in passing previously, but now that I was having a mess on my hands with properties, Column View glowed in a welcoming aura. A quick and efficient way to have a table of the computers’ serial numbers (or users, or model…) right next to the header, in a comfortable layout? I needed this. I could set columns per header if I wanted to, which meant more fiddling around with the specific laptop-setup task in my old method. Or… I could just have a set-up dedicated org file which will already have the column view for laptop-set up built in with the properties… Wait a minute.

Dedicated File…Dedicated Filesss… Filessssss… BOOM.

Using Multiple Org Files

That was it. Everything I learned to this point came together in a torrent of thoughts. The solution to the properties problem was to have dedicated org files for each big project. After all, that’s exactly what the agenda was made for: to be used as the “glue” between them. This was why I was supposed to use the agenda in the first place! As long as I had a task scheduled, it didn’t matter what file it was in. All I need to do is to tab in, just like I did with tasks.org. The idea of having a list of tasks in one file was so ingrained in my head from all the apps I used in the past, I was blind to see what was under my nose.

With the realization of needing to break tasks back to different files came the realization that I am probably squeezing too much data into one org file. After all, setting up laptops as a task is a project in itself. It should include a heading for each laptop, along with a checklist of steps I do for each, with a log describing different issues I am having in the process. From the agenda, this looks just like another project I do during the week. it fits inside the tasks list and on my phone just the same.

I was in for another pleasant surprise: In my agenda, since now I separated the setups to a different file, my category problem suddenly disappeared as well. All my laptop setups were shown as “setups” as a category because this is the file they are in. Wow. So this is what happens when you stop fighting something and start using it the way it was meant to be used.

Next Steps

I’ve been using the system above for almost a week now (this post has been a week in the making). Here are some ideas about what’s next.

  1. adoptive capture template for tickets this was done this morning! I now have a capture template that automatically prompts me for the properties needed for each ticket I need to work on. It is then filed as a ToDo task in my weekly “tasks.org”
  2. Should Figure out what to use tags for: work in progress. While categories have their place, tags are more fluid. For now, it seems like I’m gravitating toward creating “mind keywords” of certain topics or terms that I’m familiar with. These, in turn, should be good for searches since I think in these terms when I’m looking for something. For example, a task flagged with my boss’s name tells me this is a task he’s viewing actively, or a “wiki” tag tells me there’s some good info stored in the notes of the tasks that I should probably store for later.
  3. Learn to trust the system. It’s hard to let myself create tasks in different files. I still need my weekly “mind-dump” of a place where I throw in quick captures and tasks that are not big enough for their own file, but I should stop thinking of it as my weekly list of things I’m doing since it’s misleading. The agenda is what reflects that now.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.