It Rains In Triplicate

I heard rumors that apple will come out with a new version of its iphone software and a new model iphone in June. I had to do some real digging to figure that out, including my own inside source (meaning, apple’s own press releases and public senior level interviews which tout the new version of the software and state that they intend to release new iphone models in the June timeframe)

Not that discussions about the features and the like isn’t interesting, but the rumor community is almost always either an order of magnitude overly optimistic or blazingly obviously.

Somedays I like the flexibility that unix provides, somedays I hate the convoluted ways that unix uses to provide said flexibility, and other days it is a mix. I think today is one of those days.

The problem I was trying to solve is that dreamhost installs subversion 1.4.x and my tools, increasingly, are looking for subversion 1.6 features. Instead of working around the incompatibility, I wanted to see if I could get a local 1.6 installation working under a shell account. Short answer is that it works, but requires some … interesting fiddling.

Steps:

1. Create a new dreamhost shell account to store SVN repos, we’ll call this account “repoacct”
2. Log into repoacct and download and compile subversion 1.6.1 as per http://wiki.dreamhost.com/Subversion_Installation with the minor correction that –with-zlib in the configuration command line should be –with-zlib=/usr/lib
3. Create an ssh key and put install it in the repoacct authorized keys, as per http://wiki.dreamhost.com/SSH#Passwordless_Login (if you don’t want to be password-less, set a password on the public key, but the next step needs the public key installed)
4. (convoluted step) modify authorized_keys as per http://svn.collab.net/repos/svn/trunk/notes/ssh-tricks. Specifically, I modified my key’s line to read:

command=”/path/t/svn-install/bin/svnserve -t -r /path/to/repo/directory” ssh-rsa <my key> seth@machine.name

The dirty simple installation, where I just install subversion and add it to my local path does not set svnserve, so it still ends up using the 1.4.x binaries when I try to work with the repo. This method ensures my new svnserve binary is being used.

… but haven’t we, as a people, moved beyond sticking your hand in a diaper just to see if it is dirty?

I have a full drobo in my basement that is really being used as dead storage, storage for files that I do not want to lose, but do not need to access any more than, say, once a year. Ideally, I’d burn this stuff on tape or optical media and throw it in storage or a safe deposit box, instead of keeping it in a RAID or RAID-like box.

Tape is way too expensive, on the order of thousands to get a good setup, and current optical is too small. Individual drives are too unreliable. You’d think dead storage would be easier or cheaper than live storage …

Two more pieces …

I’ve been doing the review of my task items on a monthly basis. As a result, the process has been about rejiggering due dates, not cutting irrelevant items and working priorities. So, I’m moving to a monthly review.

Second, an issue. Previously, I could recognize my progress (or lack thereof) by whether or not I finished everything on my daily list, and, more importantly, how many things I had to defer.

So, while in this iteration, I, hopefully, won’t have anything artificially due, so, hopefully, less deferring, but also a limited ability to track progress. I wonder if it is worth throwing together an applescript to post how many tasks I finish in a given day or week.

So I am more seriously working on a GTD process with OmniFocus, abandoning my tried and true, yet broken and distracting strictly date ordered ToDo lists for a proper GTD, next action-based practice.

Since there are new, interesting, and messy drains on my time, yet I’m only working a half time schedule for the next month or so, now is a good time to rework my organization syste ..

So, I’m trying to document it here, if only so I can go back and laugh at myself if it fails terribly and I go back to my daily TODO list.

The goal is to be able to work more fluidly, not to have to schedule all items I need to accomplish during the day, but, instead, all me to juggle priorities in real time. There are some items that are date driven, but my entire day should no longer be date driven.

So, three parts to discuss, tags / context structure, project structure, and workflow (and lists)

CONTEXTS

Ideally, contexts are used to determine what action I will accomplish, based on my location, environment, etc.

In terms of contexts, I used to have a few basic ones. Office, Lab, Home, Anywhere. There was a 5:1 split between anywhere items and everything else. The goal is to have a reasonable split and some better granularity within the categories.

The new split has five top level categories / folders: Anywhere, Home, Store, Work, and Waiting For.

Anywhere is now split between Email, Computer, Reading, Thinking, and Calls.

Store is split between various categories of stores, i.e., Market, Baby Store, Electronics Store, Pharmacy

Work is split between work locations

Waiting For is inspired directly by GTD, which are those actions that are “sitting around” actions, where I am waiting for input from someone before I can continue on with a project.

PROJECTS

My project structure was pretty broad and flat, which made a next-action structure meaningless. I had a few top level areas of responsibility, such as Work Stuff, Projects, Personal, and Writing. Under each of these AORs, I would have a bunch of individual projects, with lots of different tasks. The tasks would be further organized into folders within projects, but the dependencies between the tasks was unclear.

So, I have a two tier AOR system. The first is the general category, right now Personal, Projects, Work, House, Writing.

Within each category are projects or further AOR categories. For example, my major work project <PR> has a bunch of folders for each major product line I’m working. Within those folders are actually projects and tasks. The idea is that a project should be a sequential set of steps, not just an unordered bag of tasks. Projects and their contents should be fluid and short lived, not immutable and long lived.

I also have incubating projects collected, but mark as “on hold” so they don’t clutter my workflow.

For those ideas that are really just incubating (i.e., a couple of sentences that I want to go after someday), are collected into another “on hold” project for each major category above.

WORKFLOW

So, where does the rubber meet the road?

There are two major types of tasks I want to accomplish, those that are due today (time critical) and those that progress my goals / projects to completion. Further, I need to capture, categorize, and promptly forget about any pending tasks that are not the next thing I need to do on a goal / project or due today.

So, workflow …

First, the INBOX. Anything new ends up being immediately captured and categorized, or ends up in an inbox. I have three inboxes:

1. if I’m in my car or away from a computer, I use my JOTT account to capture a voice note, which is emailed to an account that deposits the note into the OmniFocus INBOX
2. if I’m in a place where I can grab my phone, I’ll use OmniFocus for iPhone to write a quick note into my OmniFocus INBOX
3. If I am in a meeting, I’ll mark potential TODO items in my notebook and copy them into OmniFocus at the end of the day

Second, I have a bunch of perspectives that I use.

For day-to-day work, I have three, besides the INBOX:

1. Due Today, which is a list of everything that is flagged or marked as due today. I have daily / weekly / monthly / yearly repeating actions and actions that have a real due date. I can use flags as a short cut for things that I nominate for getting done today, without having to put in artificial due dates
2. Next Action, which is the next action for all projects, regardless of AOR, that are marked as active (i.e., not completed, on hold, or dropped).
3. Waiting For, which are those items that are pending someone else’s input

Finally, I have a Review list for the weekly review and the raw Projects and Contexts list, which show everything, either from a project or context perspective.

So, on a daily basis, I’ll start with my Due Today items. Once I finish those, I’ll go on to my next actions. Interrupt driven taskers will come and go dealt with or recorded.

Issues

So, the only issue I see is that my daily items sometimes have more context than just the day they are due. i.e., one of my daily tasks is to enter my hours for the day. That isn’t relevant until the evening. One solution is to divide up my Due Today list into Daily Checkin / Checkout lists.

come from the factory 80% full, just like hard drives.