Done last week (20160724Su - 30Sa)

Fairly productive at work this week, though I lost Friday to an all-day training session. (I made up some of that yesterday in between lab work and my doctor's appointment. Went in early because the appointment wasn't until 1:40, and I wanted to be able to have my coffee before noon.) I appear to be in pretty good health; my blood pressure was 129/75; which is decent.

The training Friday was a workshop on Scrum. Tl;dr: we've been doing it wrong. Which is not unusual. My impression has always been that it works best for things that can be built incrementally -- the idea is to break things down into "features" (corresponding to "user stories") that can be built in one sprint -- typically two weeks -- and end up done, in production, and demonstrated to the customer at the end of that. The theory is that the team gets more and more familiar with their product and their process, so they get better at estimating. And there's an expectation that developers are mostly fungible -- anyone can pick up any of the tasks and finish it in a couple of days. (Specialists like QA, tech writer (we should be so lucky!), and maybe a web developer, don't count.)

So let's look at the project I'm currently on: We have four developers. One is building a new service, one is working on the web front end (and just came on board), and two are working in different, pre-existing services that they've never worked on before. The work being done in the latter case is such that a sizeable number of pieces have to be in place in order for anything to work. Meanwhile, other teams are working on other parts of the same services, with somewhat different requirements. Theoretically, each of the three main developers could work on any of the tasks, but in practice there's a lot of context in each of those sub-projects that it would take a long time for anyone else to ramp up on.

It doesn't help that the manager and web developer are in Vancouver, and that most of the design was done almost a year before the work started, under a different manager, by three developers one of whom got pulled off to work on a totally unrelated project. This leaves only two of us with any real context.

On the other hand, I've been having fun with configuration files and makefiles. The latest hack was adding color-coded labels to the workspaces in my xmonad setup. You say "ws 2 to.do", for example, and you get a color-coded label at the top of the screen in workspace 2. The labels use standard resistor color codes, and include a clock (because the quick thing was to base them on xclock). Here. (Need to get this onto github soon.)

Writing: met my minimum goal of 500 words two days a week, but just barely. Both were in PJ (short for Private Journal), so not on DW or the website where you can see them. Sorry about that.

( Notes & links, as usual )