All too commonly, people work on things they despise. Programmers are no different, especially since computers are particularly suited to repetitive tasks. I experienced it recently but devised a challenge helped me overcome it.
I was recently given a task so monotonous, it would make any programmer miserable. I was asked to check a series of stock codes to find which were singularly-listed (i.e. listed on only one stock exchange). Presented with 160 rows of Excel data, I was expected to search Reuters for each stock code, one-by-one by hand.
Having experienced similar soul-destroying data entry work in previous employment, with half a Computer Science degree under my belt, I took a different route and wrote an algorithm to do the work for me. Work smart, not hard.
I took the task and devised a way of making it interesting. I challenged myself to write an algorithm, using full test-driven development (TDD) principles, in a language I have little experience with (Python 3), all within one hour.
After clarifying the requirements, I outlined the core algorithm on paper. I’d save time by manually copy/pasting the stock codes into a plain-text file and reading in each to query against Reuters. Closer inspection of the Reuters page showed a redirect occurred when the stock was singularly listed. I therefore planned to listen for the 301/302 HTTP redirect code on each request and log each triggering stock to a file.
It may have taken longer, but the large majority of those three hours were of solid focus; “wired in” as it’s known in The Social Network.
Note, this was three hours of elapsed time (rather than sole keyboard bashing). Why? Because “how do you account for interruptions, unpredictable bugs [and] status meetings…”? You can’t, really.
Only the people who actually write the code — the developers — can estimate how long a task will take, but only by recording honest velocity (the division of the estimate by the actual time spent) do the developer’s estimates hold any weight. Experience makes you a better estimator.
Without going into too much depth, Python 3’s library module for requesting web
urllib.request — has it’s limitations. Since redirect responses
are silently ignored, I found a mature third-party module —
httplib2 — to overcome this.
Despite strong praise in Dive Into Python 3 (DIP3, my main and excellent
resource for learning Python 3), I ran into a few bugs that prevented me from
using it. I lost around 1.5 hours troubleshooting, settling instead to redesign
my algorithm to scrape the Reuters page, compromising increased complexity for
Python 3 is a very clean and expressional language, but as the author of DIP3 reminds us, it’s still very much in active development, with a low adoption rate (so far) over Python 2.
Having a customised, pre-configured setup particularly helped when the pressure was on. Everything was at hand — windows managed by xmonad, complete editing efficiency with vim, rounded off with continuous integration testing using nosier — helping me concentrate on the task at hand.