Thursday, 4 October 2018

Get drunk and try to break shit

There is a whole world of discussion about the nature of Quality - not least as the key theme in ZAMM (Zen and the Art of Motorcycle Maintenance) by the late, great and much lamented Robert Pirsig. It's one of those things that we all think we understand until we have to define it. Many of the graduates from The Institute go into Food or Pharma to try their good-pair-of-hands at QC - Quality Control. That's a vital job, which sounds ever so slightly boring to butterfly me, but requires persistence, reliability and technical competence. Quality Assurance QA is sort of the same: I gather that QC concentrates on keeping the product to standard while QA is about maintaining, streamlining and idiot-proofing the process. In the software industry QAs are paid to "Get drunk and try to break shit"

I was browsing through metafilter on Sunday last and came across a surreal insider joke which a lot of commenters were getting off on but which I failed to understand. I share it with you here:"QA Engineer walks into a bar. Orders a beer. Orders 0 beers. Orders 999999999 beers. Orders a lizard. Orders -1 beers. Orders a sfdeljknesv". As always, check the comments for war-stories: "I took the latest build home for the weekend and tested it on one Mac and one PC (because those were the ones I owned). I came back on Monday with a list of over 400 bugs or other issues I’d found. The list was not happily received."

I should have got it because I was a software engineer back in the day although I didn't think of myself in those terms: not least because the software I wrote worked, was reasonably robust but the sky wasn't going to fall if it tripped out or went clunk. And I was my own QA insofar as I tried to think of the dopiest thing a user [almost always my later self or someone in the same lab] might try to do with my code and try to catch the problem before something blew up.

The one piece of software which I wrote for publication was, like all my code, was a kludge. Bits and pieces of code were clagged together into a menu-driven compendium to analyse a particular subset of DNA sequences. It was menu driven because clickable buttons weren't really a thing in 1992: certainly it was before the WWW. The version that worked (for me) presented a list of 9 options and paused after the instruction:
Enter a number between 1 and 9:
maybe the colon : blinked in anticipation. I had worked to ensure that an input of "2" fired up the 2nd lump of the program. But when I gave it to the students in the lab to see if it worked for them, one of them entered l "el" instead of 1 "one" and the whole thing stopped with deeply unhelpful

No comments:

Post a Comment