Worst case scenario

August 4, 2009 Adam Milligan

Years ago, after I finished college but before I started working professionally with software, I spent a couple years working as a paramedic. I learned a lot from that job, not least about interacting with people who really, really don’t want you in their lives.

One of the calls I remember most vividly happened around three in the morning, not long after schools had let out for the summer. A group of recently graduated high school girls had rolled their Ford Bronco on the highway. When we arrived an engine company was on scene, busily cutting the remains of the car into fun size Bronco strips. I followed the trajectory implied by the hole in the windshield and found my patient, the driver, on the pavement some distance from the car.

While I set about preparing to package her up for a quick trip to the hospital another engine company arrived. As I started my cursory physical exam the lieutenant rushed over and demanded I stop. To understand his reasoning you have to realize that the process of emergency medicine affords little or no dignity to trauma patients: life comes before limb; modesty comes well down the list.

So, what the fire lieutenant objected to was that I was cutting the clothes off a sixteen year old girl in the middle of the highway, directly illuminated by the halogen scene lights from our bus. He demanded that we package and transport the patient fully clothed. If you haven’t worked with firefighters, know this: they travel in packs. My partner was hanging IV bags and working the radio for another bus, so it was me, a supine patient, and five firefighters. They got what they wanted.

Of course, to protect trauma patients’ spines you have to package them fairly thoroughly. You basically strap them down to a six foot board so they can’t move, and once you finish it’s pretty much impossible to get their clothes off or do a half decent physical exam without jeopardizing their spinal cords. Which means, as the attending medic, when we got to the ED I was the one who had to explain to the trauma surgeon and ED physician why I was handing over a patient who could have had a piece of windshield glass the size of a grapefruit sticking into her kidney for all I knew. Not a shining career moment.

I remember that call not because the patient was badly hurt (just some broken bones; she was lucky), or because I made a huge difference in someone’s life, but because of the lieu’s argument for not doing the full, expected, physical exam. As his minions packaged my patient like gift-wrappers at Macy’s, and my partner made a break for the driver’s seat, he told me these exact words:

“She’s suffered enough trauma. She’ll be okay.”

Now, I knew at the time he was quite likely right (her chances having escaped major injuries were actually better than you might imagine), and it’s actually quite difficult to explain to patients why you have to cut their clothes off (try it some time), poke them, shock them, or tie them down. It’s very, very tempting to do the easy thing in these cases, and 99 out of 100 times if you avoid making the patient, or yourself, uncomfortable they do fine. But, one out of 100 times the patient dies from a ruptured spleen or flash pulmonary edema. That’s the worst case scenario; and that’s the only scenario medics really care about.

Now (finally on to my point, dear reader), I recently submitted a patch to Rails that I, and many of my colleagues, believe will help prevent invalid functional tests, and therefore prevent bugs. The response from the Rails core team:

The vast bulk of tests just pass the id in those places and they’ll work fine. Users overloading to_param are in the minority and we shouldn’t spam everyone else just to satisfy them.

Ignoring for a moment the somewhat unscientific characterization of the prevalence of #to_param overriding, the message is that the patch is potentially annoying for people with improperly written tests and most of those improperly written tests probably won’t be a problem. It’s more comfortable for the Rails team to let broken tests slide and hope all will be well, than to bother developers over the cases in which broken tests lead to broken production.

I realize that comparing HTTP 500 responses to ruptured spleens might seem overly draconian. But, consider this: while the consequences of doing the wrong thing in software are much less dire compared to medicine, it’s so much easier to do the right thing. If you doubt this, drop me a line; I’ll be happy to nasally intubate you. I guarantee after that experience spending half an hour fixing your controller tests won’t seem so bad.

About the Author


Full Text Search with Rails
Full Text Search with Rails

Wolfram Arnold shares his experiences implementing full-text search in a Rails-based translation applicatio...

"Missing host to link to!" Rails 2.3 Upgrade Issue
"Missing host to link to!" Rails 2.3 Upgrade Issue

During the process of upgrading a project from Rails 2.2.2 to Rails 2.3.2 several of our tests were breakin...