#def programmingtag(blogger):

[another-normal-anomaly]

I want to flip a flag in a piece of code. The unit tests for that code are written so as to depend on the flag being turned off in places that really should not depend on it, including places that test the behavior where the flag is turned on. As in, they have the flag off for the first 3/4 of the test and then turn it on mid-test because if you turn it on before the test starts (you know, like it would be in the real world) something that looks completely unrelated breaks. I pointed out this problem to someone on the team that wrote this whole thing, and he wants me to fix it. Dude, I don't know this codebase, I don't know what your tests are supposed to be doing, I just know they're doing it fucking badly.

[another-normal-anomaly]

> load testing our monitoring service

> see how many pings we can send in a short time

> send all the pings to google.com because we know it's up and can handle them

> get blocked as suspected DDOS

> I don't know what I expected.gif

[another-normal-anomaly]

Working on a hobby programming project is an excellent cure for being blocked on stuff at work all day. 

[another-normal-anomaly]

@evolution-is-just-a-theorem and I just compiled our very first program in our new language! True, the entire program was “fn a(b): \n   x = 5 + 6″, but it still counts!

[another-normal-anomaly]

I deleted 7000+ lines of zombie code today!

[another-normal-anomaly]

So I've got this user data at work, right? And I need to mung it, store the munged form in a database, then fetch it back later, unmung it, and give it back to the user exactly as they gave it to me.

I have a library with a bunch of functions named stuff like Mung(), Mung1(), MungExceptWhenFoo(), and MungClockwise(); and other functions named UnMung() (//Warning: not the exact inverse of Mung()), Unmung4(), and UnMungWiddershins(). I spent most of last Friday and all of today looking for a pair of those that 1) does exactly what I want and not any other thing, and 2) are the inverses of each other, but everything that worked on the test data screwed up the real data and gave back stuff different from what it took in.

I finally realized this afternoon that MungClockwise() and UnMungWiddershins() *totally are* the inverse of each other, and the problem is that users have been giving me strings that are already partially or entirely munged. So when I fully mung and then unmung them they end up different from how they started.

[another-normal-anomaly]

Today I discovered a piece of code that will display a message if and only if that message is a single specific string. I have not yet tracked down the responsible party and demanded satisfaction on the field of honor, but I haven’t ruled it out either.

[another-normal-anomaly]

Unit Test Stockholm Syndrome

When the tests fail so many times that you get used to the pattern [run tests->look at error->change stuff], and when the tests finally pass you're like "The heck am I supposed to do with this green 'Passed' message? What an uninformative error!"