Another terrible rant from the inhereted Hydra source code. So deep in the dark dungeon of that code I noticed something interesting. They declare this INT32 array with an incredibly long (like 200 values) list of hard coded magic numbers. Something along the lines of:

INT32 array[200] = {-1,0,1,21,4,7,19,33...};

However, the resulting output was incorrect. After spending a fort night and a good chunk of my remaining sanity I had overcone the 437 levels of indirection left by the previous programmers, and narrowed it down to this line. But it looked perfectly fine.

I pull up the diffs and notice someone had checked in a change to the source. I track it to this line and find what the original data had been.

INT32 array[200] = {-1,0,1,2l,4,7,19,33...};

In VS the default font shows l and 1 as fucking identical. Someone had accidentally made that change to 21 from the original 2l and checked it in. I mean I can't really blame them. Who the fucking hell inatantiates a fucking int32 array and peppers in a fucking 2l (long) for no fucking reason?!

The single most asked question when you're a unwitting 'code archaeologist' working on an inherited code base:

'Why in the Lord's name can't I just start from scratch!?'

So at one point I worked on an inherited project that had the worst code I've ever seen. I mean bad, so bad there may no quantifiable measure that can accurately convey how bad. We ended up naming the thing 'the hydra', cause it had a million issues and they just kept growing as we fixed things. To my point, in C++ they implemented their own primitive type Boolean32 as a signed int32 pointer. If that wasn't enough they used it as an octal bit mask. They also switch the value using logical and / or between 2 numbers, 037777777777 and 000000000001. So essentially they only switch this value to 1 or -1 and end up comparing it to their own const true or false. In c++ any value not 0 is == true...apparently not in this code.