Step 1: Create and deploy feature. Get feedback: It works fine!

A year passes...

Step 2: Add another small feature that is completely independent from previous feature.

Step 3: Notice that feature #1 is broken.

Why the HELL did it stop working?!? I didn't change anything related to that feature. I can't explain that!

Step 4: Examine bug and find out something unknown about the technology used in feature #1.

Why the HELL did that ever work in the first place?!?

Step 5: Get time to fix it. Don't get time to find out why it worked for a year...

The problem is solved, the mystery remains...

I want more details!

Schrodinger's bug

@filthyranter
In the case that got me to post this, we had the lucky case of roughly finding out what happened.

It's about and industrial sensor. We got the runmode configuration wrong and it should not have worked, but we also used sensor commands at certain moments that were not intended to be used while in that runmode and those commands forced the sensor to step out of his current runcycle and start anew, which basically "emulated" the behaviour that we wanted from the orginally intended runmode.

It's like getting someone to sing a song from the beginning by choking him into unconsciousness for a moment instead of just asking him.

Every analogy i can come up with is cruel and somehow i really feel a little sorry for that little sensor. :-(

@CWins Okay, now I also wonder how the fuck that went well for a whole year.

Also, welcome to devRant!