Two of the biggest sources of incidents I’ve seen are:
1. Legacy code
2. Migrating away from legacy code

The conclusion is clear: you should only write non-legacy code

@norootcause My personal definition of "legacy code" is "code that has customers" so, yes
@nat No customers, no incidents!
@nat @norootcause someone at an open space put it, “the thing you hate the most in your codebase is probably also the thing that makes you the most money”
@norootcause only write code (or do any work/job) that can’t cause an incident.
@becomingwisest @norootcause can't believe no one's thought of this
@norootcause While we’re at it, we should improve quality by writing code with fewer bugs.
@stuartmarks Right??? We should just stop writing bugs altogether.
@norootcause @stuartmarks While we're at it, we should also be more clairvoyant so we never need to deprecate anything
@jawnsy @stuartmarks The real root cause of incidents is our inability to see into the future.
@norootcause @stuartmarks Wow Lorin, I think you've just solved the problems of several fields, let's pack it in, we've figured everything out!!!
@norootcause @jawnsy That’s simple to fix: just use AI.
@stuartmarks @norootcause Sprinkle some AI on it! Yaaassa

@jawnsy @norootcause AI MPD FTW!

(MPD = magic pixie dust)

Sean Murthy (@[email protected])

All code is legacy code. Eventually. #code #software

Hachyderm.io
@smurthys Not all code! Only code that still gets used.
@norootcause this isn’t a choice of old code with bugs or write new code with bugs it is about doing a better job of designing and writing applications. Decades of research and knowledge and we still make the same mistakes.
@norootcause Also, version 1 is typically buggy and in need of refinement. Therefore always start with version 2.
@norootcause You joke, but I've seen people write instant legacy code, no maturing required.
@norootcause i strive to make 2. my legacy
@norootcause where "legacy code" is "any code written by someone else, or written by you more than 6mos ago"