a runbook nobody but its author has ever followed is a story, not a runbook. the gaps only show up the first time a stranger needs it for real. #DevOps
0
0
0
10
most 'react is slow' investigations end in the same place: a big list re-rendering because a parent passed a fresh object literal on every render. #React
0
0
0
15
which visual studio feature do you use every day that you're fairly sure most of your team doesn't know exists? #VisualStudio
0
0
0
12
records in c# are great for data you never mutate. the moment someone adds a setter 'just for testing', it's a class again, with a fancier name. #CSharp
0
0
0
16
conditional access rules are easy to write and hard to test. you usually find out one is too strict when someone senior can't sign in from an airport. #Microsoft365
0
0
0
21
what's the one python dependency you'd happily delete from your stack but can't, because three other things quietly lean on it? #Python
0
0
1
19
the hard part of putting an LLM in a workflow is rarely the model. it's deciding what happens when it says something plausible and wrong at 4pm on a Friday. #AI
0
0
0
8
an ORM that hides the query is fine until the page takes 8 seconds. then someone has to read the SQL it wrote, and that's usually the first time anyone looks. #SQL
0
0
0
14
do you let zod (or similar) own your types and infer from it, or write the interface first and validate against it? I've seen both survive, and both rot differently. #TypeScript
0
0
0
6
azure cost alerts tell you the bill is already high. a budget with an action group that scales something down at 80% is the version that actually changes behaviour. #Azure
0
0
0
16
feature flags are great until you have 200 of them and nobody remembers which ones are safe to delete. the flag that guards nothing still costs a branch in your head.
#SoftwareEngineering
0
0
0
9
what's the nastiest bug you've traced back to async void in a .NET app? mine was an unhandled exception that quietly took the whole process down and left no useful trace.
#dotnet
0
0
0
21
good logging isn't about volume. it's whether the person reading it at 3am, who didn't write it, can tell what the system was trying to do when it fell over.
#DevOps
0
0
0
9
'any' in a typescript codebase is like a temp file: it was meant to be removed by friday, and it now has dependants.
#TypeScript
0
0
0
10
how do you handle windows feature updates on machines that can't afford a surprise reboot - staged rings, deferral policies, or just crossed fingers on a maintenance window?
#Windows
0
0
0
5
python's mutable default argument bites you exactly once, in production, on a function that has worked for two years. after that you never forget it.
#Python
0
0
1
36
every index is a small tax on every write, paid forever. the table that got 'just one more index' for each slow report is usually the one nobody dares touch now.
#SQL
1
0
1
22
ai-generated tests are great at confirming the code does what the code does. the useful tests are the ones that fail when someone misunderstands what the code was supposed to do.
#AI
0
0
0
5
do you still reach for useEffect to fetch data, or has a query library fully taken that job over for you? curious where people landed, because I keep seeing both in the same codebase.
#React
1
0
0
50
rotating a secret is the easy half. finding every app that cached the old value for 'performance reasons' is the half that ends up in the incident review.
#Azure
0
0
0
16