adegoodyer.com
DevOps and Infra Developer 👨💻
BSc Software Engineering 👨🎓
K8s, Go and Linux 🖥️
Eight blind men and an elephant 🧘♂️
Pro European 🇪🇺
74 posts
33 followers
47 following
Regular Contributor
Active Commenter
comment in response to
post
The long way to a small angry planet - magical
comment in response to
post
I know you like book recommendations and feedback so I'll recommend 'The Measure'
comment in response to
post
Time is one resource that cannot be borrowed or bartered.
Spend it like it's the most precious thing you have, because it is.
comment in response to
post
This is exactly what gad feels like. Could also be a bit of pre-burnout behaviour.
Either way, you can tackle that feeling of overstimulation the same way by developing good sleep hygiene aligned to you circadian rhythm.
Happy to elaborate if anyone needs.
comment in response to
post
docker-slim?
comment in response to
post
Before everyone piles in.. let's just remember yaml was supposed to be human readable and flexible from the get go.
comment in response to
post
👋
comment in response to
post
Shout out to fzf for fuzzy searching as well and it's rapid.
comment in response to
post
Not come across mosh before, having a peek now 👀
You should checkout Twingate as an alternative to Tailscale, great with k8s.
comment in response to
post
VSCode Remote Explorer is actually nice and convenient to drop right into where you need to be.
I mix vscode, nvim, vim, tmux, ssh config etc in my workflow.. whatever I need to do tbh.
You can call it laziness or you can call it efficiency - I don't mind.
comment in response to
post
I have a special vinyl reserved for this moment - probably 4-5 times a year.
Lights out, volume high, press play and dance like nobody's watching 🕺😄
youtu.be/TP9luRtEqjc?...
You are one in a million.
comment in response to
post
Or worse.. Bastions!
Have you tried Twingate as an alternative to Tailscale?
I switched almost a year ago, but both are pretty decent options tbh.
comment in response to
post
Agree, being volatile and unpredictable that also brings in a new factor to whichever codebase I'm working on.
Test generation is imo probably one of the best ways of utilising llms whist implementing code.
Perhaps not unit level if you're strict tdd, but definitely integration tests.
comment in response to
post
Cursor seems to spread the context out as far as I can tell (via ChatGPT or Claude - with the latter being larger) so gave better project based results from my experience.
comment in response to
post
Same experience here, but also with all LLM's in general - you can push them to a point and then they collapse and start getting confused/hallucinating.
Ideally the ability to set rollback points in your code would probably improve this beyond editing previous queries.
comment in response to
post
I'm almost anti-ai assisted programming these days though - good for exploration and prototyping.
Can certainly make you more productive in the short term, but I never feel fully connected to any generated code I write and so it's definitely more error prone and harder to recall months later.
comment in response to
post
It's the choice of LLM per interaction that's more important and you can easily select each time, so that's nice.
ChatGPT and Claude seem to be the leaders atm and both have their strengths.
comment in response to
post
I've recently gone back to Copilot on both vscode and neovim and can recommend.
Code completion suggestions as well as inline chat in both editor and terminal.
Played with Cursor for a while but if you think the extensions on vscode are becoming bad then that's like the wild west.
comment in response to
post
Knowing you've solved a users pain point is that they become mute to talk about afterwards.
If you can also prove that through data then you've thoroughly understood domain, problem and solution.
It's not magic tho - good communication, experience and tight feedback loops.
comment in response to
post
Sincere congratulations Sam, it's a joy to read everything you write.
You've set your own standard - never let that drop, always wonder.
So effing proud of you 🫶
comment in response to
post
Grafana would be my natural choice but if you've no CLI access then probably a no go unless you're provider has enabled if previously.
Simplest option is probably MySQL Workbench > Server > Dashboard (and possibly Performance Reports with the right grants).
comment in response to
post
Should be the default option imo
comment in response to
post
You're missing the best bit! 😊
Could have just lost it's non-stick coating but I've always found after it's clicked done, turn off at the plug and leaving for 15mins works for me.
Seems to give nicer rice as well.
Oh and always rinse your rice three times before hand to clean and remove starch.
comment in response to
post
If you're in that space and like surprising films then I'd definitely recommend m.imdb.com/title/tt3007...
comment in response to
post
Don't get me started on the weird ways that MacOS operates.
/ (aka root) is read only.
Networking is a ball ache.
Bleddy .DS_Store !!
Non GNU tooling.
Brew updating everything whenever I just want to install a package.
Containers.
comment in response to
post
Do you believe in quantum?
I'm a sceptic, but part of me thinks they might've skipped a battle.
comment in response to
post
What have you currently chosen to manage your infra?
comment in response to
post
The amount of time spent implementing, maintaining and testing Terraform ultimately turned me off of it for good. That cost is particularly massive when you consider who you're employing to do that work.
comment in response to
post
Hate is a strong word, especially without elaboration but I hear you.
For me, I actually liked the concept of Terraform initially - an API wrapper essentially and was an improvement on Ansible etc.
HCL (not great) and multiple state management (desired, actual, true) made me start to question it.
comment in response to
post
The tightest of feedback loops - this is the way
comment in response to
post
How good is the trailer btw!
The main vocab is a poem by Rudyard Kipling called 'Boots' but no idea about the origins of what the woman is saying or how either is related..
7 6 11 5 4 17 32 the day before..
Spooky! 🫣
comment in response to
post
Blind in-app notifications I think we're all pretty ignorant to now (along with ads, cookie notifications etc), but contextually aware and intelligent prompting like you suggested - I'd be a big advocate of that.
comment in response to
post
Wish I would've screen grabbed it now.
But it also made me stop and think - what else could be occurring here.
Having detected that behavioural loop - that's the ideal place to provide further information or prompt for feedback for example.
comment in response to
post
Seeing this in the wild.. this morning I started using Google Keep again and was copy/pasting in links for later reference.
After a few times was pleasantly surprised when a small dialogue appeared underneath the last list asking if I wanted to install the extension to speed up my workflow.
comment in response to
post
Thanks Richard, much appreciated.
I wanted it to look clean and not overload the viewer with unnecessary information so am pleased with that.
comment in response to
post
What's also pretty great is that fzf indexes on the fly and is still miles faster that most IDE's (which will often crash when searching large codebases or number of repos anyway).
Rust is so good at this type of tooling.
comment in response to
post
Nice! Thank you
comment in response to
post
Agree!
Also very handy for finding/searching logs/configs..
fzf github.com/junegunn/fzf
fd github.com/sharkdp/fd
rg github.com/BurntSushi/r...