bijancn.bsky.social
Senior Staff Engineer | PhD Physics
22 posts
28 followers
130 following
Active Commenter
comment in response to
post
Ironically there are no MCPs for connecting to LSPs AFAIK even though it would make a ton of sense (AI could lookup type informations or trigger exact refactorings)
comment in response to
post
But that’s so far away from the keyboard? What am I missing? Why isn’t this a layer?
comment in response to
post
If it hurts, do it more often 😅
comment in response to
post
ChatGPT ultra will be a hitman controlling drones. For just 100 K per month you can get rid off the competition for good
comment in response to
post
zsh and a fast „o-my-zsh“ clone and you will see the light
comment in response to
post
I would increase their budget and forbid them from having a car 😅.
comment in response to
post
Better start a business 🙃
comment in response to
post
I don’t think fetching data should be this hard 😅
comment in response to
post
I bet the complexity of the pipeline has grown so large over the years that no one dares to touch it at this point anymore. I saw something similar from Shopify and it was painful to even look at
comment in response to
post
Well. The data has to be extracted, transformed and loaded. Then it lands in the data lake. The data pipeline then pumps it to the data warehouse. Finally massive quality checks run which fail often so the shadow data warehouse doesn’t actually go live until the failures get resolved
comment in response to
post
What’s that app?
comment in response to
post
So you tell people to use Kafka?
comment in response to
post
Happened recently to me. Overoptimized my Supabase DB trying to fix the 500ms delay until I realized the throttle to slow 4G was still on
comment in response to
post
It’s like Ikea but with higher quality base materials and no instructions. You can build a cathedral with it. Or something that barely stands
comment in response to
post
Just tried it on my app (was also thinking about it these days). Went pretty well. Like 80 % of the screens look good, some need tweaking. And the state seems sometimes wonky but that's also fixable.
comment in response to
post
Ask cursor with Sonnet?
comment in response to
post
You are allowed to do 5 of everything manually 😉
comment in response to
post
It’s really great for queries. Sadly RLS isn’t so popular in training data so it has a hard time getting schemas right
comment in response to
post
And they should be!
comment in response to
post
Didn’t even get through the first episode
comment in response to
post
Thriller!
comment in response to
post
Condolences