OpenAI just released the largest-ever report on how consumers are using ChatGPT.
It looks at millions of (anonymized) messages over the last 3 years to analyze usage evolution over time.
Key takeaways 👇
FlashList V2 is a complete rework and is in alpha. Please feel free to give it a shot and raise issues. We're going to continue to optimize it.
V2 is fast and doesn't require any estimates! It has many new features! Read the following for more details: https://t.co/bjoQzvZm64
Why do ethnic minorities in the UK live in households that spend so much more income on housing than White British adults? Turns out the answer is complex.
Our latest report, released today, dives into the root causes. Let’s explore the key insights👇🏾
Most of the time that we get a client coming to us complaining of performance problems with React Native, it ends up being something from the following:
- massive unnecessary rerenders because they're not following basic React principles (React Compiler will solve a lot of this, but isn't magical!)
- poor understanding of state management, using context poorly
- using old versions of React Native, not keeping up with perf improvements
- super slow back ends, and no graceful handling of this on the UI
- super heavy unoptimized images and fonts
- missing really basic configuration (cache, inline requires ... more on cache later)
- waiting for all API requests to finish before rendering (skeletons help, as well as Suspense)
- unoptimized lists, not virtualized, inefficient (waiting for @jmeistrich to solve this forever)
- rendering far too many unnecessary views, not using fragments
- loading heavy native modules on app start, some of which are rarely used (TurboModules helps...!)
- not using cache, or having cache misses that shouldn't happen (and not understanding this was happening)
- not prewarming the cache for likely user flows
- overloading initial renders by drawing offscreen/non-critical content instead of using lazy loading
- not using React Navigation's new preload
- excessive loading spinners (for shorter load times, loading spinners increase perception of slowness; avoiding them altogether or doing something more subtle makes it seem faster!)
- using third-party libraries without understanding their performance tradeoffs
- doing stuff on the UI thread that they shouldn't
I'm probably missing some, but this is what came to mind this morning!
These are mostly due to lack of knowledge of React as well as not measuring as they go, so small perf problems pile up and they don't notice until it's a real ball of yarn to untangle.
Hey @mrousavy, with all the special knowledge you got with custom video pipelines, can you tell me how Instagram and TikTok can encode videos so incredibly fucking fast? Me trying out stuff with ffmpeg on device is not even close compared to those big players.
React Native 0.76 will ship with a single merged dynamic library called `https://t.co/w2nNNDov97`:
https://t.co/peZ08rvbtS
This comes with major space savings for users as well as performance wins
@_iamodb Agboola na weyrey 😹 after the test I remember a guy from 400L who tutored us in that course back then. While he was teaching us, Agboola walked by and saw him solving, not knowing that the guy was tutoring us, Agboola said “you this boy, you want to fail again” 😹