@erikras Conception sometimes feels like you're not programming and you're in a hurry to actually get to the programming part, but if you take shortcuts and the foundation is fragile the end product will suffer from it.
@erikras The dev equivalent would be conception. Applying masking time is time consuming but you can't take shortcuts, badly applied tape will create leaks and you will have an ugly paint job. It feels like you're not painting but it's a super important part of the job.
you choose to run. Anti-AI open source project are in no way opposed or an obstacle to malleable personal software, on the contrary. No programmer (AI assisted or not) is entitled to having their contributions upstreamed in an open source project regardless of how good it is.
The point of open-source was never to let everyone contribute. SQLite is one of the biggest and most successful open-source projects and don't let ANYONE contribute. The point of open-source is that anyone can modify the version of the software they choose to run.
"This is a protectionist tale as old as time. And the justifications are just as tired: It's about quality! It's about attribution! It's about workers! Spare me. It's about you, your insecurities, and your privileges." https://t.co/SP6DubrXXh
AI contributions (or contributions in general) are being restricted but in the vast majority of cases it's because the humans mass-opening pull requests are terrible at communication and are not competent. EVEN THEN it doesn't change the fact that you can fork and patch the copy
@TheGingerBill@speech_ka_@awesomekling@ladybirdbrowser You could resize the sidebar to only show a couple letters + the favicon I presume. It's "wasting" horizontal real estate just like regular tabs is "wasting" vertical real estate, and most screens (especially ultra-wide) have a LOT more horizontal real estate. But you can choose!
> yEaH BrO bUt RaYcAsT iS pAiD
free in @raycast 👇
manage windows
clipboard history
find files
calculate everything
create @calcom invites
control @Spotify
manage issues on @linear
join meetings
local models
block distractions with focus
bring your own key
stay up to date with @github notifications
create virtual AI assistants
check the current price of crytpo
control screen @screenstudio's recording
fire emoji cannon
pick emojis with natural language
browse your screenshots
create snippets
create quicklinks
open window/tabs/spaces and search @diabrowser
control @CleanShot
create reminders
search screenshots
make your text 𝓯𝓪𝓷𝓬𝔂
interact with @OpenAI chatgpt
extract text from images
take notes
check @vercel and @Netlify deployments
check the weather
translate using @Google translate
open tabs/windows/launch configurations from @warpdotdev
open/manage projects in @code
search/install brew packages
pick color
search @NotionHQ
search/open @1Password
search/see unread @SlackHQ messages
start @zoom meetings
search @Wikipedia
manage @cursor_ai agents
fully uninstall apps
monitor your @claudeai code usage
track flights
look up word definitions
convert text cases
eject discs
empty trash
automatically quit applications
practice your typing skills
search for gifs
set audio device
image transformations
view 2fa codes from imessage/email
browse google fonts
...plus thousands more
@peduarte Have you considered other frameworks than React or did you stick to it because of the extensions ecosystem? Compiled frameworks like Vue Vapor, Solid, or Svelte are a lot more efficient than React (both CPU and memory-wise) with granular signals and no VDOM
@mikker Would love to have your perspective on optimizing memory footprint since you work on Tuna. SSDs are really fast nowadays, if people can stream gigabytes of LLM weights while doing inference surely my launcher can stream the MBs of indexes and text only when it's needed
@mikker e.g. should the clipboard history be in-memory at *all time* not just loaded from disk when I use the command? even then 150k words of prose would probably be about 3MB, 1MB if you compress with zstd which is very fast to decompress. Even the MacBook Neo's SSD reads at 1600 MB/s