Chat Logs

  1. * phaleth joined #primate
  2. * blue_ joined #primate
  3. blue_hi phaleth
  4. phalethhi blue
  5. blue_so I had a discussion with my brother yesterday about how clankerdom stifles innovation
  6. phaleththere will be less and less people learning all sorts of skills, not just software development
  7. blue_like, why do people keep using next despite its terrible record with rsc (cve of 10.0!) and the general fact that it's a terrible piece of software
  8. blue_because the clanker recommends it
  9. blue_basically at this point you're better off using vanilla js and html than next
  10. phalethand php and raw sql
  11. phalethall in a single php script
  12. phaleth:)
  13. blue_yes, perfect
  14. phalethin any case nextjs got picked up by many because it had exquisite docs
  15. blue_anything better than this clanker infested nonsense
  16. phalethone does really have to think like a junior dev and describe things that way
  17. blue_look at this nonsense
  18. blue_phaleth: https://dpaste.com/BZWAEHADJ#wrap
  19. blue_this is what a vanilla next installation produces
  20. blue_you can't argue that's sane
  21. blue_phaleth: https://dpaste.com/GP9BAKEET#wrap
  22. blue_that's what a comparable primate (+react) installation produces
  23. phalethlooks like they use tailwind
  24. blue_and the bulk of those bytes are just nonsense headers that my dumb browser sends
  25. blue_that's why primate takes <0s to load on comp, whilst next is around 2-3s
  26. phalethnowadays next is obviously able to hide a lot of important details behind their easy to use framework
  27. blue_if someone has any shred of understanding of security, he would never let rsc run in any of his apps
  28. blue_the complexity of that is terrible
  29. blue_and for no gain
  30. phaleththey probably just state next is blazing fast due to having all those rendering strategies and that it
  31. blue_I don't even understand this chunking nonsense that next does
  32. blue_the esbuild approach is easier: you chunk only on dynamic imports
  33. blue_it's kinda funny it's 2026 and people still use subpar software for everything
  34. blue_anyway, I'm gonna upgrade the RP repo to primate 0.38
  35. blue_it's still on primate .36 and I wanna continue dev
  36. phalethit's 2026 and people just want an exe
  37. blue_phaleth, did you see this
  38. blue_https://github.com/primate-run/primate/commit/485cba0b86144c60cff81802353156310ae44158
  39. nevetmove to tsgo · primate-run/primate@485cba0
  40. phalethdrag and drop it into the browser and it deploys
  41. blue_lol
  42. phalethyeah, I did
  43. phalethI assume that's also not breaking the primate website deployment from master branch
  44. blue_you assume correct
  45. blue_in fact, you overassume. apps/website does not use typescript directly
  46. blue_none of the apps does
  47. blue_typescript in the primate monorepo is used for two things: compiling packages, which is increasingly rare because I'm using live TS during, and type-checking in the editor
  48. blue_when you build apps/website, what happens is that esbuild pipes all of the TS into a single JS file
  49. phalethok, can you take a look at the PRs then?
  50. blue_I could, but I wanted to ask you. you've fixed this woff thing a few times, do I keep changing things that it keeps regressing?
  51. blue_I anyway wanted to find some solution for woff because I keep needing custom code for it and it's annoying me
  52. blue_https://github.com/primate-run/primate/blob/master/apps/website/config/Website.ts#L18-L28
  53. blue_I have 100% the same piece of code on repopack's website repo
  54. blue_I was thinking of maybe adding {loaders": Record} to the config
  55. blue_so you'd just do {loaders: {woff2: "file"}}
  56. blue_but that feels like circumventing the module API a bit
  57. blue_maybe the best way is providing a primate helper
  58. blue_`import loader from "primate/loader"; ... export default config({ modules: [loader("woff2", "file")] })
  59. blue_translates interally to a Module, but without all the boilerplate
  60. blue_but anyway your first PR only makes sense if woff2 is used
  61. blue_and a part of me feels back about hardcoding woff2 support into the code
  62. blue_feels bad*
  63. blue_what does the patch do, create a preload script tag for woff2?
  64. blue_do we need this? esbuild inlines woff2 with the module I just showed
  65. phalethyou can move that from the core if you want to, that's totally up to you
  66. phalethI'm just fixing it again, but it's not really a regression
  67. phaleththis time the problem is in a different place
  68. blue_it just adds a preload script tag, or what does it do?
  69. phalethyeah, the code for adding that preload tag is fine
  70. phaleththe bug is not there
  71. phalethI mean the preload tag creation was already there :)
  72. blue_oh, so what is the bug
  73. phaleththe bug is that woff2 files are ignored by that magical thing dynamically processing assets
  74. phalethI also think that the build system should take care of pre-rendering the tags, but that'd mean lot of rework
  75. blue_the script tags?
  76. phaleththink header at least
  77. blue_it does create script tags
  78. phalethat build time?
  79. blue_oh, you want at build to create html files with the script tags already baked in?
  80. phalethyeah, but don't worry about that now
  81. phalethit's fine as it, cause it just works
  82. blue_lol, I just did `npx primate build`, and build/server.js isn't minified
  83. blue_when did that get lost
  84. phalethheh