Chat Logs

  1. bluehi phaleth
  2. phalethhi blue
  3. phalethI've been lurking and found https://github.com/railwayapp/railpack which looks like something we might need
  4. nevetGitHub - railwayapp/railpack: Zero-config application builder that automatically analyzes and turns your code into an image
  5. phalethlooks like it builds debian based images
  6. phaleththe idea itself is good I think
  7. phalethran into this bug https://github.com/containers/podman/pull/28454 well, actually that's the bugfix, just not released yet
  8. nevetfix(#28429): OCIRuntime field in SpecGenerator is ignored when called through HTTP Rest API by aadi58002 路 Pull Request #28454 路 containers/podman
  9. phalethit's good that someone reported the problem
  10. bluewhat's it do?
  11. blueoh btw yesterday I discovered podman-tui, tried to install via aur, it went straight to go compilation so I ditched it
  12. phalethlets you choose the container runtime, via CLI that works, it's a bugfix for REST API
  13. bluearch is so bad sometimes, I'm way too old to be compiling software
  14. jreicherblue: k9s?
  15. phalethyeah, well, I'm trying to avoid CLIs for the automation
  16. bluejreicher: I won't touch kubernetes with a 9-yard-stick
  17. bluephaleth: and yes, the principle behind railpack is essential the same like repopack
  18. blueexcept of course, using a web app and not spitting out a docker file
  19. bluethis is the deal, right? I *still* need to deploy the thing somewhere
  20. phalethrailpack does not output Dockerfile
  21. phalethit uses buildkit LLB, which I think is something worth taking notes about
  22. bluebut we will also have a `rp` cli that does essentially the same, but deploys to rp.com
  23. bluemy idea is something like `rp deploy`, then you get `deployed to https://[hash].repopack.app`
  24. bluebut we will need something like `rp login` or so because obviously you need some user association there
  25. bluebut essentially my dream flow is running just one command in my repo and getting a url that I can open
  26. blueand then rince repeat
  27. phalethoh, you mean CLI for the enduser, that's for sure gonna be very handy as people want to script things
  28. blueya
  29. bluebtw, phaleth, I was looking at Temporal last night, and oh boy, that is a dumpster fire of complexity
  30. blueso node 26 has it, deno apparently has it too, bun doesn't have it because webkit doesn't have it
  31. bluehttps://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Temporal
  32. nevetTemporal - JavaScript | MDN
  33. bluelook at this bible-long garbage
  34. blueit took them 9 years to come up with THIS?
  35. bluethe amount of don'ts is terrible
  36. blue"Don't assume that era is a user-friendly string; use toLocaleString() to format your date instead."
  37. blueyeah, good luck giving devs a bunch of don'ts
  38. phalethyeah, well, dealing with datetimes and timezones has always been a pain
  39. blueI think I will skip that and create @rcompat/time. dates are complex, but this API is *not* a good answer
  40. phalethnah, just don't bother
  41. blueI have good ideas, and you know how it is when I have good ideas
  42. blueI waste time on them
  43. phalethjust stitch together something quick and dirty and hardcode it to repopack
  44. phalethif it works out you can create a lib out of that later
  45. blue`import time from "@rcompat/time"; const t = time.ref(Date | string | number); t.stamp();`
  46. blueyeah I don't actually need any of this for RP
  47. phaleththat's good, then no need to bother at all
  48. phalethfor repopack you can always convert time on frontend anyway, will that make the site slower? yes, but who cares, it's a tiny bit slower
  49. bluethat's what you should do, anyway
  50. phalethagreed
  51. bluewe will store TIMESTAMPTZ in the postgres db
  52. phalethnew Date().getTimezoneOffset() / 60
  53. blueand we will get a date object
  54. bluebecause p.date maps to TIMESTAMPTZ
  55. phaleththat should return -2 for you, and timestamps in postgres are all GMT
  56. blueUTC, yes
  57. phalethyeah
  58. phalethnow you need to figure out how to calc time diffs to be able to add those -2 or whatever and that way you replace the whole library
  59. blueya
  60. bluebut you know, I just want simple APIs. the two mortal sins of CS are caching and overengineering
  61. blue`epochNanoseconds`
  62. bluereal use case right there right now. JS in airplanes, coming to your next starport!
  63. blue`Temporal.PlainYearMonth`
  64. bluecome ON
  65. bluethis is so bad
  66. blueit took them nine years to come up with this monstrosity
  67. blueno thank you, I prefer `Date` over that with all its gotchas
  68. phalethyeah, waste of time
  69. blueI think they copied some stuff from https://docs.oracle.com/javase/8/docs/api/java/time/temporal/Temporal.html
  70. nevetTemporal (Java Platform SE 8 )
  71. blueI mean, the Java API is its own level of crazy with stuff like `ThaiBuddhistDate`
  72. blueMinguoDate ???
  73. bluewhy is everything so bad
  74. bluewait till we colonise mars, they'll invent java.time.chromo.MarsTime
  75. bluechrono*
  76. blueanyway, point being, I ain't using Temporal in JS
  77. bluephaleth: given repopack.com/[0]/[1], do you prefer calling 0 org or namespace?
  78. phaletha participant
  79. phalethor a member
  80. blueso conceptually it can be a user or an organisation
  81. bluemember sounds good to me for now, we can also call it user and have user types
  82. phalethforgot to say I need to reboot the server
  83. phalethblue: any clue what's wrong here? https://upaste.de/raw/FDq
  84. bluephaleth: did you restart the server?
  85. phalethtrying to add the --prod flag to pnpm install
  86. phalethblue: yes, I did, forgot to warn
  87. blueall good
  88. blueI'm just monitoring contabo a bit for reliability
  89. bluehm, that's pnpm being a pain in the butt
  90. phaleth--prod seems to have effect https://upaste.de/raw/m79
  91. blueso the problem is (peer)deps inside the monorepo
  92. bluesome imports might point to .js files, but if `npm run build` hasn't been run at least once, those files won't exist
  93. bluepnpm is just a garbage tool, that's all. if you run `npm run build` once inside the repo, the next pnpm install would work
  94. bluemaybe we'll switch to bun from pnpm
  95. bluebtw I love the new comp
  96. blueI only just now finished configuring niri and the gang, and it's fabulous
  97. phalethI've tried switching to bun from pnpm but could not figure out how
  98. phalethnew comp?
  99. blueyes, the thinkpad t480 that I'd order like half a year ago
  100. phalethoh yeah
  101. blueand then I was slowly building up superarch on it
  102. blueit's on gitea
  103. phalethpnpm install just worked two days ago, nothing has change on primate master since
  104. bluewasn't there the switch to tsgo?
  105. blueoh, this is still on github, I need to move it to gitea
  106. bluehttps://github.com/superarch-org/config
  107. phaleththe switch to tsgo wasn't a problem two days ago
  108. phalethand there is no new commit since
  109. bluepnpm sucks
  110. blueis the container running?
  111. phalethyeah, but bun cannot replace pnpm
  112. blueI can pop in and fix it
  113. phaleththe container is fine
  114. blueso buildctl is failing?
  115. phalethyeah, image build is failing
  116. blueRUN pnpm approve-builds
  117. bluetry this phaleth
  118. blueprior to
  119. blueRUN pnpm install --prod
  120. phalethok
  121. blueor just drpo the --prod flag
  122. bluewhat do you need it for, anyway
  123. bluewhat's it do?
  124. blueRUN pnpm install
  125. blueRUN pnpm -r build
  126. blueor that other command which builds only the packages relevant for apps/website
  127. phalethhttps://upaste.de/raw/AAo
  128. phaleth--prod means it will not bother with devDependencies on pnpm install
  129. phalethbut it was already failing without --prod
  130. bluewas that the first link you sent?
  131. phalethyeah
  132. blueI get the same warnings when I clone locally
  133. phalethI think there is a new pnpm version
  134. blueanyway the prod flag isn't gonna help you
  135. blueyou need TS to build stuff
  136. blueand TS is a devdep
  137. blueif you were an app, it would be OK, because you'd be consuming js from npm or having your bundler transform non-js
  138. bluebut here you need the TS devdep
  139. bluejust put
  140. bluepnpm install && pnpm -r build
  141. blueinto the build
  142. bluethat should work
  143. bluewith tsgo it's gonna be fast anywy
  144. phalethright, that's true
  145. phalethRUN npm install -g pnpm@10 <- avoids the problem
  146. blueyeah, let's pin things to an old version that no one will support anymore
  147. phalethwelcome to the JS ecosystem
  148. bluehaha
  149. bluealso known as the DS ecosystem
  150. bluedumpster fire script
  151. phaleththat's DFS
  152. blueyes but that's already like depth-first search or some other nonsense
  153. phalethdepth first traversal
  154. bluejavascript is a joke, and we all know it
  155. phalethI mean pnpm 10 works better than bun
  156. blueit's always been a joke to be honest, but the sad thing is that no one's trying to fix it
  157. phalethyou can try to build the monorepo with bun, but then you will have to change things in the repo I guess
  158. blueso the reason I haven't done that is simply that most people currently run node
  159. phalethyeah, bun just comes with their own solution that comes with it's own problems
  160. blueand I don't *feel* the need to verify that stuff is working on deno/bun, because I have rcompat
  161. blueyou have no idea how comfortable it is just to write stuff and know it works on all runtimes
  162. blueusing non-crazy APIs
  163. phalethI wish npm was able to build these monorepos
  164. phaleththe way pnpm screws around in the node_modules dir is very odd
  165. blueyeah, pnpm is doing so odd voodoo magic
  166. bluesome*
  167. bluethey all do, because the JS ecosystem is like that