Chat Logs

  1. phalethhttps://primate.run/blog/typescript-monorepos-with-proby
  2. bluephaleth: yes!
  3. blueproby is not outclassing vitest / jest
  4. blueis now*
  5. phalethit's cool that node supports ts
  6. blueit's been so for a while with a flag, but they deflagged in 24 and 23 (backported)
  7. bluephaleth: if you don't mind, could you post the rcompat blog post to HN?
  8. phalethsome nice ideas about server side TS are also here https://github.com/elixir-volt/quickbeam/blob/master/README.md
  9. phalethI can try
  10. bluecool
  11. blueooh, zig
  12. bluephaleth: https://github.com/elixir-volt/quickbeam/tree/master/priv/c_src
  13. blueit uses quickjs :P
  14. blueI should known by the name...
  15. phalethyeah, take a look at the bottom of that readme how it handles npm packages
  16. blueshould have*
  17. blueoh cool
  18. bluemaybe we should add erlang/elixir support to primate's backend
  19. blueit would be somewhat the same as that
  20. phalethupvote https://news.ycombinator.com/item?id=47636948
  21. nevetTesting TS monorepos without a build step with rcompat and proby | Hacker News
  22. blueCool!
  23. phalethyeah, well, the thing about that experiment is that there are two runtimes needed to run the app
  24. phalethQJS might be small
  25. blueyup
  26. blueanyway, the genius thing about this new proby nonsense is that it reuses the customConditions in your tsconfig.json
  27. blueI'd been using those to do jump-to-live-source forever, otherwise you jump to the compiled js
  28. bluephaleth: the redeployment loop on primate website is insane
  29. bluearound 3s
  30. bluegit pull && npm run build && systemctl restart primate-web
  31. blueI have never seen anything like this in the wild
  32. blueobviously in real conditions you'd do ab it more, like pnpm install and potentially pnpm build for the dependents, but with a caching enabled, that'd be still 3s unless any package changed / dep updated
  33. phalethyeah, but that deployment is so hacky
  34. blueit's really straightforward. get the git repo, install deps, build necessary packages, build app, run
  35. blueif you got most of those on a hot path, a redeployment boils down to the build app step which is 2-3s
  36. blueso if I only change website content, redeployment is lightning fast
  37. blueif I change packages that apps/website depends on, it might be a bit longer
  38. bluebtw if we use bun to run, we could skip the package build step entirely
  39. bluebun can run ts from node_modules. it's only node that's retarded
  40. phalethyeah, but bun is leaking memory
  41. blueso yeah friggin' sucks
  42. bluebun should really stabilise
  43. blueinstead of putting more effort into nonsense optimisations
  44. phalethheh, building standalone bundle is fine and moving it to server where there is only the JS runtime is also a better idea
  45. blueoh yeah let's optimise Response(null) to be a million times faster
  46. blueyeah, and primate already supports that
  47. blueif the node devs weren't such idiots and agreed to run ts in node_modules, all of this wouldn't be even needed
  48. blueyou'd publish ts to npm, done
  49. blueI like it how the argumentation was "but you then don't know which ts version". that's not an argument, node strips types
  50. phalethbun's only goal was to win the benchmark that the guy on youtube keeps doing, and bun won that benchmark, usability is not what bun is concerned about
  51. bluethat friggin' sucks
  52. phalethv8 should support TS, makes no sense to strip types
  53. blueso I was thinking that the JS parser I'm writing for flog, it should just support TS natively
  54. blueso yeah, what you're saying for v8 essentially
  55. bluewhat we really need though is a new language, like valuescript
  56. bluethat if I just function foo(a: string): string, that nonsense is actually guaranteed in runtime instead of being stripped away
  57. blueif I just do*
  58. bluewhich is kinda funny, because javascript already maintains this information. it has a type system otherwise typeof wouldn't work in runtime
  59. bluephaleth: looking good for primate 0.37 today
  60. blueeaster primate :)
  61. bluenot quite the bunny you'd expect, but close enough
  62. phalethok, hopefully that release contains something usable for repopack
  63. blueyes, the db migrations, the appfacade#env
  64. bluethe revised modules are also a boon
  65. bluein short, all ofit
  66. phalethcool
  67. bluewelp, release is ready
  68. blueI'm gonna head out to the store and cut it when I'm back
  69. chovyhow many stars on gh are you guys up to now?
  70. chovyi would love for this to be successful and topple everyone else
  71. chovybut i didn't see that happening.
  72. dreamrealblue: just published a *different* thing on BCN - I'm holding the primate post for Monday. Today's post is about something that's a bit crap IMO, and I was using it to test something in the rendering.
  73. dreamrealprimate's gonna get front billing monday morning.
  74. dreamrealunless something unusual happens, I suppose, but that's the PLAN.
  75. phalethchovy: claude code CLI published it's source maps recently and so nowadays people are busy starring all those reconstructed github repos called claw code or whatever and that's where all all the github stars go
  76. phalethmyself I do not understand why would a CLI need source maps in the first place, but I guess it's always good when code gets leaked
  77. dreamrealI don't think so.
  78. phalethwell at least everyone can see how terrible that cli is
  79. chovydreamreal: "Front Billing" wdym?
  80. chovyphaleth: is their code bad??? lol
  81. chovyfucking enterprise closed source garbage
  82. phalethyeah, most likely
  83. dreamrealWhat do you mean? I'm not sure what that response is saying.
  84. dreamrealI mean, I respect work product: if you worked on something and *sent me your code*, awesome. If you worked on something and asked me to use it and it worked well... and you didn't send me the code? I'm not going to go grasping for it. It's YOURS.
  85. dreamrealClaude's code is THEIRS. How is this any different?
  86. chovynot sure what you're getting at dreamreal
  87. chovyi have been working in silicon valley for 30 years as an engineer and ALL corporate codebases are a fucking disaster.
  88. chovyat least with foss you know its clean
  89. dreamrealchovy: it's their code. What does its state matter? I've been in the industry for a minute, too.
  90. dreamrealReally?
  91. dreamrealI mean, I've seen a lot of FOSS...
  92. chovyat chegg we had a User.js file that was 100k lines of code no modules
  93. chovysorry no that was fuckign IBM lol
  94. chovywe rewrite that shit after 1 year
  95. chovyired
  96. dreamrealI mean... so?
  97. chovywe rewrite that shit after 1 year of being acquired
  98. dreamrealThis is a matter of ownership and ethics.
  99. chovywho is goinna read a 100k file?
  100. dreamrealDunno! Does it matter?
  101. dreamrealAn AI probably would, and man, that'd be fun
  102. dreamrealIf something is MINE, what right do you have to it unless I give it to you?
  103. chovyi hear what you're saying but in reality most code is just fucking spaghetti. not saying i'm any beteter but i try not to go over 100 lines in a file
  104. dreamrealI mean... who cares?
  105. dreamrealThat's not bad advice (to keep files short) but ... so what? What does that have to do with claude?
  106. bluechovy: 280ish, I believe
  107. bluechovy: in 2026, we're *still* the only framework that supports js as a first-class citizen
  108. bluedreamreal: primate 0.37 is coming out today. if you want to publish a story, THAT is way more important than the rcompat thingy
  109. bluealso it's Easter Primate
  110. dreamrealreleasing on SATURDAY? fsck, that IS something you need to write
  111. chovydreamreal: nothing. now. ai is far better than the people i ever worked with
  112. bluedreamreal: not ANY Saturday. EASTER Saturday!
  113. dreamreal... I feel like we're having different conversations, chovy
  114. dreamrealblue: If you're trying to minimize impressions, that's a good way to do it :D even my programming-obsessed friends are taking today off, it's morons like me who're still paying attention
  115. dreamrealeaster saturday isn't relevant for me, but shabbat is and even so, here I am :D
  116. bluedreamreal: I don't care about that, and you know that
  117. blueI'm not a marketing guy, not in this life, nor will I ever become, and I'm not gonna try, if the release is done I'm gonna release it
  118. dreamrealI do, but it's worth making the decision deliberately. I'd *love* to have primate releases announced on BCN, for what it's worth
  119. dreamrealif it's today, today it is.
  120. bluethere are people waiting for features, I'm not gonna block them for marketing purposes
  121. dreamrealSo what does 0.37 add? the rcompat stuff I already queued for monday, I guess, but ...
  122. bluedreamreal: https://primate.run/blog/primate-037
  123. nevetPrimate — Primate 0.37: Revised modules, database migrations, and typed environment access
  124. dreamrealDo you support oracle or db2?
  125. blueno, since I don't know how to run them myself
  126. blueI'd like to support oracle, but for that I'd need to setup an oracle db locally
  127. dreamreal*nod* there are docker containers - derby is an implementation of db2 in java, not sure how good oracledb support would be. The PK stuff - it sounds like strings ARE NOT valid primary keys any more?
  128. blueyes
  129. bluewell it's still strings. it's just uuid strings
  130. dreamrealSo if I WANTED a pk of "dreamreal" it'd have to be a UUID with a unique index on a name field?
  131. dreamrealFWIW: container-registry.oracle.com/database/free:latest
  132. blueprobably, yes. it just means that key.primary and key.foreign won't allow you to use p.string
  133. bluenot using uuid means a can of worms of "how do I generate new pks" or "user has to do it manually". it could be a use case, but not one I find particularly attractive unless convinced otherwise
  134. dreamrealsure, it's an opinionated thing
  135. skillnews12Primate 0.37: Revised modules, database migrations, and typed environment access - https://primate.run/blog/primate-037
  136. nevetPrimate — Primate 0.37: Revised modules, database migrations, and typed environment access
  137. blueif key.{primary,foreign} were ever to support to p.string again, it'd have to come with a generation strategy
  138. dreamrealOh, THAT's why nevet took forever! It wasn't quoting US, it was quoting skillnews!
  139. blueso, key.primary(p.string, () => { /* return unique new string */ });
  140. bluethat being said, nothing stops you from using p.string without key.primary and setting the index in the db manually
  141. blueyou just don't participate in orm goodness in that world
  142. blueor well, pk-based orm goodness
  143. dreamrealDo you want to see what I drafted before I publish? I have thoughts, and you might not like them, but it's your project.
  144. dreamreal(and there's time: I still don't want to publish this TODAY.)
  145. dreamrealI am going to put it on social media, I think, and I'm not goingto say "two days ago I published..."
  146. bluerelease song: https://www.youtube.com/watch?v=Ci_h4nJRga0
  147. nevetYouTube: The Diary Of Jane | Breaking Benjamin - Topic
  148. skillbot04,01► 14,01YouTube :: The Diary Of Jane
  149. bluedreamreal: ya?
  150. dreamrealhold on, revising
  151. dreamrealI tell you what, I'm gonna be low-key pissed if BCN never takes off :D
  152. bluewhy shouldn't it
  153. dreamrealmarketing, actual appeal, people don't read
  154. blueI have mostly stopped caring about what people think
  155. dreamrealmonetization is also a factor: it's a passion project for now, but passion only carries so much
  156. dreamrealif I'm the only one rowing the boat, it ain't gonna go far
  157. dreamrealI've had a few people say "ooo that'd be great to be published there" and when I say "yeah, write it up" it's crickets
  158. bluewell, people suck
  159. blueI've been the only one rowing the boat on primate for like 2 years and it's only got a handful of stars on gh, I recently stopped caring
  160. dreamrealI hear you (one of those stars is mine, for the record)
  161. dreamrealokay, you on google docs or anything like that?
  162. blueyes, potentially
  163. chovydreamreal: i guess so...idk
  164. chovythought we were just talking
  165. chovynot debating
  166. chovyanyway i trust oss more
  167. dreamrealchovy: we were, I just don't think we were talking about the same things
  168. bluechovy, he just means you're talking past each other
  169. chovythat's the main reason i use it over anything proprietary
  170. bluechovy: it's not the main reason, it's the master reason!
  171. blue:P
  172. chovyi like the foss+cloud model
  173. dreamrealblue: check DM
  174. dreamrealI do too but I also like to eat and have a roof over my head
  175. dreamrealIt's a decision I make deliberately to use open source as a distribution model, or CC-by-SA
  176. dreamrealbut the thing is: if I lost my day job, BCN goes away.
  177. dreamrealBecause if I were to go to a "hey, I gotta eat" model for it, nobody would pay for it REGARDLESS of value: they'd say "aw man, I sure wish someone would pay the author to create such great stuff" without ever once thinking "hey, why didn't *I* pay the author to create such great stuff"
  178. dreamrealblue: that article draft is NOT formatted for google docs, obviously
  179. blueya
  180. dreamrealediting that doc isn't going to help me much, but comments sure would since they'd give me a delta to work with
  181. dreamrealchovy: https://bytecode.news/posts/2026/04/layoffs-at-oracle
  182. bluedreamreal: ACK, good write-up
  183. blueI don't really have any comments, it's pretty much the gist of the release, the evaluation is also par for the course
  184. dreamrealOkay, cool. It's going to get pride of place on monday, I think.
  185. blueEaster Monday? :P
  186. dreamrealIs that a thing?
  187. dreamrealBut yes
  188. dreamrealpeople here work on monday after easter, it's not that kind of holiday
  189. dreamreallater
  190. blueso in germany, easter friday to monday are typically free
  191. dreamrealshould i wait until tue?
  192. blueno
  193. blueif easter monday ain't a thing in the us, then publish
  194. blueactually publish it whenver you want
  195. blueI'd publish it now :P
  196. bot[Primate Blog] Primate 0.37: Revised modules, database migrations, and typed environment access - https://primate.run/blog/primate-037
  197. dreamrealhttps://bytecode.news/posts/2026/04/primate-0-37-migrations-typed-env-and-a-cleaner-module-api
  198. bluedreamreal: awesome!
  199. dreamrealgave me a chance to flesh out a lot of factoids, too
  200. bluehttps://x.com/primate_run/status/2040447712659722314
  201. bluehttps://tboteproject.com/systemdfindings/
  202. nevetsystemd birthDate Merge: Corporate Filings, Conflicts of Interest, and Governance Failure
  203. bluethis is pretty crazy
  204. onosendiblue: have you tried liquid dnb? https://www.youtube.com/watch?v=s17e_5zK21s
  205. nevetYouTube: #304 KushSessions (Liquid Drum & Bass Mix) | KushSessions
  206. skillbot04,01► 14,01YouTube :: #304 KushSessions (Liquid Drum & Bass Mix)
  207. blueonosendi: I haven't!
  208. bluetrying now
  209. bluenot bad
  210. chovyso this is why i switched from primate to next/react i can build any app with any interface on any platform using that stack. i don't think you can do web extensions with primate
  211. chovyor mobile apps
  212. chovyor desktops or clis
  213. bluechovy: what's web extensions?
  214. blueand of course you can do that with primate. next isn't doing anything in particular that makes it mobile or deskopt ready
  215. blueyou're probably using sometihng else like react native or so
  216. chovyyou don't uswe next on mobile or web extensions
  217. chovyjust react
  218. chovyi use electroin for desktop
  219. chovyyes react native
  220. chovyfor mobile
  221. chovyi can build tuis with react now lol
  222. chovythat shit is everywhere!!
  223. chovyclaude won't let you work more than 12 hours at time before they temp ban you
  224. chovyhow you gonna become a billionaire working 12 hours a day?
  225. blueI thought you hated next chovy
  226. bluewhy are people using that nonsense, it's so terrible
  227. bluelike even if you don't use primate, there are *better* alternatives