ByteCode.NewsSubmit
Use quotes for an exact word.

Subscribe

RSSPodcast
Log in

Chat Logs

  1. phalethonly 20 seconds to build primate website, used to be about 45-50 seconds
  2. phalethflog really needs trust me bro script and windows builds
  3. bluehi phaleth
  4. blueyes, the primate website now doesn't need to build packages anymore, runs TS from source
  5. bluebtw phaleth, this is something that flog/dog are gonna solve generally for TS. no build step
  6. phalethhi blue
  7. bluehi
  8. phalethyou mean there is no call to flog build?
  9. blueexactly. no packages are being compiled to JS
  10. bluethe 20s are, according to my analysis, 14s docker nonsense, 6s primate build
  11. bluedocker stuff can be cached. the real redeploy time of primate website is 6s
  12. bluethere's also flog install involved, but that is really, really fast. faster than pnpm currently
  13. phalethbut in the Containerfile there is RUN flog run --workspace apps/website build
  14. blueyes. flog run jst runs scripts
  15. bluein that case, the script is `primate build`
  16. phalethah, so before the call was: RUN flog build --workspace apps/website
  17. blueyes. `flog build` builds TS -> JS, recursively
  18. bluewith dog around, `flog build` should be kinda obsolete I hope
  19. bluebtw dog's motto: One paw at a time. I came up with it and clanker approved
  20. phalethnah
  21. bluedreamreal found it hilarious for some reason
  22. phalethI just hope clanker is done with flog except for the issues I've mentioned
  23. blueflog is only getting started. https://repopack.com/project/flog/tasks/73
  24. bluewe're getting `flog format` and `flog lint`. but for that we need dog, so it'll arrive later, probably in 2027
  25. blueflog -> everything related to monorepo maintainance, dog -> just running stuff
  26. bluethe split bun never had, but should have had
  27. bluemaintenance*
  28. phalethbut for just running stuff it can use node
  29. blueyes but node is slow and stupid
  30. phalethor the go backed in runtime
  31. bluewell goja doesn't have jit, dog has jit (already)
  32. phalethnode runs on v8
  33. blueya, dog is gonna be faster than v8
  34. phalethomg :D
  35. phalethI thought flog is gonna be generally useful for people who can't be bothered with pnpm, but cannot fully migrate
  36. blueyes, well it's gonna replace npm/pnpm/yarn/nub/bun
  37. phalethanyway, JS engine is ton of wasted tokens cause of web standards
  38. bluethe web standards are actually the easiest part because clanker has ref
  39. phalethgoja is slow, too bad
  40. bluejit is the hard part, but dog is already getting fast there
  41. blueit already has 0 recompilations for some warm operations
  42. phalethok, so that's what is clanker doing
  43. bluethat doesn't mean a lot right now, but it's only going to get better
  44. blueand btw remember it's currently only optimised for amd64
  45. blueunlike v8, which has instructions for many, many platforms. I just need linux servers for now
  46. phalethis JIT compilation doable with golang?
  47. blueyes
  48. bluea friend of mine built wago (wago.sh) which is nearly the fastest wasi runtime, has jit
  49. bluebeats the v8 wasm runtime
  50. bluethat's currently the inspiration for dog, wago showed it's possible
  51. bluewago is top of many benchmarks and 2nd or 3rd (to jsc) in others
  52. phalethnah, clanker has no chance
  53. bluewago is 100% clankered
  54. phalethyeah, but wasm is simple
  55. blueyes
  56. blueI'm not saying it's gonna be as easy as wago
  57. bluebut it is possible
  58. phalethwhat is this error? flog: binary target is missing
  59. blueno idea
  60. phalethtrying to call: flog run --workspace apps/website build
  61. bluebinary target sounds like esbuild stuff?
  62. phalethah, so package.json and other things got update in primate repo
  63. phalethso I need to pull down that
  64. blueyes
  65. bluephaleth, you'll notice it's really really fast
  66. blueit's actually ridiculously fast
  67. phaleththat @primate/source is some magic
  68. phalethwell, at least pnpm repos can still use the previous: flog build --workspace apps/website
  69. blueit's just runtime keys. with dog it's all gonna be standardised into the "source" runtime key
  70. bluethe idea is, no need to compile TS, ever
  71. blueit's perfect
  72. bluebtw this reminds me, once primate 0.43 releases and flog.run can serve binaries, we gotta move rcompat and repopack to flog
  73. blueget rid of pnpm completely
  74. phalethyeah, also bun
  75. blueyes bun is dead anyway. sumner threw out all the team and is now playing with claude alone
  76. phalethnot sure what you mean by: compile TS
  77. blueTS -> JS
  78. phaleththere is still build/server.js right now
  79. blueyes but that's primate build, don't confuse it
  80. blueprimate build is about building the primate server with esbuild. it also strips TS but that's not a costly operation
  81. phalethso tsc is not called on invididual primate packages?
  82. phaleththe 6 ones that the primate website needs
  83. blueright now? no. it used to, when we had `flog build` in the containerfile
  84. phalethhmm
  85. bluetsc (which is tsgo since version 7) is fast, but still costs a lot of time
  86. bluethe reason we went down from 44s to 20s are mostly factoring out the TS build time
  87. blueand a few optimisations in flog
  88. phalethmagic
  89. bluewell it's hardly any magic. bun has supported running ts directly for a while. node has had it since v24 I think
  90. blueit's not really magic anymore
  91. phalethwell, if rcompat can be built that way then it's legit I guess
  92. bluebtw phaleth, one of hte reasons I'm building flog, is that node doesn't allow loading TS from node_modules
  93. phalethah, so that's why
  94. bluewhich is a very, very backward decision
  95. blueone of the reasons I'm building dog*, sorry
  96. bluebun can load TS from node_modules
  97. bluebut node refuses. it has the capability, it just refuses
  98. phalethyeah, bun is pushing node around, good
  99. phalethwell, what a shame for people paying for all those CI costs
  100. phalethso dog is gonna be the first TS only runtime? JS completely banned?
  101. phalethI mean it should be exactly that
  102. phalethcause there are lot of optimizations to be had from the typesystem
  103. blueidk it needs to ban JS completely
  104. blueidt*
  105. bluebut dog is gonna be focussed on avoiding that pesky build step
  106. blueand also you to publish source-packages to npm
  107. bluein fact, encourage you to do so
  108. blueif you publish both src and lib, and point to src with the "source" runtime key, you allow dog to run your packages directly from source
  109. bluethe problem of newer runtimes, like deno and bun, they always tried to reshape reality around their own egos
  110. blueI think it's important to play along with npm/node even if I don't like it
  111. phalethsounds good, but it'll take a lot of time
  112. bluetime is all we got
  113. blueof course if we had funding, this would be done yesterday
  114. dreamrealSo getting streampack set up for another network has been... instructive
  115. phalethwhat's the tech stack?
  116. dreamrealthe backend server is postgres+jvm, the front ends are variable: everything's OCI. The main front end on bytecode.news is primate; there's a spring+JS UI at pudl.bytecode.news
  117. dreamrealthe UIs are "bytecode.news" that use the streampack API, they're not "streampack"
  118. dreamrealThe idea was to build an ontology on the backend, not using *ontology tools* because nobody really thinks of ontologies very well, and the translation from ontology to usable API isn't worth it for the most part. But the API says "here's what you can do," and the UIs and services make those operations appropriate for their domains.
  119. dreamrealso there's an ontology, and an IRC service, and... well, there's nevet right there
  120. dreamrealI had thought of actually doing an ontology, blazeds or rdf, but... again, ontologies
  121. phalethah, it's that nevet thing that is still deployed on the VPS next to repopack
  122. dreamrealyeah, I don't know the state of that at all, baby steps!
  123. phalethontology makes you sound like a doctor
  124. dreamrealDoes it? To me it makes me sound like someone who works with knowledge graphs, because that's what ontologists do
  125. phalethheh, ok, I never heard that term
  126. dreamrealan ontology = a body of knowledge
  127. dreamrealdesigning one is utterly nontrivial
  128. phalethI remember nevet deployment with primate as frontend was rather smooth, only one problem with jdbc
  129. dreamrealnevet's factoids are an actual limited ontology; a factoid has info, urls, seealso, tags, maven coordinates, etc., as a very limited ontology
  130. dreamrealoh yeah, primate stores the tokens in a db as a limited scope
  131. dreamrealthe nextjs front end doesn't, and probably should, but I'm trying to sort of drop that one as unmaintained except for security issues: if someone wants to write a GOOD nextjs front end I'll be happy to deploy it but it's not MINE and I don't like that UI
  132. bluenextjs and good are not typically words that appear together in one sentence
  133. phalethhmm, I think nextjs or in other words react on backend is wrong, should be all TS at this point in time
  134. dreamrealyep, I suppose it's doable but not by me and I have no interest in it
  135. dreamrealI WOULD like to try an HTMX UI, though, it'd be a lot like the pudl implementation
  136. dreamrealcould probably use a lot of the same server code
  137. bluedreamreal: https://primate.run/docs/frontend/htmx
  138. nevetPrimate — HTMX
  139. dreamrealheh
  140. dreamrealgood point :D
  141. dreamrealSee discord, BTW: I'm working on an article about rcompat
  142. blueprimate is kinda awesome: you could have both marko *and* htmx, in ONE app, in two disjunctive bundles
  143. bluelooking
  144. dreamrealIt feels like rcompat mostly needs a good thorough documentation workover
  145. phalethhtmx tends to go well with any view/template layer, but the framework has to be able to support partial html responses
  146. phalethbut might as well not use any framework and put all markup in TS
  147. phalethI mean template literals are a thing, not much support from IDEs, but clanker does not care
  148. bluedreamreal: you're also welcome to submit patches to rcompat (docs), if you want. otherwise I'll fix those things myself as soon as I can
  149. phalethnew rcompat favicon is very original
  150. bluephaleth: primate uses partial html responses on every htmx request, automatically
  151. phaleththat means dreamreal does not have to pester clanker with TS
  152. phalethI mean template literals
  153. blueya
  154. phalethactually, I gotta check if I pushed the favicon
  155. phalethrepopack site thinks favicon is binary
  156. dreamreal0ver.url=https://0ver.org/
  157. nevetZeroVer: 0-based Versioning — zer0ver
  158. dreamreal0ver
  159. dreamreal0ver.url=https://0ver.org/
  160. dreamrealFactoidUpdatedEvent(selector=0ver)
  161. nevetok, dreamreal: updated 0ver.
  162. dreamreal0ver=<reply>"0ver" refers to a versioning scheme that uses a significant version of "0," forever, used by *many* popular projects across many ecosystems.
  163. dreamrealFactoidUpdatedEvent(selector=0ver)
  164. nevetok, dreamreal: updated 0ver.
  165. dreamreal0ver.seealso=semver
  166. dreamrealFactoidUpdatedEvent(selector=0ver)
  167. nevetok, dreamreal: updated 0ver.
  168. dreamrealsemver.url=https://semver.org/
  169. dreamrealFactoidUpdatedEvent(selector=semver)
  170. nevetok, dreamreal: updated semver.
  171. dreamrealsemver=<reply>Semver is "semantic versioning," a version scheme that uses major, minor, and patch version numbers, indicating the severity of change from *prior* numbers.
  172. dreamrealFactoidUpdatedEvent(selector=semver)
  173. nevetok, dreamreal: updated semver.
  174. dreamrealsemver.seealso=0ver
  175. dreamrealFactoidUpdatedEvent(selector=semver)
  176. nevetok, dreamreal: updated semver.
  177. bluedreamreal: https://rcompat.repopack.app got a logo... courtesy of phaleth
  178. nevetrcompat — JavaScript standard library
  179. dreamrealI HATE IT! Let me go look at it
  180. bluehow can you hate it if you didn't look at it >_>
  181. dreamrealno favicon yet, is it the flowchart-looking thing?
  182. dreamrealI was being funny
  183. dreamrealI am hilarious, you know, everyone tells me that after I pay them to say it
  184. blueyes, no favicon yet, because I was lazy
  185. dreamrealNo worries, one bit at a time
  186. blueweird, there is a favicon
  187. blueit's just not getting loaded
  188. blueit's not explicitly in the code, btw. it relies on browsers trying for it automatically
  189. bluelike primate.run does
  190. bluebut I think the favicon may be botched, will need to ask phaleth
  191. bluedreamreal: https://rcompat.repopack.app/
  192. nevetrcompat — JavaScript standard library
  193. bluefavicon works
  194. blueI do have some other fixes
  195. dreamrealwoot
  196. bluewhat woot
  197. dreamrealfixes + favicon
  198. blueya
  199. bluehttps://bytecode.news/posts/2026/10/emergent-design-outside-the-codebase
  200. bluethis is an incredible article, but I recommend listening to it rather than reading
  201. * dreamreal shudders
  202. dreamrealthe narrator's a goofball and is not good at this
  203. dreamrealThe author also needs to start putting "hey, share this if you think it's worth sharing" somewhere
  204. blueoh come on, don't go yt on us
  205. dreamrealWell, here's the thing: people are not sharing this
  206. dreamrealthey're leaving it up to me and that's not going to be sufficient to build critical mass, it's just the way it is
  207. blueI mean, I *get* the sentiment. people don't share unless you ask them to. I might be an exception, but still, the "subscribe, hit the bell icon" thing is what annoys me the most in yt videos
  208. dreamrealso I can do it all myself OR I can ask people to step up and help
  209. dreamrealI might ask and then nobody thinks it's worth sharing, that's entirely a possible outcome
  210. bluewell this is *good* content, and it's essentially for free, I don't see a reason not to share it
  211. dreamrealbecause sharing is an action and I don't have any calls for action, so people... don't
  212. blueand btw it's good, as we already established, despite the primary sources often being trrible
  213. dreamrealso the exposure is mostly limited to MY networks and I don't enjoy networking well
  214. dreamrealthere's some but not a lot
  215. bluebtw, you can add bytecode.news here as a feed too, I'd say
  216. dreamrealand I can't see myself doing the whole "click 'like' and subscribe!' thing
  217. blueit might not always or often be primate relevant, but it's almost always worth reading
  218. dreamrealrss feed subscribe https://bytecode.news/
  219. dreamrealfeed subscribe https://bytecode.news/
  220. nevetSubscribed to "ByteCode.News"
  221. dreamrealgrr
  222. dreamrealthere
  223. bluenice
  224. bluefeed list
  225. nevetjustine.lol - https://justine.lol/rss.xml Next.js Blog - https://nextjs.org/feed.xml Read the Tea Leaves - https://nolanlawson.com/feed/ The JetBrains Blog - https://blog.jetbrains.com/feed/ W3C - Blog - https://www.w3.org/blog/feed Terminal Trove - https://terminaltrove.com/blog.xml Distant Province - https://distantprovince.substack.com/feed That Privacy Guy! Blog - https://www.thatprivacyguy.com/blog/feed.xml James Shore - https://www.jamesshore.com/v2/feed Entropic Thoughts - https://entropicthoughts.com/feed Feed - https://hollycummins.com/rss.xml Eignex - https://eignex.com/feed.xml Nate Meyvis - https://www.natemeyvis.com/feed/ FREEZE64.COM - https://freeze64.com/feed/ purplesyringa's blog - https://purplesyringa.moe/blog/feed.rss Angular Blog - Medium - https://blog.angular.dev/feed Typotheque - https://www.typotheque.com/rss Ziglang.org News - https://ziglang.org/news/index.xml Martin Fowler - https://martinfowler.com/feed.atom bun.com - https://bun.com/rss.xml Svelte blog - https://svelte.dev/blog/rss.xml The Astro Blog - https://astro.build/rss.xml Improbable Research - https://improbable.com/feed/ Diátaxis - https://diataxis.fr/atom.xml Terence Eden’s Blog - https://shkspr.mobi/blog/feed/atom Ember.js Blog - https://blog.emberjs.com/rss.xml Random Notes and Cheat Sheets - https://www.kodewerk.com/feed.xml Deciphering Glyph - https://blog.glyph.im/feeds/all.atom.xml Docker - https://www.docker.com/feed/ Node.js Blog - https://nodejs.org/en/feed/blog.xml Simon Willison's Weblog - https://simonwillison.net/atom/everything/ Chicory Blog - https://chicory.dev/blog/rss.xml tRPC Blog - https://trpc.io/blog/rss.xml Primate Blog - https://primate.run/blog.rss OpenTelemetry - https://opentelemetry.io/index.xml Blog posts | RSS Feed - https://www.cockroachlabs.com/rss.xml Rust Blog - https://blog.rust-lang.org/feed.xml ByteCode.News - https://bytecode.news/feed.xml Nushell - https://www.nushell.sh/atom.xml Vanish is... - https://vanish.is/feed.xml Paged Out! Magazine - https://pagedout.institute/rss.xml JoshLong.com - https://api.joshlong.com/feed.xml Nature Reviews Computing - https://www.nature.com/nrcomput.rss David M. Lloyd - https://word-bits.flurg.com/rss.xml Biome | Blog - https://biomejs.dev/blog/rss.xml Deno - https://deno.com/feed Spring - https://spring.io/blog.atom OpenAPI Initiative - https://www.openapis.org/feed TypeScript - https://devblogs.microsoft.com/typescript/feed/ Enigmastation.com - https://enigmastation.com/feed.xml Open Liberty - https://openliberty.io/feed.xml MDN Blog - https://developer.mozilla.org/en-US/blog/rss.xml Fernando Borretti - https://borretti.me/feed.xml The Nuxt Blog - https://nuxt.com/blog/rss.xml The Vue Point - https://blog.vuejs.org/feed.rss cekrem.github.io - https://cekrem.github.io/index.xml Vlad Mihalcea - https://vladmihalcea.com/feed/ Piotr's TechBlog - https://piotrminkowski.com/feed/ Apache Fory™ Blog - https://fory.apache.org/blog/rss.xml Vercel News - https://vercel.com/atom Poimandres - https://pmnd.rs/feed.xml Apache NetBeans - https://netbeans.apache.org/blogs/atom React Blog - https://react.dev/feed.xml Christian Rackerseder | Blog - https://www.echooff.dev/rss.xml thoughts from the red planet - http://feeds.feedburner.com/ThoughtsFromTheRedPlanet WebKit - https://webkit.org/feed
  226. blueah, it's all feeds
  227. dreamrealgah, that's a horrible command
  228. bluenot just this channel's
  229. blueI forgot :/
  230. dreamrealyeah, I need to fix that
  231. dreamrealfeed subscribed
  232. dreamreala whole lot of commands I forgot how to use
  233. dreamrealfeed
  234. bluenevet should print help I think
  235. bluefeed
  236. dreamrealyeah, the IRC side has a lot of trouble with help because it's so verbose
  237. dreamrealfeed subscriptions
  238. nevetbun.com - https://bun.com/rss.xml Node.js Blog - https://nodejs.org/en/feed/blog.xml Primate Blog - https://primate.run/blog.rss ByteCode.News - https://bytecode.news/feed.xml Deno - https://deno.com/feed
  239. dreamrealthere
  240. bluenice
  241. blue"The broad, general, forward-looking interface built for unknown future uses is exactly the speculative abstraction YAGNI forbids"
  242. bluevery good
  243. bluewhat's funny about this article dreamreal is that it describes rcompat almost to t
  244. blue"In both cases the design arrives when something uses it, and not before."
  245. blueit was always designed around my needs in primate
  246. dreamrealyes, emergent design isn't new
  247. blueto a t*
  248. dreamrealbain's book was mid-2000s and wasn't new THEN
  249. bluemaybe I should get it
  250. bluebtw: why don't you do affiliate links?
  251. blueI think you link to amazon relatively frequently
  252. blueif I buy the book anyway, I don't see any harm in you getting something out of it. it's a win-win situation, since amazon has enough to spare
  253. dreamrealI am allergic to such things :/
  254. dreamrealDon't want to feel like I'm trying to sell people on things, including thinking
  255. blueyeah I get it
  256. bluebut at some point it would be nice to have a business model around BCN.. maybe subscriptions
  257. dreamrealyeah, I know
  258. dreamrealI just hate the ideas I've had around it
  259. blueya
  260. bluehttps://news.ycombinator.com/item?id=50132798
  261. nevetEmergent Design, Outside the Codebase | Hacker News
  262. blueeveryone pls upvote
  263. bluedreamreal: https://repopack.com/project/flog/tasks/74
  264. bluea way forward for dog/flog
  265. bluehttps://repopack.com/project/rcompat/tasks/46
  266. bluesoon coming
  267. bluedog is screaming along...
  268. bluecurrently working on a measured array-index speed paw
  269. bluedreamreal: ^
  270. bluewe also need `flog test --coverage`