Chat Logs

  1. bluelioloc: hey!
  2. bluehttps://repopack.com/project/primate/tasks/153
  3. nevetRepopack · Sign in
  4. bluecheck out that, will be part of 0.43
  5. bluedreamreal: primate 0.42.1 is out, including updates to @primate/core and @primate/marko. It is a security-hardening and stability release, I recommend updating
  6. dreamrealgah, sorry, underway
  7. bluewell, you don't need to update *now*
  8. blueit's just a general info
  9. dreamrealI'm bad about enduring old versions when I know a fix is out there
  10. dreamreal... another reason to vastly dislike windows
  11. blueyou know, there's a tension between having a lockfile, and receiving updates. I strive to try to diffuse this tension in some form with repopack, with trusted publishing (npm is also going in that direction), but the fact of the matter is that lockfile prevent you from receiving updates simply by redeploying a service
  12. bluelockiles*
  13. bluewhich somewhat defeats the purpose of patch updates, as per npm policy
  14. bluenpm is generally very problematic from a security point of view, but their solutions never feel like they've thought them through. they are most reactive, much less proactive
  15. blues/but/and/
  16. dreamrealTrying to update the nextjs repo now, which is always a drag. The dependency audit for nextjs is great fun: "this is a mandatory update but BTW nothing will work if you apply it"
  17. dreamrealgee, I wonder why I agonized over switching to something different for all of 15 ninutes.
  18. bluenextjs is just not worth it
  19. blueI sent you the video by lioloc where he compares them
  20. bluenextjs is very, very heavy
  21. bluelioloc: so cool, your branch can start a desktop app with gtk 4 on linux. but a revision was still requested
  22. liolocwill take a look
  23. bluelioloc: btw https://primate.run/blog/primate-042#production-diagnostics
  24. nevetPrimate — Primate 0.42: Virtual route groups, structured forms and production diagnostics
  25. bluethis was missing from the original blog post
  26. blueit's kinda nice :)
  27. liolocneat
  28. liolocvery cool!
  29. bluewe're gonna use it on repopack.com, that when an error 5xx is generated, we report it automatically
  30. bluewe might show the user an id for reference
  31. liolocwhat a great idea
  32. lioloci'll have to try it myself
  33. blueyup
  34. bluedreamreal: whilst reviewing primate:
  35. blue"Error: cyber_policy: This content was flagged for possible cybersecurity risk. If this seems wrong, try rephrasing your request. If you’re doing authorized security work that requires more cyber permissive safeguards, apply for Daybreak access via https://platform.openai.com/settings/organization/status-and-access before retrying."
  36. bluegreat, now I can't even review my own OSS code
  37. blueidiots
  38. dreamrealblame aggressive demands by the US government, and idiots trying to use AI to break others' code maliciously
  39. dreamrealSorry, man
  40. dreamrealhits me too, and that's actually MY INDUSTRY - I have access to models that do not have those safeguards but I literally have to be in a military base to use them
  41. dreamrealand thus I do not use them
  42. blueand the annoying thing is that they now flagged the entire session. which is OK, I can start a new one and rp is pretty good at bootstrapping, but yeah
  43. dreamrealoh, that reminds me: do you want #primate's logs to be public as well?
  44. blueyes, it's ok
  45. bluethanks for asking
  46. dreamrealI actually need to sanitize the logs some
  47. bluebut I am bound by these network's policies to advertise that, btw
  48. blue(as are you, if you have it on any of your channels, btw; be advised)
  49. bluebest is to put it in the topic, which I will do now
  50. dreamrealhmm, hat might be an interesting feature: "here's the logs, filter for potentially insecure/illegal content and sanitize profanity"
  51. bluedreamreal: https://bytecode.news/logs?provenance=irc%3A%2F%2Flibera%2F%2523primate
  52. nevet#primate, October 7, 2026
  53. blueis this the shortest url I can get?
  54. blueit is rather ugly, I must confess
  55. dreamrealYES, darn it!
  56. blueok
  57. dreamrealit is ugly, yes, because provenances are ugly
  58. dreamrealhmm
  59. bluenicer: bytecode.news/logs/irc/libera/primate
  60. blueyou can map this with a [...].ts route in primate
  61. blueand yes, I know there are # and ## channels, and that is an abomination to map to a url, where # means the hash/anchor
  62. dreamrealyeah, primate's not the only option here, and I'm going to file an issue against all three UIs, and damn it nextjs sucks and I'm considering deprecating it for maintenance
  63. dreamrealthere are actually FIVE UIs, but two of them are BADLY outdated because I don't give a single crap about them and they're not even deployed live
  64. bluethis is *rather* easy to in primate. create a route as thus described above. then: const `[protocol, server, channel] = request.path.toJSON();`
  65. bluesorry, the opening backtick should be been before the const
  66. blueshould have been*
  67. bluedreamreal: https://libera.chat/policies/#public-logging
  68. nevetNetwork policies
  69. bluejust fyi; I will make sure this is properly advertised in the topics of #primate and #repopack; you take care of your flock
  70. bluejust so we're transparent about it, the public logging by nevet for this channel is authorised by me (channel owner) until indicated otherwise
  71. dreamrealI'm working on the issue, have a few other things in the queue
  72. blueno worries
  73. blueI'll change the topic as soon as you're up with the prettified url
  74. blue... or with the non-prettified one if you decide elsewise
  75. dreamrealI'm working on it, issue has been filed and shouldn't be long
  76. blue+1
  77. dreamrealdang it you casually created six issues for streampack :D
  78. blueyes well, you're carrying too many frontends :P
  79. dreamrealOnly two issues are in the front ends, nextjs gets ignored for this one
  80. bluewhat are the other four?
  81. dreamrealbut this is WHY I wanted multiple front ends, to get the backend API solid
  82. dreamrealnextjs [deployed, deprecated], primate [deployed and is the 'main site'], spring [pudl, deployed and is sort of the 'admin site'], spring (basic) [undeployed, unmaintained, was the proof for the design as a balance against...], reference (nextjs) [undeployed, was the 'site' before the nextjs 'main site' was built]
  83. bluebtw dreamreal, I feel like both things can be true at the same time. repopack exposes an API (for `rp`), *and* it has a very fast direct path for the ui. there is an inherent sync problem that arises from this, but the dividends for the direct path, eliminating the costliness of API requests, and enabling a direct db access essentially, is huge
  84. dreamreal*nod*
  85. blueI'm not saying you *need* to go there, it's your decision. but a two-path'd approach can pay out
  86. blueyour design*
  87. blueof course, the fact the backend is written in kotlin makes it very hard for you to port it to primate. hence I was thinking about a @primate/kotlin wasm bridge
  88. dreamrealI'm not sure what you mean: streampack DOES have a direct API. You mean such that UIs have access to the data directly?
  89. blueessentially yes. on repopack.com, the first load is always *one* request; on bytecode.news too, by masked by another fetch request in the background, to the api
  90. bluebut masked*
  91. dreamrealAh, no
  92. dreamrealI totally understand and it WOULD make everything faster
  93. dreamrealNO ARGUMENT THERE at all
  94. bluebut it would break your API design, I get it
  95. dreamrealbut the intent was NEVER actually for me to maintain UIs that didn't directly interest me
  96. blueya
  97. dreamreallike, I can see me being directly invested in the pudl UI, but honestly, I'd far rather SOMEONE ELSE maintain the primate API because it's not my strength area
  98. dreamrealand that means trusting users to not be damn fools
  99. dreamreallike, there ARE secrets in the database: they're encoded and you could capture the database without having secrets exposed, BUT someone could THEORETICALLY spend a lot of computing power on those secrets
  100. bluethe question here is a matter of coupling. you have a very loose coupling at the moment, but it comes at some price. I am thinking if you can keep it decoupled to your heart's content, and still get the benefits of traditional stronger coupling
  101. dreamrealblue: right, but that comes with TRUST. I could do that with pudl, as the interested maintainer. But not with the other UIs.
  102. dreamrealRight now interactions are pretty fast, and having a common API gives me not only viability tests for the API but common reference points
  103. bluehow so? trust that the server implementation isn't compromised, you mean?
  104. dreamrealTo be clear: you're wrong exactly 0%
  105. dreamrealno, trust in the maintainers of the UIs
  106. bluebut *you* are the maintainer
  107. dreamrealright now there is no trust, and there's no need for it: the UIs can't even pretend to have write access because there is no access
  108. dreamrealright
  109. dreamreal... of pudl
  110. dreamrealand technically primate as well
  111. dreamrealbut I really, really want BCN to have enough critical mass to break that pattern
  112. blueanyway, there are other challenges as well. I'm not sure how much kotlin wasm is ahead, and there are also to my knowledge some limitations with wasm java
  113. bluebut I must confess I have been eager to try out java wasm since java 25
  114. dreamrealwait until the next major release comes out
  115. dreamrealthe java LTS releases are quite something, compared to the history
  116. bluebut 25 is LT
  117. blueLTS, no?
  118. bluesupported until 2033
  119. bluethe next lts seems to be 29, sept 2027, that's some ways off
  120. dreamrealyes, but the next LTS is gonna *rock*
  121. dreamrealI'm staying on 25 until the next LTS but they keep adding features the community's been wanting and needing since, like, 2006
  122. bluethere's also the question of whether it makes sense to do a @primate/java or a @primate/kotlin first
  123. * dreamreal has no recommendation there, not enough experience with WASM in the field
  124. blueya
  125. dreamrealblue: backend is being merged for release now
  126. bluedreamreal: nice!
  127. dreamrealand now it's being set for deployment
  128. dreamrealprimate update following, but it needs backend deployment because the API is changing
  129. bluewith the latest primate updates?
  130. blueah ok
  131. dreamrealand yes, it's going to have the latest primate updates
  132. dreamrealalthough I'm not migrating the *code* very much
  133. bluethere's no code changes in 0.42.1 to be done
  134. blueshould be 100% transparent
  135. dreamreal*nod* just being clear, because I'm not using the route groups, etc
  136. blueI know
  137. bluethe route groups have got some stability fixes
  138. bluebut they're still opt-in, and will remain so
  139. bluedreamreal: https://repopack.com/project/primate/tasks/153
  140. nevetRepopack · Sign in
  141. dreamreal+1
  142. dreamrealGrrrr, trying to get some tuning issues resolved, HOLD ON. One step at a time. The backend deployment needs to work before we progress.
  143. bluethere's an asset chunking bug, though it's not surfaced on bytecode.news
  144. dreamreal*nod* Sorry for the join spam, the AI model interaction changed :/
  145. dreamrealit's rejecting "thinking" because I'm not turning it off properly yet
  146. dreamrealgolly, it's WORKING now but still not right
  147. dreamrealgrrrr
  148. dreamrealOkay, the new AI stuff works, and the openapi endpoints have been cleaned up. Now the primate and pudl updates for the logger entrypoint.
  149. phalethI thought pudl is just a design lang
  150. dreamrealit's an implementation as well, and in THIS context, it's a UI that uses pudl
  151. dreamrealit's also the most pessimistic consumer of the openapi spec, which is why it's a gate
  152. phalethok, I never understood the openapi spec, all I know is that primate has https://openapi.primate.run/manifest.json
  153. phaleththis UI that uses pudl uses pudl at build time I hope
  154. phalethright?
  155. dreamrealerr... no, no
  156. dreamrealthe bytecode API that uses openapi is talking to api.bytecode.news
  157. phalethoh, there is a SwaggerUI, and I was just like curious if you have SwaggerUI and you really do
  158. dreamrealI do, yes
  159. phalethwell, anyway, openapi is like I read odata and microsoft and I'm like "nah, I've had enough of that"
  160. dreamrealit's a drag, yes, but it's probably the best drag available
  161. phalethI also think that odata is one of the good things that microsoft has done, but still
  162. phalethbunx is used to run proby in the primate monorepo, that means it's still vendor locked in
  163. dreamrealblue: zeroing in, found a gap in pudl's endpoints
  164. dreamrealbackend is stable, so closing the gap and then fixing primate
  165. dreamrealAND THEN https://bytecode.news/logs/irc/libera/primate should work
  166. bluevery very good
  167. bluewe have a @primare/core patch coming up
  168. dreamrealwhen and for which? The migration thing?
  169. blueno, a 0.40 thing
  170. dreamrealanything that would affect BCN's deployment?
  171. blueno
  172. bluejust optimising network request size
  173. bluetrimming fat
  174. dreamrealok
  175. dreamrealblue: https://bytecode.news/logs/irc/libera/primate
  176. nevet#primate, October 7, 2026
  177. blueawesome
  178. dreamrealand thank you for the suggestion
  179. phalethdreamreal: the pudl frontend also feels kinda lightweight, which is nice
  180. dreamrealboth it and primate are pretty zippy
  181. phalethyup
  182. dreamrealpudl just uses an application paradigm, which isn't what most users would want or need
  183. dreamrealand then contrast it to nextjs.bytecode.news
  184. dreamrealwhich isn't sluggish but is *by comparison*
  185. phalethyeah, these windows are kinda odd on a website
  186. bluedreamreal: you are welcome
  187. dreamrealphaleth: for *writing* it's fantastic, I use the pudl endpoint for all admin access, AND it gives you a console :D
  188. phalethblue: optimizing network request size?
  189. blueyes, removing the names of chunks from json transfer objects
  190. phalethdreamreal: cool, I like CLIs
  191. phalethblue: nice, every byte counts
  192. dreamrealphaleth: see discord, #offtopic
  193. phalethdiscord? I'll pass
  194. dreamrealscreenshots, tho!
  195. phalethheh
  196. dreamrealI could paste one here but, um
  197. dreamrealsrsly, check it! Just this once!
  198. phalethmaybe later
  199. dreamrealit's the impressive! ... dang it
  200. dreamrealblue: do you see it? Are you not entertained?!?!
  201. bluesee what
  202. dreamrealI guess I could use an image pastebin, but I don't know of any
  203. dreamrealblue: discord, primate, #offtopic
  204. blueah, I saw now, yes
  205. dreamrealWhen I write on the pudl api, NOT ONLY does it have a "suggest tags" and "update tags" feature in the editor, but I can also just crank up a console and type into it just like I would here
  206. dreamrealcalc 42*pi
  207. nevetThe result of 42*pi is: 131.94689145077132
  208. dreamrealopenapi
  209. nevetopenapi is a vendor-neutral, machine-readable spec for HTTP APIs (formerly Swagger): describe the paths and schemas once, generate clients, servers and docs from it. URLs: https://www.openapis.org and https://spec.openapis.org/oas/latest.html
  210. dreamreal(this output is in the screenshot phaleth so CALLOUSLY refuses to see)
  211. dreamrealcalendar hebrew
  212. dreamrealcaloh
  213. dreamrealtoday hebrew
  214. nevetToday is Wednesday, 26 Tishrei, 5787 (Hebrew)
  215. dreamrealtoday
  216. nevetToday is Wednesday, October 7, 2026
  217. dreamrealI'm going to eventually move the factoid suggestion stuff into primate too but it's admin-only, as most of the AI stuff is gated to admins
  218. dreamreal(one aspect of the factoid derivation is heuristic, not AI, but the most useful features leverage AI)
  219. dreamrealI am, BTW, adding a filter for the backend such that it has patterns in URLs to ignore, so I can add "repopack.com/project" and have it not echo those url titles
  220. bluethat's AWESOME
  221. blue(however, there is the intention to make some semblance of public projects on repopack.com; although as phaleth noted, we will be limiting the ability of bots to get the entire js bundle from that, and that's a task for primate to solve)
  222. dreamrealWell, the thing is, the intent is to say "ignore urls that match THIS PATTERN" - not just a string, but a pattern - and also ignore urls that have text like "log in" or "sign in"
  223. blueya
  224. bluephaleth: I've deleted most of my stuff on github, and it'll be moving to repopack soon
  225. bluein particular scld and others I'm going to start creating projects for
  226. blueprimate is going to stay for a while, rcompat will move over completely soon (I already closed most issues). rcompat will be archived and eventually deleted
  227. blueI'm gonna commit fully, I don't see a way forward if I don't
  228. phalethok, it's up to you, let me know when I should get rid of gitea
  229. dreamreal*nod*
  230. bluesoon
  231. phalethalso make sure the UUIDv7s are in the db before anything
  232. bluewe're not going to GA any time close, uuidv7s will be in before that
  233. bluebut they might need to land in primate first
  234. blueI mean, they already are in primate. I'm just not sure the translation in @primate/postgresql is
  235. phalethyeah, makes sense
  236. dreamrealblue: check discord, it is to amuse
  237. dreamrealI'm still giggling
  238. dreamreallike, "holy cow, are you ... what... what's wrong with you"
  239. dreamrealI know giggling isn't manly but... come on!
  240. blueyes
  241. dreamrealand that fool knows the history and visibility
  242. dreamrealand apparently hasn't actually thought "gee, when someone says something they MAY ACTUALLY mean it"
  243. bluetrue
  244. dreamrealIt's back! and I'm adding log moderation capabilities to the UIs now
  245. blueAwesome sauce.
  246. blueYou're on FIRE on nevet today, dreamreal!
  247. dreamrealblue: can you join #nevet so I can intentionally abuse you and trigger the reports?
  248. dreamrealjreicher: found the problem for openliberty BTW
  249. dreamrealjreicher: https://github.com/OpenLiberty/openliberty.io/issues/4367
  250. nevetAtom feed republishes the ArgoCD drift post with a new id on every site build · Issue #4367 · OpenLiberty/openliberty.io
  251. jreicherAnd there was indeed a problem?
  252. jreicherAhah.
  253. dreamrealjreicher: there was, but it wasn't nevet!