ByteCode.News
Submit RSS Atom Sign in

Back to the channels

#java

  1. * yaboishady2 joined #java
  2. * rorx joined #java
  3. * mindCrime joined #java
  4. * dmlloyd joined #java
  5. * NiKaN joined #java
  6. * ferdna joined #java
  7. * b3t10 joined #java
  8. * fgarcia joined #java
  9. * ChaiTRex joined #java
  10. * domicron joined #java
  11. * stfstfm joined #java
  12. * ChaiTRex joined #java
  13. * domicron joined #java
  14. * michele joined #java
  15. * marcel1 joined #java
  16. jreicher jdtls 1.60 is out, but they only tagged 1.59 3 days ago. I wonder what that's about.
  17. * jreicher joined #java
  18. * MikeBux joined #java
  19. * stfstfm joined #java
  20. * Krusher joined #java
  21. * kathadris joined #java
  22. * iwtga joined #java
  23. * ptomli joined #java
  24. * magla joined #java
  25. * GreenResponse joined #java
  26. dreamreal jreicher: changelogs?
  27. jreicher Good point! I'm embarrassed to say I didn't bother to look.
  28. jreicher https://github.com/eclipse-jdtls/eclipse.jdt.ls/blob/main/CHANGELOG.md
  29. javabot jreicher's title: "eclipse.jdt.ls/CHANGELOG.md at main · eclipse-jdtls/eclipse.jdt.ls · GitHub"
  30. dreamreal And so the mystery is solved. Not sure why those changes were significant so quickly, but eh
  31. dreamreal I probably do the same: "There, it's done! Out into the world you go, product! OFF WITH YOU! ... wait, I forgot to ..."
  32. * julemand101 joined #java
  33. Para Junie hasn't yet managed to start generating changelogs.
  34. Para I'm annoyed because they do multiple releases per week.
  35. * julemand101 joined #java
  36. dreamreal I still haven't really used Junie as far as I know
  37. dreamreal if it's integrated lightly into IDEA, then I have, but I haven't explicitly used it
  38. Para JetBrains kinda also messed up their LLM support so it's a mess.
  39. Para Sort of tl;dr is that if you haven't touched Junie by name or AI Chat, you probably haven't used it.
  40. Para There's several things going with it, maybe this helps https://www.youtube.com/watch?v=J8fHU4WFd_c
  41. nevet YouTube: From One AI to Any AI: JetBrains rethinks the approach to AI tooling | JetBrains
  42. Para Poor summarization is that 1) Junie and direct AI Assistant is soft deprecated 2) ACP based AI tools are now the way to go 3) Junie CLI is the new main entrypoint for all things AI 3) Junie CLI (which is also called just Junie) links to IDEA, which detects Magic, which links it to AI Assistant, which works as base for AI Chat, which allows for using Junie (CLI) managed ...stuff
  43. dreamreal You know, I find that telling me something via insta or youtube or whatever is remarkably ineffective for me
  44. * magla joined #java
  45. dreamreal I'm pretty much always like "yep, there's a video or whatever, cool" and I never watch them
  46. Para At least they start that video with admission that they screwed up :P
  47. Para Junie was supposed to be teh bestest prompt tool or whatever, turns out you shouldn't invent the same wheel everyone is inventing at the same time.
  48. Para ...or something. The history is eating the timeline and facts quite quickly.
  49. dreamreal heh
  50. Para Junie (CLI) is also heckin' buggy.
  51. dreamreal yeah, that sounds like a screwup :D
  52. Para My biggest issue at the moment is that Windows+WSL2 simply doesn't work with Junie CLI for custom models. Sure, not the most common use case but it does cover a non-insignificant chunk of uses.
  53. dreamreal As soon as you mentioned Windows I was like "hah-" and then WSL was the killshot
  54. dreamreal found your problem
  55. dreamreal most of my team relies on windows and a significant portion uses WSL2 to make it tolerable and they wonder why I just manage to ignore all their OS follies
  56. dreamreal I love it when they mention their OS, I know that it's no longer relevant to me, it's something they're screwing up
  57. Para The technical problem is that the *nix mount in IDEA is under Windows' magic path \\wsl.localhost\<fs>, but Something(tm) makes IDEA try to open that in CMD which is the original DOS based prompty thing, but those paths actually work in a reasonable way in PowerShell only.
  58. Para And somehow Windows version of IDEA just can't fathom that if I tell it to use PowerShell, maybe that's what I mean.
  59. dreamreal * dreamreal nods wisely pretending to care
  60. Para And why WSL2...well, the local inference software exists mostly for Linux only :D
  61. Para And I have Windows on my home PC because this is for watching cat videos and playing Factorio, not software development.
  62. * mixfix41 joined #java
  63. Para Friend of mine is already "subtly" hinting me towards NixOS, though.
  64. jreicher Para: what do you need WSL2 for? (I should say I'm not familiar with Junie, so I might be missing something obvious)
  65. dreamreal uh huh
  66. Para So maybe I'll just completely go insane in near future and turn into yet another continuation of Torvalds bros.
  67. jreicher oh I think you answered that. SOrry
  68. Para jreicher: Toolchains hate Windows, it's really nothing more than that.
  69. Para And the feeling is mutual, but alas...
  70. jreicher But does it have to be WSL2? I find MSYS2 is much better and works for almost all the same things
  71. Para I'm not polluting my Win11 any more than I have to. It's already a garbage fire, I don't need it to have any extra smells.
  72. jreicher I have a lot of sympathy for that, but you might want to make an exception. MSYS2 seems pretty hygienic to me.
  73. dreamreal I, uh, know of an office where the IT group keeps enabling and disabling virtual systems on the managed Windows installations...
  74. jreicher Eww
  75. Para If storage prices weren't what they are I'd be solving this with a dedicated Linux disk. Now I'm just angry at things.
  76. * MikeBux joined #java
  77. dreamreal Well, at least anger is fun.
  78. * metalmaniac joined #java
  79. * internecine joined #java
  80. internecine How many people here feel like they've brought significantly less value at work because of AI?
  81. dreamreal not me
  82. dreamreal have you?
  83. Para It's 80/20 to me.
  84. Para 80% of time is still farting around customer who can't decide things at all and 20% of the time what LLMs give me is a better baseline for doing actual stuff.
  85. Para So the actual "win" percentage is maybe 5% if even that.
  86. Para Sometimes it does things for me which are immediately useful and the kind of things I'd spend days on for little gain, but that's not really "it did two days of work in 30 minutes", it's "it created a thing in 30 minutes which no one wanted to do because it would've taken two days for just a nice to have thing"
  87. Para A good example would be custom binary format parser template for HexFiend, which I did earlier this week. Pointed Claude to docs, gave it the binary file spec and told it to Do The Thing.
  88. Para It's a nice thing to have but no one had cared enough earlier to actually do that. Plus Tcl is not exactly a common skillset either these days, so the extra time would've been in learning how it works for a oneshot script.
  89. * stfstfm_ joined #java
  90. * Odyss3us joined #java
  91. dreamreal I can't tell if I'm just really good at telling the LLMs what to do and scope and restraints, or if I'm just extraordinarily lucky
  92. dreamreal because honestly the horror stories about token consumption and how awful the code is, etc etc just rarely seem to apply to me: I don't exceed token allocations, and the code it generates is *generally* pretty useful, and when it's not, I'm able to head it off at the pass pretty quickly
  93. dreamreal which makes me think it's a skill issue, but I don't know if I'm, like, good at it or not
  94. Para There's a feedback loop on chat length. The longer one chat gets, the bigger the context and "thinking" seems to take a wild run after certain point.
  95. Para So if your initial prompt+context is already good enough it doesn't go off the rails and you get to keep your monies.
  96. * Betal joined #java
  97. * five618480339176 joined #java
  98. * hwpplayer1 joined #java
  99. * acidsys joined #java
  100. Chronos dreamreal: My guess is there's quite a bit of confirmation bias on "both sides."
  101. * punit01 joined #java
  102. * mindCrime joined #java
  103. * monkeyPlus joined #java
  104. * punit01 joined #java
  105. * punit01 joined #java
  106. * punit01 joined #java
  107. * domicron joined #java
  108. * Ragnor joined #java
  109. * hwpplayer1 joined #java
  110. * metalmaniac joined #java
  111. Zapek I have seen all kind of horrors: from people writing tickets using AI, then others implementing it using AI and finally others doing the reviews using... AI as well. in the end nobody knows how things work anymore. the favorite answer to any question has become "just ask claude"
  112. * monkeyPlus joined #java
  113. Tenchi Zapek: there exists a notion of hopeful optimism that a time will come when the industry will need, once again, the people with the knowledge of how to dig everyone out of this hole being dug 😅
  114. GreenResponse of course nobody gives a f--k about environmental strain so called AI takes on nature https://i.imgur.com/4J71nxE.jpeg
  115. * sponkz joined #java
  116. * Odyss3us joined #java
  117. * jonp` joined #java
  118. * hwpplayer1 joined #java
  119. * Anaphaxaway joined #java
  120. dreamreal both claims are doing a lot of work, for what it's worth
  121. * hwpplayer1 joined #java
  122. * iwtga joined #java
  123. jreicher javabot can't manage IRC memos, can it? I've checked the source and it doesn't seem to.
  124. javabot jreicher, what does that even *mean*?
  125. dreamreal it does not have a memoserv, no
  126. dreamreal does libera not have one?
  127. jreicher I don't mean not having a memoserv; I mean "forwarding" to memoserv.
  128. dreamreal oh, no, it does not
  129. dreamreal it's actually service-neutral, athame is only one ircd
  130. jreicher The nice thing about that is if I do something like "javabot tell dreamreal Pat Metheney is overrated" then other people here also see the message
  131. dreamreal indeed, although it DOES have a "say" operation
  132. dreamreal and a tell, actually
  133. jreicher Yeah but the tell is to recite a factoid, right?
  134. dreamreal tell is, yeah
  135. jreicher Can't find say...
  136. dreamreal I'd have to look through the operations again and I haven't looked at javabot's source for a while
  137. jreicher It looks like a message echo?
  138. jreicher (I'm looking at the source)
  139. jreicher javabot: say They all ask'd for you
  140. javabot jreicher, what does that even *mean*?
  141. jreicher Nope...
  142. dreamreal I dunno, I'd have to look, it's been a while, maybe I'm misremembering - I have only worked on maybe eight infobots :D
  143. jreicher https://github.com/evanchooly/javabot/blob/master/src/main/kotlin/javabot/operations/SayOperation.kt
  144. javabot jreicher's title: "javabot/src/main/kotlin/javabot/operations/SayOperation.kt at master · evanchooly/javabot · GitHub"
  145. jreicher Not that you would need my help finding it
  146. dreamreal I wasn't looking because it wasn't my need :D
  147. jreicher Oh wait I addressed the bot incorrectly didn't I
  148. jreicher ~say They all ask'd for you
  149. javabot jreicher, what does that even *mean*?
  150. jreicher Man...
  151. jreicher Oh I think I have to be an op. I overlooked that in the code before.
  152. * Anaphaxeton joined #java
  153. * hiredman joined #java