Time Nick Message 12:13 sfan5 merging #17721, #17615 in 15m 12:13 ShadowBot https://github.com/luanti-org/luanti/issues/17721 -- [no sq] Floating point precision issue related fixes by appgurueu 12:13 ShadowBot https://github.com/luanti-org/luanti/issues/17615 -- [don't squash] Some refactoring + Lua CLI proof of concept by sfan5 12:17 sfan5 so I've been wondering: do we consider #17718 to be in violation of https://github.com/luanti-org/luanti/blob/master/doc/developing/ai_policy.md? 12:17 ShadowBot https://github.com/luanti-org/luanti/issues/17718 -- Separate align_style for overlay tiles by andriyndev 12:17 sfan5 the policy says "Generating substantial amounts of code with AI is strongly discouraged and [...]. AI assistance may always be used for simple tasks, such as [...]". 12:18 sfan5 comparing that to "AI was used to generate the initial version, which was then carefully reviewed and improved manually." makes me tend towards it not being allowed 12:20 MTDiscord It says discouraged, not banned 12:23 MTDiscord If anything, something like https://github.com/luanti-org/luanti/pull/17733 and most likely un disclosed ai and just a giant all in one should be closed 12:23 sfan5 it should but that's not the point 12:24 sfan5 the point is where to draw the line between "discouraged" and "banned" 12:25 MTDiscord Useful and slop 12:25 sfan5 looking at #17352 it was worded like this to give the us coredevs leeway to decide, but I'm wondering on what basis we should be deciding 12:25 ShadowBot https://github.com/luanti-org/luanti/issues/17352 -- Proposal: Luanti generative AI policy by v-rob 12:25 sfan5 (the mentioned PR looks fine to me IMO) 12:33 ZenonSeth other than what wsor said about useful vs slop, I thought part of the policy of discouraging is to avoid people opening too many big PRs frequently to avoid a case where everyone can put together some quick llm generate 1k-2k line PR every other day 12:34 ZenonSeth i mean they don't even have to be big, just frequent llm generated PRs that are slop also contribute to that i think 12:53 sfan5 sneakily merging #17722 too for good measure 12:53 ShadowBot https://github.com/luanti-org/luanti/issues/17722 -- Android: throttle unzip progress updates so the completion signal isn't lost by 9LJ9 13:00 [MatrxMT] Undisclosed should just lead to permaban imo 13:02 [MatrxMT] As we're building community, not only software. A contributor not willing to admit to have used AI or not caring enough to read is not a contributor we want, I'd say 13:02 [MatrxMT] Plus the waste of our time 13:23 sfan5 will push this https://github.com/luanti-org/luanti/commit/1140cdc8359404e8f15e2f7198b5cfaf7d4fe617 if it works 13:43 sfan5 while the gcc_9 failure doesn't seem to be consistent it at least didn't fail with this commit 13:43 sfan5 so I'll push