Time Nick Message 09:19 sfan5 merging #14300 in 10m 09:19 ShadowBot https://github.com/minetest/minetest/issues/14300 -- [no squash] NetworkPacket and other improvements by sfan5 13:13 grorp merging #14211 in 15 min 13:14 ShadowBot https://github.com/minetest/minetest/issues/14211 -- Only pause rendering if the Android activity is stopped by grorp 15:35 grorp It seems odd that there are still three regressions from #13146. 15:35 ShadowBot https://github.com/minetest/minetest/issues/13146 -- Inventory mouse shortcut improvements by OgelGames 17:25 Sokomine problem with latest git and signs_lib (at least on yourland server) with tons of warning messages: ERROR[Main]: generateImagePart(): invalid X or Y offset for part_of_name="[combine:291x102:10,0=signs_lib_color_16px_0.png:138,0=signs_lib_color_16px_0.png:266,0=signs_lib_color_16px_0.png:10,0=signs_lib_font_16px_49.png:14,0=signs_lib_font_16px_66.png:20,0=signs_lib_f 17:27 sfan5 can you put the entire error message on a pastebin? 17:29 Niklp see #14312 17:29 ShadowBot https://github.com/minetest/minetest/issues/14312 -- MineClone 2 "mob head" broken by tile generation changes 17:35 sfan5 signs_lib seems to be doing something different hence the question 17:38 Sokomine sfan5: added comment to that issue you mentionned 17:40 Sokomine sfan5: we suspect signs_lib of beeing at least partially responsible for the horrible lag when walking down the main street on the yl server 17:42 sfan5 do you have an example with signs_lib_color like above? 17:54 Sokomine sfan5: looking for that. hm. seems these error messages spread over multiple lines. perhaps due to the textures beeing so big. shall i combine into one string? 17:56 sfan5 was just wondering if there's any negative numbers in there 17:58 Sokomine sfan5: can't find one in that particular one. but in other textures that likewise caused an error 18:01 Sokomine sfan5: added further error messages and hope they help 18:02 MTDiscord Sokomine: Have you tried #14314 yet? 18:02 ShadowBot https://github.com/minetest/minetest/issues/14314 -- Two tiny doc things, combine regression fix by appgurueu 18:03 sfan5 >ERROR[Main]: generateImage(): Failed to generate "[combine:32x64:56,20=blank.png" 18:03 sfan5 ok now what about this 18:04 Sokomine no idea. but feel free to join the your land server. many error messages there. but you might not make it to anywhere outside the main town as there are many posters and signs 18:06 MTDiscord Sokomine: Is there something like a feature that lets you toggle capes? 18:07 MTDiscord (It looks like either (1) someone made a mistake, this is a bug; (2) someone hackily disabled capes; (3) there is some kind of feature at play to disable capes at runtime hackily.) 18:07 Sokomine luatic: the capes might be actually my fault and up to me to fix. they can in theory be worn by NPC (though to my knowledge none there in the vicinity does) 19:17 sfan5 found more efficiencies in [combine by working on the code for 5 minutes... 19:18 sfan5 inefficiencies* ofc 19:18 MTDiscord this code should be reworked to use proper limits in the for loops 19:18 MTDiscord then we'll get the proper behavior naturally 19:19 MTDiscord i can do that 19:20 sfan5 honestly half of the file should be thrown away and rewritten 19:22 Sokomine if you have imporovements that might indeed help a lot in many ways 19:25 chmodsayshello When updating a PR, should I do so by merging master into my branch or rebase? 19:25 sfan5 if you want us to merge by rebase, the latter 19:25 sfan5 otherwise choose one 19:27 sfan5 @luatic my current suggestion https://github.com/minetest/minetest/commit/7829584aabf02a48aebcc01c0ff845c571c83824 19:28 sfan5 (also dropped the unnecessary copy, all images returned by generateImage should are already ECF_A8R8G8B8) 19:29 sfan5 s/should // 19:30 MTDiscord I see. What I was thinking was restricting the blitting for loop to the intersection area between the source and destination. 19:30 sfan5 good idea regardless 19:36 sfan5 http://sprunge.us/9iupLH?diff forgot a tiny thing 19:36 Sokomine sfan5: your proposed patch eliminates the scrolling warning messages for me as well 19:36 sfan5 @luatic if you agree on the approach you can import that into your PR 19:37 Sokomine another thing i havn't been able to pinpoint: recently i get "connection timed out" errors when connecting to the your land server. pretty early on and pretty fast 19:39 MTDiscord sfan5: I'm not sure. It depends on whether we want to consider blitting with zero overlap a feature, in which case we don't want a warning for it. I'm inclined to say it is a legitimate feature. Besides that it looks good to me though. 19:41 sfan5 I think it still has the possibility to help modders find bugs 19:41 sfan5 requesting the engine to do something that has no effect objectively only has downsides 19:42 sfan5 starting from trivial circumstances like "more data to send on the network" 19:43 MTDiscord It has the upside of you not having to do the work yourself, if you have a legitimate use case where something might go out of bounds. 19:43 MTDiscord That said, I'm not sure that texture modifiers should really be compared to the "blit" methods found in other game engines. 19:43 MTDiscord I can live with the warning. 19:45 sfan5 should I open a PR or want to amend yours? 19:48 MTDiscord whatever you prefer. you can also edit mine to replace my commit with yours. 19:51 sfan5 done 20:26 sfan5 this emergeBlock second parameter is a major footgun 21:21 Krock sfan5: do you know by any chance why we're invalidating peer IDs in the first palace? 21:21 sfan5 in playersao? 21:21 Krock in regard to #14313 21:21 ShadowBot https://github.com/minetest/minetest/issues/14313 -- Crash on exit with MineClone 2 in "Host Server" mode 21:21 Krock right 21:22 sfan5 if the peer is already disconnected that would be a reason 21:22 sfan5 (at that point) 21:22 Krock as long PlayerSAO exists I'd expect it to accept connections 21:22 Krock s/connections/packets/ 21:24 sfan5 so the flow is Server::handlePeerChanges -> Server::DeleteClient -> PlayerSAO::disconnected 21:24 sfan5 no longer makes sense to send packets at that point 21:24 sfan5 issue is you can't delete a SAO instantly due to references 21:25 Krock hmm 21:25 sfan5 curiously the invalidation of RemotePlayer only happens later in PlayerSAO::unlinkPlayerSessionAndSave 21:26 Krock RemoteClient::getPlayerSAO could as well return nullptr when the thing is gone 21:26 sfan5 I wouldn't bet on someone having put thought to designing the lifecycle of these objects 21:26 sfan5 rather it just happened to come into existence like this and mostly works 21:26 sfan5 (like many parts of the engine) 21:26 Krock story of most of Minetest's internals 21:46 nrz Sfan5 i remember starting clean up this part 4 years ago, but i didn't moved much because it was fragile 21:47 nrz I decoupled many session and sao links, but some liké you just show remains. 22:35 MTDiscord Can it use some or none instead?