Time Nick Message 10:18 pgimeno can someone please share the patch to disable zstd saving? 10:20 sfan5 http://sprunge.us/t45EiP?diff 10:32 pgimeno thanks! 12:43 MTDiscord what did it break/why was the patch needed? 12:43 MTDiscord compatibiltiy with 5.4.0 12:43 MTDiscord so with 5.5 release it will be live? 12:43 MTDiscord or no? 12:50 sfan5 it didn't break anything 12:50 sfan5 you just can't use new maps on older clients, never could 13:00 pgimeno maybe it would be a good idea to add read support in one version and write support in the next 13:01 pgimeno that way, if the new version is released with critical problems, people can go back to the previous one 13:02 pgimeno to be clear, I mean to disable zstd writing in this version 13:03 pgimeno is zstd the only reason to bump the map version? 13:08 sfan5 yes 13:15 MTDiscord when was the last time a map version incompatibility like this was implemented? 13:15 MTDiscord and more importantly, how was that handled? 13:17 MTDiscord 5.1 or 5.2 if i recall correctly 13:19 sfan5 0.4.16 and it wasn't any different 13:25 MTDiscord https://github.com/minetest/minetest/commit/049256573b6fc379ab4cd1f867f33504de6abeee#diff-a9b4955fbefbb32a04ffcf1df1b952a5898eb223ab2d8d61e1640f4b7fd4c443 version 28 was 5.1, thats what i was remembering 19:12 x2048 So quiet here... erlehmann, if you are here, I saw you posted a link to property based testing, and I'm wondering if discussion went anywhere. 19:15 erlehmann x2048 the person working on unit tests is josiah_wi, i advise you to check how that is going 19:16 erlehmann x2048 do you want advice for something specific? like have specific code you want to test and do not know how? 19:33 x2048 erlehmann, well, I was just thinking how we could apply PBT to Minetest as a whole and had two ideas, so I wondered whether there's work ongoing for hollistic testing and whether it makes sense to start with it. 19:35 x2048 One idea is to create a 'fuzzer' game that would continuously play sequences of game actions, like modifying world or moving enities. 19:35 x2048 Another idea was to create a bot client. 19:44 erlehmann x2048 the bot client would be a good idea for the CI. look into waspsaliva, it is a client made by cora. 19:44 erlehmann x2048 you know cora 19:44 erlehmann right? 19:45 erlehmann x2048 regarding a fuzzer game, if --random-input was deterministic with a seed that would help A LOT 19:45 specing Test_User has a bot client 19:45 erlehmann or how the option is called 19:45 specing (headless) 19:45 erlehmann oh nice how 19:46 erlehmann i mean i assume anon5 has some kind of bot client too 19:46 specing I believe that it involved a lot of swearing 19:46 specing due to the interesting things that happen with authentication 19:46 erlehmann show and tell 19:46 erlehmann where is the code lebowski 19:46 specing contact test user for that 19:47 erlehmann specing, literally how 19:49 specing how what? Ask him 19:52 erlehmann where? forum, irc, oysterity, somewhere else? 19:52 specing this without -dev 20:59 x2048 Had to step away... Thank you erlehmann, specing. I'll try to find josiah_wi and Test_User. erlehmann, have you got experience with PBT? 22:39 MTDiscord x2048 I keep track of this development channel from the relay. I am not currently focused on making tests in general, but rather bringing back a unit testing setup for IrrlichtMt, because I wanted to write some quick functionality checks for WebP stuff and found there were no tests to use as a reference. 22:42 MTDiscord I'm taking the time to review and modify some of the old tests that were removed so that it can be useful to other devs. I'm leaving out any tests that I can see are not unit tests or are particularly flimsy. So most of what's getting restored are simple functionality tests of core Irrlicht data types, that had better not break or there's a problem.