Time Nick Message 18:01 proller https://github.com/minetest/minetest/pull/11910 - 3 months of nothing! 18:41 v-rob Any thoughts on whether we're having that core dev meeting soon? 18:45 MTDiscord Oh, my, goodness. Can I join? 18:48 rubenwardy it's public in the channel 18:48 rubenwardy How about this Saturday? 18:48 rubenwardy v-rob: make sure to add things to the agenda if there's something you want to discuss 18:48 rubenwardy so we don't forget 18:48 rubenwardy !dev Meetings 18:48 ShadowBot Meetings - Minetest Developer Wiki -- http://dev.minetest.net/Meetings 18:50 v-rob I'm open Saturday, so that's good for me 18:51 MTDiscord When will 32 bit builds be deprecated? 18:54 v-rob Will they? I've never heard anything about that 19:00 ROllerozxa why should they be deprecated? 19:03 MTDiscord macos, ubuntu, and windows (latest versions) all dont even have 32 bit OS versions anymore. that said, no reason to drop support unless there is some advantage to it 19:04 erlehmann what advantage should there even be? 19:05 MTDiscord code cleanup probably? dunno. only outlier is andriod which appears to still support 32bit, but isnt very prevalent 19:06 MTDiscord i don't think we should drop android support 19:06 MTDiscord it's quite popular 19:07 erlehmann i'd be very sceptical of claims that writing non-portable code is “code cleanup” 19:07 erlehmann after all. minetest doesn't just run on x86 and x86_64 19:08 ROllerozxa how would dropping 32-bit windows binaries lead to code cleanups? 19:09 ROllerozxa only thing it would mean is one less build for each release 19:09 MTDiscord 10% of andriod is 32bit in 2019, or so a quick google says 19:09 v-rob I don't think any code would change at all, really 19:10 ROllerozxa keep also in mind minetest android arm64 builds are rather unstable, which is why the 32-bit armeabi-v7a builds are still recommended 19:11 MTDiscord yeah, doesnt the play store force 64bit builds on 64bit devices tho? 19:11 erlehmann also using 32bit binaries on 64bit systems can give you some small performance boost IIRC 19:12 rubenwardy Doubt 19:12 MTDiscord doubt 19:12 v-rob Anyhow, we still support Windows below 11, so it's not like 32-bit has disappeared. I'm pretty sure most Linuxes still support 32-bit. 19:12 v-rob Although I'm no Linux user myself. 19:12 MTDiscord seems unlikely any ubunutu based distros will offer 32bit versions 19:13 ROllerozxa erlehmann: really? I'd expect the opposite to be the case 19:13 erlehmann „ubuntu based” lol 19:13 Krock 32-bit versions are becoming difficult to get 19:13 erlehmann usually stuff is debian-based, not ubuntu-based 19:13 rubenwardy I don't necessarily think there's a reason to deprecate or drop support for 32bit, but we could not produce official builds or make them less prioritised on the website 19:14 rubenwardy If we had an installer, we could potentially make it 32bit and then install the right version based on the system, which would clean up the website UI for Windows 19:14 MTDiscord pop os, mint, elementary os. some big ubuntu based ones 19:14 v-rob Well, Ubuntu isn't everything. 19:14 MTDiscord never said it was 19:15 v-rob Where is our minimum supported OS stuff? 19:16 v-rob I can find stuff on the wiki, but it says XP is still supported, which is patently false (I've tried it myself :) ) 19:16 Krock Minetest 5.0.0 did work on XP. Don't know what's changed since then. 19:17 v-rob I think it works up to 5.2.0, or maybe it was 5.3.0 19:17 ROllerozxa minimum supported windows version seems to be windows 8, from looking at the download page 19:17 v-rob Oh, so it does. I'm willing to bet Windows 7 still works, though 19:18 v-rob I like how we don't even officially support Windows 11 =P 19:18 ROllerozxa ah, the joys of explicitly listing versions instead of just being like "windows 8+" 19:19 rubenwardy Supported doesn't mean works 19:19 MTDiscord ^minetest on windows in a nutshell 19:19 rubenwardy Something can work but not be supported, or can be supported and not work. Supported just means that we'll accept issues, fix things, and provide help 19:20 v-rob True 19:21 erlehmann rubenwardy Jonathon, i do not know, but if pointers are twice the size, doesn't that imply that your object code is larger and your instruction cache and memory bandwith is effectively smaller? 19:21 erlehmann linux has the x32 ABI for that reason 19:22 erlehmann https://en.wikipedia.org/wiki/X32_ABI 19:22 v-rob Anyhow, we probably don't need to change how we do anything about 32-bit builds unless there's anything significant that it would do for us. 19:22 erlehmann > It allows programs to take advantage of the benefits of x86-64 instruction set (larger number of CPU registers, better floating-point performance, faster position-independent code, shared libraries, function parameters passed via registers, faster syscall instruction) while using 32-bit pointers and thus avoiding the overhead of 64-bit pointers. 19:22 erlehmann > The best results during testing were with the 181.mcf SPEC CPU 2000 benchmark, in which the x32 ABI version was 40% faster than the x86-64 version.[3][4] On average, x32 is 5–8% faster on the SPEC CPU integer benchmarks compared to x86-64. 19:23 erlehmann rubenwardy Jonathon what do you say to that? 19:23 v-rob Ooh, pointer size optimizations. What is this, DOS with different compile-time memory models? :D 19:24 v-rob Still, 32-bit builds for Minetest exclusively use the x86 instruction set, no mixmatch between x86_64 and 32-bit pointers 19:24 erlehmann what i'm saying is, it depends on the workload. you get a penalty for the pointer size and you need to offset it somehow. 19:25 erlehmann which you can do, of course, by taking advantage of lots of 64 bit math or so. or needing more than 4GB RAM per process? 19:25 erlehmann as always, benchmark. 19:25 v-rob 32-bit might be faster, and 64-bit might be faster. I'd guess that it's largely system and processor dependent. 21:24 MTDiscord Oh my God I'm sorry for starting this 21:24 erlehmann jordan4ibanez so why did you ask in the first place, like what was the thought behind it? 21:27 MTDiscord Looking into the code clauses of if aarch64 in one of the files thinking that, wow that's going to be 98% of the time, why is this even still a thing? 21:27 erlehmann uh, what? 21:28 erlehmann no idea how you get to 2% of users who have an aarch64 system, but if you wanna drop support for that, i'm pretty sure some mac users will be upset lol 21:28 erlehmann and also some reform2 users 21:41 v-rob How in the _world_ do you log into the dev wiki? I never got any email to set my initial password, and attempts to reset the password results in no emails either. Maybe the wrong email got used somehow? I don't know... 21:42 rubenwardy you should have received an email, if not it'll be because celeron55's server is broken 21:43 MTDiscord No, I meant the inverse of what you're saying, aarch32, win32, Linux 32, check etc etc etc