Time  Nick      Message
08:58 sfan5     major documentation bug: our "Compiling on GNU/Linux" section includes instructions for Alpine, which isn't GNU/Linux
09:44 celeron55 i'd say it's close enough. they haven't even gotten rid of GCC as their default compiler, and musl and busybox mostly mimick GNU anyway
10:40 red-001   sfan5: I would consider Android 5 ancient, maybe anything below 7.1.2 is ancient not merely old
10:42 red-001   but maybe should move that up to 6, since that's not getting google service updates anymore it seems
10:43 red-001   and clearly the documentation needs to be changed to say GNU/LInux/ulibc/Linux/
10:43 sfan5     whether google play services is up-to-date isn't really our concern
10:51 sfan5     if our app wasn't suspended I could tell you how many users/active installations on very old android we have
10:53 red-001   I can't imagine it runs all that well on most of those surely, quite rude of google to cutoff the metadata feed just because they took down the app
10:55 sfan5     oh actually https://kitsunemimi.pw/tmp/serverlist_stats_2025-11-06.txt
10:55 sfan5     careful estimate <2% of users on older than Android 7
10:59 celeron55 chinese phones and tablets used to run older androids even though they were fairly powerful. and they don't self destruct - as long as you don't drop them on a rock, they still work
11:00 red-001   not 5 or 6 really, they might run 10 or something like that
11:00 red-001   seems there's still a fair few on 6
11:00 red-001   and not really anyone on 5
11:00 celeron55 more like android 8
11:00 celeron55 think of huawei honor 8 or something like that
11:01 red-001   " I would consider Android 5 ancient, maybe anything below 7.1.2 is ancient not merely old"
11:01 red-001   hence perhaps I would consider Android 8 to be merely old
11:01 celeron55 maybe. altough that phone was originally released with android 6
11:02 red-001   if it got upgraded to 9 or 10 that would for me put it into unsupported but not actually that old category, like Windows 10 versions
11:03 red-001   sfan5: more Windows 7 users than I expected, I guess better avoid breaking compatibility for them
11:04 celeron55 of course one chan shout "think of cybersecurity" but that's not necessarily realism for a poor family with many kids. it works fine as a device to run luanti and for making phone calls, and will do that essentially forver as it supports LTE
11:04 celeron55 forever*
11:04 celeron55 can*
11:04 celeron55 (i can't type today)
11:05 celeron55 (or actually, spell)
11:05 red-001   well you didn't have to admit the second one, I was perfectly willing to believe you spelled the words perfectly and merely typed them wrong
11:06 red-001   eventually the batteries will fail
11:07 red-001   or the screen will get cracked or moisture ingress or what have you
11:07 celeron55 that's true. the replacement interval is essentially the point where buying a more recent used device is cheaper than buying a battery or screen for the old one
11:07 red-001   actual flash memory failure is a big one too, had a otherwise perfectly mangled phone fail like that, was very sad
11:08 celeron55 altough, the screen doesn't break on its own, so it only comes down to the battery
11:08 red-001   and flash does fail on its own
11:08 red-001   *more or less on its own
11:09 red-001   still have the failed mainboard somewhere, was super annoying cause I got the higher end model of that phone with 128gb and never even used up all the storage
11:09 MTDiscord <.zenonseth> lots of those old phones used to have replacaeble batteries too, was one of the best things about them, my phone as late as 2016 had one
11:10 celeron55 anyway, it would be useful to define what's the age of android devices that we think can be dropped. i think going above 10 years probably isn't necessary. the honor 8 was released in 2016 but that's only start of production. they were probably produced for 2 to 3 years
11:11 celeron55 this means dropping android <=6 would probably be reasonable by now. so i agree with red-001
11:11 red-001   figured out it was flash when I was trying to understand why the EDL flash mode didn't work, eventually read the error log instead of just assuming it was on me and turns out it couldn't bootstrap the UFS
11:12 MTDiscord <.zenonseth> ive missed the start of the conversation, but i assume there's something gained by dropping android <6 ?
11:12 red-001   I suppose it could have been physical failure, I didn't try reballing the BGA chip on there, don't have the tools to do it and my data wasn't worth that much
11:12 red-001   celeron55: not even saying to drop it, more of an observation, if we run into issues with adding a new feature however and it's cause of support for android <= 6, I think can drop it then
11:13 celeron55 i don't think there's any immediate gain for us, so there's no need to drop right now
11:13 celeron55 but it's good to keep the threshold in mind if things come up
11:14 red-001   like the engine I believe doesn't support Windows XP or what have you I believe, Android 5.0 feels about as obsolete as XP
11:15 red-001   contrast that with windows 10 which is technically EoL but still widely used
11:15 celeron55 it's the same thing as with C++ versions. once support in the field is met by our definition, then a developer can propose a version requirement bump
11:15 red-001   or even 7
11:15 celeron55 (given that it gives some immediate benefit)
11:16 red-001   https://github.com/MultiCraft/MultiCraft/commit/f11b304d6fda7671cb54af4a51b66e022b741adf what is going on in MultiCraft world
11:16 red-001   oh
11:16 red-001   quickly Matrix bridge is down everyone say why you like IRC better
11:16 MTDiscord <jordan4ibanez> It has more text
11:16 red-001   "core.change_password() CSM API" is crazy work
11:17 red-001   I really want to know what that API is for
11:28 red-001   I'm surprised that for jsoncpp you can't just use string view and let it construct a std::string on older versions
11:30 [MatrxMT] <the4spaceconstants> matrix bridge still saw that
11:38 red-001   oh right explicit constructor
11:49 MTDiscord <rollerozxa> there's basically no reason to drop old android versions unless the NDK forces us to
11:50 MTDiscord <rollerozxa> the upcoming NDK r30 LTS still supports android 5, but I think the next one might bump it up to android 6
16:04 sfan5     13:03 <red-001> sfan5: more Windows 7 users than I expected, I guess better avoid breaking compatibility for them
16:04 sfan5     i think windows 7 users are already using "extension packs" (I made this name up) to patch in newer functions these days
16:05 sfan5     or basically no reasonable apps would still work
16:16 MTDiscord <luatic> "service packs"?
16:17 sfan5     no those are from microsoft
16:17 sfan5     but I have heard of community-made stuff that allows you to run software targeting w10 on w7
16:19 MTDiscord <luatic> ah
16:21 MTDiscord <wsor4035> windows xp community does the same, but a bit more insane
16:25 MTDiscord <rollerozxa> I'm pretty sure luanti still works on a fully up to date version of windows 7, with UCRT and network security stack updates
16:46 MTDiscord <rollerozxa> yeah
16:46 MTDiscord <rollerozxa> https://cdn.discordapp.com/attachments/747163566800633906/1543300812363530382/image.png?ex=6a945e62&is=6a930ce2&hm=05ac486d522ff65f7629210e36fbdb4daba7c41b000caff5272aa7973e41fed2&
16:47 MTDiscord <rollerozxa> obviously I know it's not officially supported by luanti but libc++ still supports windows 7+ so there should be no reason it shouldn't work unless that begins to change
18:25 sfan5     merging #17425, #17514 soon
18:25 ShadowBot https://github.com/luanti-org/luanti/issues/17425 -- Lua API: Improve ItemStack handling by SmallJoker
18:25 ShadowBot https://github.com/luanti-org/luanti/issues/17514 -- [no sq] Update jsoncpp to 1.9.8 + fixes by sfan5
18:37 sfan5     done
18:41 sfan5     will push something to fix the prometheus-cpp build failure too
18:57 sfan5     >install libcivetweb-dev
18:58 sfan5     >project refuses to configure because /usr/bin/civetweb is missing
18:58 sfan5     >install main package too
18:58 sfan5     >random linker errors
18:58 sfan5     guess not
19:03 sfan5     oh cool the docker image has a different solution
19:03 sfan5     someone else can fix it
21:28 red-001   #17515 uses a plugin to fetch the CA certs now
21:28 ShadowBot https://github.com/luanti-org/luanti/issues/17515 -- Add CA bundle for TLS verification. by red-001
21:29 red-001   no point caching this in tree if the build needs internet access anyway