Time  Nick       Message
00:35 paramat    tested, merging #7495 in 15 mins (trivial)
00:35 ShadowBot  https://github.com/minetest/minetest/issues/7495 -- Mgvalleys: Fix 'mgvalleys_spflags' setting name by paramat
01:08 paramat    aarg forgot, merging now
01:11 paramat    done
09:53 ANAND      Why do formspec elements use v2s32 instead of v2f32 internally?
09:54 ANAND      for position
09:54 ANAND      How are they able to store float values?
09:55 rubenwardy pixels
09:55 ANAND      ah..
09:55 rubenwardy there are two spaces - inventory space and pixel space
09:55 rubenwardy and it translates from the former to the latter with parsing
09:55 ANAND      understood, thanks :)
09:55 ANAND      ok
10:00 rubenwardy might be worth getting something like uservoice, then moving features there: https://leaflet.uservoice.com/forums/150880-ideas-and-suggestions-for-leaflet
10:01 rubenwardy hmm, although technical discussion
10:01 ANAND      https://github.com/minetest/minetest/blob/master/src/gui/guiFormSpecMenu.cpp#L1215
10:01 ANAND      Why is stof() and float typecasting used here? Aren't they being assigned to a variable of type s32 anyways?
10:02 ANAND      Also, why the 7.0/30.0
10:02 ANAND      ?
10:02 rubenwardy I guess to defer the rounding
10:02 rubenwardy yeah
10:02 ANAND      hmm
10:03 rubenwardy for example,     0.5*100 + 3     is different to        0*100+3
10:03 rubenwardy :)
10:03 rubenwardy even if they're both rounded
10:03 ANAND      True
10:04 rubenwardy like,          round(0.5*9+1)=6    and     round(0.5)*9+1 = 10
10:04 ANAND      :)
10:04 red-001    100 / 0.4
10:05 ANAND      Oh yes
10:43 Wuzzy      https://github.com/minetest/minetest/pull/7251
10:44 ANAND      #7251
10:44 ShadowBot  https://github.com/minetest/minetest/issues/7251 -- Add function `minetest.read_schematic` by paly2
10:48 red-001    unable to replicate #7334, possible close?
10:48 ShadowBot  https://github.com/minetest/minetest/issues/7334 -- Player backend migration skips some players.
11:00 nerzhul    how is this possible ? the loop just loop over the file :p
13:42 Wuzzy      about #7251: This would be very useful for my mod [schemedit] because then I could finally make schematics loadable
13:42 ShadowBot  https://github.com/minetest/minetest/issues/7251 -- Add function `minetest.read_schematic` by paly2
13:42 Wuzzy      so can someone review above PR plz? already has 1 approval by SmallJoker
13:52 ANAND      What prevents assigning fractional positions to the container[] element?
13:52 ANAND      e.g. container[1,1.5; positions the container at 1,1 and not 1,1.5
13:53 ANAND      https://github.com/minetest/minetest/blob/master/src/gui/guiFormSpecMenu.cpp#L287
13:53 ANAND      ^ Here's the function that parses the container formspec element (provided for convenience)
13:55 Krock      ANAND, we have an issue about that
13:56 Krock      it only deals with entire numbers but not floats
13:56 Krock      #7452
13:56 ShadowBot  https://github.com/minetest/minetest/issues/7452 -- Container[] doesn't work with fractional values
13:56 Krock      let's see whether it's actually just one line to change. testing.
13:56 ANAND      Yes ik, but is that fixable?
13:56 ANAND      nvm
13:57 ANAND      I placed container_stack.push below the pos_offset.{x,y} lines
13:57 ANAND      Got weird results... :/
13:58 ANAND      I'm sure the fix is quite simple, because all the other formspec elements support fractional values too
14:14 Krock      hmm.. it's more than just changing one line. the positions are all calculated with v2s32 vectors
14:20 Krock      ANAND, do you know whether it's intended that labels are unaffected by containers?
14:21 Krock      ^ rubenwardy
14:23 Krock      nvm, gotta have to find out what UpperLeftCorner is supposed to do
14:49 nerzhul    Hello, i just opened #7496 to track cherry picked commits for a new android build
14:49 ShadowBot  https://github.com/minetest/minetest/issues/7496 -- Stable 0.4.17.2 preparation (Android only ?) by nerzhul
14:51 paramat    nice :)
14:54 nerzhul    just to answer
14:54 nerzhul    i'm always on 0.4.1
14:55 nerzhul    0..17
14:55 nerzhul    erfff
14:55 nerzhul    0.4.17
14:55 nerzhul    sorry i rebuild some electricity in my house this afternoon, my hands are tired :)
14:56 nerzhul    games/minetest_game> git status
14:56 nerzhul    HEAD detached on 0.4.17 (d6a6686)
14:59 nerzhul    build & game are working well, now building android part
15:01 nerzhul    note: we forgot to update deps on last Android release, i will update them (curl, sqlite at least)
15:24 nerzhul    Krock: we didn't rewrote the float serialization over network yet?
15:25 Krock      not yet
15:25 nerzhul    what was the current proposition ?
15:25 Krock      there were some ideas but they always have a performance penalty
15:25 Krock      or precision issues
15:26 nerzhul    can you remember me what are the current propositions ?
15:28 Krock      nerzhul, this https://github.com/minetest/minetest/pull/6312#issuecomment-339429303  and sqrtf(x)
15:30 Krock      I did some precision measurements with sqrt() using Lua, but it wasn't too great
15:31 Krock      found the script -> https://pastebin.com/raw/L7nBhD3y
15:37 Krock      So does #7496 work now? Is it tested?
15:37 ShadowBot  https://github.com/minetest/minetest/issues/7496 -- Stable 0.4.17.2 preparation (Android only ?) by nerzhul
15:38 Krock      nvm, it seems that you're on it already
15:45 nerzhul    Krock i'm on yes
15:47 nerzhul    nice singleplayer load properly
15:47 nerzhul    it's a good release now :)
15:49 nerzhul    trying to rebuild it another time to ensure it's good
15:49 nerzhul    the game is quite fast in debug mode, it's nice
15:49 srifqi     Any suggestion on this? https://github.com/minetest/minetest/commit/b1e58c9c35ae3eb6167a0745086cdb0ddb1dd9d7#commitcomment-29532111
15:49 srifqi     nerzhul: nice!
15:50 nerzhul    Krock: why not split the float in two parts ? one int64t with the int part, and another u64 which is 1 billion * fractional part
15:51 nerzhul    this permits 9 numbers after comma it should be sufficient
15:51 Krock      srifqi, after looks definitely better. Taller buttons were already requested earlier
15:53 srifqi     Is it? (^_^"
15:53 srifqi     Okay, I'll create a PR.
15:53 nerzhul    Krock what do you think about my idea ? it can permit to have a very high precision, we can do 1000 billions to be as precise as needed
15:53 Krock      Yes, Wayward1 had a hard time pressing those on a small screen
15:54 Krock      nerzhul, increasing the float value to 16 bytes? Definitely possible but proportional behaviour does surely not cover most of the float range
15:54 nerzhul    12 numbers after comma should be sufficient
15:54 Krock      right now it's 4 bytes
15:55 Krock      so from 10^12 to 10^-12 and negative values should be covered
15:55 paramat    srifqi +1
15:56 nerzhul    negatives are in the first int64
15:57 nerzhul    then uint64 for decimals
15:57 nerzhul    we can use a int32 to be as precise as float
15:57 nerzhul    and keep int64 for double
15:57 nerzhul    it's better
15:58 nerzhul    int32 a = (int3
15:58 nerzhul    oops
15:59 nerzhul    int32 a = (int32) f; uint32 d = 1000000000* (f - a);
15:59 paramat    srifqi maybe even extend the button boxes upwards a little too, since players will tend to have their thumb centred on the icons?
15:59 nerzhul    4 bytes are not important for the float context
16:00 nerzhul    and we have 9 decimals, it's sufficient
16:00 nerzhul    our coordinate are not in nanometers :p
16:02 nerzhul    what are the attentions point on the android build ?
16:02 nerzhul    i tested singleplayer & multiplayer
16:02 nerzhul    is there anything more to test ?
16:04 Krock      using a single u64 plus sqrtf we could reach the same precision (value range from 7E19 to 9E-19)
16:04 srifqi     Done. #7498
16:05 ShadowBot  https://github.com/minetest/minetest/issues/7498 -- Android: Resize jump and crunch buttons' touch target by srifqi
16:05 Krock      whereas values would be off by 0.1 per billion
16:05 srifqi     paramat: How much?
16:05 nerzhul    sfan5, paramat, rubenwardy if you can look at #7410 ? it's mainly having your opinion about the PR, the code is correct
16:05 ShadowBot  https://github.com/minetest/minetest/issues/7410 -- Modernize lua read (part 2): C++ templating assurance by nerzhul
16:05 nerzhul    Krock, it's same precision as my model, sqrtf is more complex than just calculating simple things like i proposed :)
16:07 Krock      yes, same precision with a much smaller value range 1E9 .. 1E-9
16:08 nerzhul    exact, but is this needed, we are not so precise on coordinates
16:08 nerzhul    testing the release version now
16:11 srifqi     There are 777 issues and 111 PRs.
16:11 srifqi     7 : 1 ratio
16:12 Krock      not necessary a good thing :/
16:12 srifqi     ._.)
16:15 nerzhul    https://play.google.com/apps/testing/net.minetest.minetest
16:16 nerzhul    can you test you have access to the beta program ?
16:27 Krock      will test
16:29 Krock      * "We know that 0.4.17 ...
16:30 Krock      Media loads.. joining server with existing account. WIll test singlepalyer afterwards
16:32 nerzhul    when you are okay, post approvals on the PR, i will merge, cherry pick the release commits to master & publish app
16:33 nerzhul    we currentlyu have 7k installs on a total of 246k
16:33 celeron55  installed the non-testing version. singleplayer failed like was described; updated to the beta release, singleplayer runs (with some GL_INVALID_ENUM errors but runs anyway)
16:34 Krock      singleplayer works too. joining a new server
16:35 nerzhul    and: it's the good MTG :)
16:35 nerzhul    like previous release
16:35 nerzhul    (you can unzip the release apk to verify it)
16:36 nerzhul    i think to prevent future problem we should use the beta program to publish APK to have some testers, and publish to production after validation of 2 or 3 different people, what do you think about this in the release process ?
16:36 Krock      sounds good
16:37 sfan5      doesn't google play support partial rollouts?
16:37 nerzhul    rollout to upgrade, not downgrade
16:37 nerzhul    downgrade is not possible
16:37 nerzhul    and i performed a full rollout upgrade like every release, but it was a failure
16:37 rubenwardy Would be good to use betas in future when releasing release candidates at the start of a freeze
16:37 nerzhul    rubenwardy, noted
16:38 sfan5      what I mean is roll out something to 10% of users, then check the "android vitals" tab to see if error rates have increased
16:38 nerzhul    but remember the version code should be updated in beta too, i think. (except if i can post a debug release, but i didn't tested it)
16:38 sfan5      if not, roll it out to the remaining 90%
16:38 nerzhul    sfan5 10% is not sufficient for us, but it can be an indiciator, but it's too late, we need to have beta before :p
16:39 sfan5      sure this is not the best way
16:44 nerzhul    sfan5 can you take 5-10 mins to test the beta app
16:44 nerzhul    if it's okay to you and Krock i can promote to production
16:44 sfan5      if you tell me how to install it, sure
16:44 nerzhul    https://play.google.com/apps/testing/net.minetest.minetest/
16:44 nerzhul    go there on your smartphone or your browser, then install the app
16:45 nerzhul    or upgrade on the play store
16:45 rubenwardy I'm installing the update
16:45 srifqi     nerzhul: The beta app seems to be working on my phone.
16:45 rubenwardy I can confirm it works now
16:45 Krock      however, I uninstalled the 5.0-dev build I had - to ensure there aren't conflicts (dunno about the directory structure)
16:45 nerzhul    nice srifqi
16:47 sfan5      nerzhul: works
16:49 nerzhul    nice, is this okay for a release then ? we have 4 approvals :)
16:49 nerzhul    (on the playstore, not MT)
16:49 nerzhul    and merging #7496 too as it's based on it
16:50 ShadowBot  https://github.com/minetest/minetest/issues/7496 -- Stable 0.4.17.2 preparation (Android only ?) by nerzhul
16:50 sfan5      no need to officially bump the version to 0.4.17.2 btw
16:50 sfan5      (IMO)
16:50 nerzhul    yeah, i don't think it's needed
16:51 nerzhul    https://dev.minetest.net/Android it seems very old and outdated :p
16:52 nerzhul    especially the sfan5 part
16:52 sfan5      yes you can throw that away
16:52 nerzhul    i will rewrite the second part :p
16:53 Krock      trivial change in march. the other stuff is two years old (or more)
16:53 rubenwardy shame you didn't get it done before sapier :'(
16:53 rubenwardy proper official cmake would be so much better than the crud that is currently used
16:54 nerzhul    rubenwardy, did you see my pr about it ?
16:54 nerzhul    it's a very very huge part
16:54 rubenwardy lol
16:54 nerzhul    not because of minetest, but deps
16:54 rubenwardy link?
16:54 nerzhul    #7123
16:54 ShadowBot  https://github.com/minetest/minetest/issues/7123 -- Android: switch to cmake based standard builds by nerzhul
16:54 nerzhul    it's painful, and i started to port some build parts to gradle directly but it's a very huge part :p
16:55 nerzhul    maybe i should port some parts outside of this PR to reduce a little bit the diff and the Makefile :p
16:56 nerzhul    i think i will port the gradle cleanup parts at a point, separately of this PR, this will reduce a little bit the makefile, and find another target after that
16:56 nerzhul    when everything is wrapped in gradle directly, refactor
17:00 srifqi     paramat: Done translating image.
17:02 paramat    ok
17:04 nerzhul    https://dev.minetest.net/Android#Releasing_Minetest_APK updated
17:08 paramat    yes no need for 0.4.17.2 for android, otherwise we would need to release desktop 0.4.17.2
17:08 nerzhul    sfan5, rubenward, Krock : okay for a production release & a merge ?
17:08 sfan5      yes
17:10 nerzhul    let's go then
17:12 nerzhul    merged + cherry pick done
17:16 Krock      s/merge/rebase & merge/
17:16 Krock      in case there were new commits (doubt it)
17:16 nerzhul    yeah, it was done like that :p
17:17 nerzhul    i hope nobody commits on stable 0.4 :p
17:22 nerzhul    apk published in production, thanks to testers
17:27 srifqi     Update the description maybe?
17:27 Krock      nerzhul, I'm not sure how marking the duplicates as deprecated is sexy but thanks for the review :3
17:31 nerzhul    srifqi already done :)
17:31 nerzhul    heh Krock
18:08 paramat    the description text at the play store needs updating, still has c55's notice of breakage. also, under 'what's new' it should say '0.4.17.20' not '.. .2' and make clear the recent breakage is fixed
18:08 paramat    however, thanks!
18:09 Krock      .2 because .1 was what we had in early June
18:11 red-001    0.4.17.20.....
18:12 red-001    should just switch to using the date at that point
18:13 red-001    paramat, isn't the 20 just the machine-readable version number?
18:15 sfan5      yes it is
18:18 red-001    nerzhul, is it possible to create a public test version?
18:19 red-001    could publish a release cadidate like that and ask the forum to test it
18:44 paramat    we decided to not bump to 0.4.17.2 because MT desktop has not
18:45 paramat    it's 0.4.17.1 app version 20
18:49 paramat    in the play store the '0.4.17.1'-type version number should refer to the current MT version which is 0.4.17.1. the android-specific version number is 0.4.17.20. probably a typo
18:50 Krock      can't we just leave the android build how it is right now? it works.
19:55 paramat    nerzhul or celeron55 see my last 5 messages :)
20:03 celeron55  edited some of the stuf
20:03 celeron55  +f
20:03 celeron55  (it updates slowly)
20:59 paramat    thanks
21:05 Krock      paramat, were you successful in applying the patch for smoother yaw changes? Will rebase if that doesn't help either
21:16 paramat    haven't tried yet, will try soon
21:16 paramat    -3 will probably work
21:18 paramat    closed #7439 as seems resolved
21:18 ShadowBot  https://github.com/minetest/minetest/issues/7439 -- Android 0.4.17.1: Google Play Store build crash while loading singleplayer world
21:37 Krock      Well so, didn't want to switch branches. deleted the old smooth_yaw branch, git am -3 the patch, force push. Works for me :D
21:39 paramat    hm i'm obviously not so good with git :}
21:39 paramat    oh i see, will retry
21:41 paramat    ok getting somewhere now thanks
21:42 Krock      good to hear :)
21:46 paramat    i always get this warning when compiling: "warning: ISO C forbids conversion of object pointer to function pointer type [-Wpedantic]" does anyone else?
21:46 paramat    and i mean. always, for years
21:52 Krock      feels a bit smoother with vehicles
21:54 paramat    will test more
21:55 Krock      boats are very smooth
21:59 paramat    testing, now it seems to work well
21:59 paramat    odd, when i compiled the branch that was 350 commit behind it was bad
22:00 paramat    anyway, will test more
22:00 paramat    my driftcar turns very smoothly now
22:01 red-001    paramat, in lua code?