Time  Nick            Message
07:07 fruitsnack      Hello, where entity origin should be since 5.0? Before mob implementations and player had entity origin at their legs, should it be in the middle now with collision boxes extending in both directions (for example if mob is two nodes tall collision box min and max height will be -1 and 1)
14:21 deltasquared    so just updated to 5.1.0 as shipped by my distro and... segfault on joining world :/
14:22 deltasquared    gonna check distro bug trackers first. (arch linux)
14:22 deltasquared    ... huh, nothing yet
14:23 deltasquared    > Invalid size of image for opengl texture
14:26 deltasquared    also a thread that was in something to do with font code at the time... (one sec prepping log)
14:27 Krock           dump to pastebin
14:30 deltasquared    Krock: doing so now
14:31 deltasquared    https://a.uguu.se/2v7BGupuysbg_log.trunc.txt (warning, embedded colour codes, having some difficulty getting journalctl to not emit them)
14:32 deltasquared    the two active threads I spot are getGlyphIndexByChar() and fdatasync() (apparently being driven by sqlite there)
14:33 Krock           no, that's just showing what the other threads do
14:33 Krock           the issue is probably that you didn't specify a valid font
14:33 Krock           or it's missing
14:34 deltasquared    no font changes were performed as a result of this upgrade to v5.1.0
14:34 deltasquared    furthermore this only occurs on world load
14:35 deltasquared    not at menu (I use the same font for everything)
14:35 Krock           indicates that there's an issue with the monospace font
14:36 deltasquared    shipped version of irrlicht by distro not changed. font not changed. it's still there, and is the same as I was using it on v5.0.1
14:36 deltasquared    I had minetest open before the upgrade, then re-opened it after the upgrade, and this started occurring
14:36 deltasquared    _nothing else has changed_
14:36 deltasquared    so what gives
14:37 Krock           does it also happen in master (5.2.0-dev)?
14:38 deltasquared    y'know you could at least try to phrase that bearing in mind it's slightly non-trivial to do so on a moment's notice, y'know :P but fine
14:39 Krock           also, what freetype version was used before/after? is Minetest built with freetype?
14:39 Krock           sorry?
14:41 Krock           error happens somewhere here. https://github.com/minetest/minetest/blob/master/src/irrlicht_changes/CGUITTFont.cpp#L405
14:43 deltasquared    checking now, sorry I was just kicking off a dev build.
14:43 deltasquared    I'm also checking my package logs again, one moment
14:43 deltasquared    freetype2 before and after is 2.10.1
14:44 deltasquared    I'm not sure how to check if freetype is enabled in the build seeing as I can't get at the in-game pause menu, but I'm 90% certain it is enabled
14:45 deltasquared    the pkgbuild says it is using freetype. https://git.archlinux.org/svntogit/community.git/tree/trunk/PKGBUILD?h=packages/minetest
14:45 Krock           if you have some spare time, please run Minetest in gdb. It usually provides nicer looking dumps
14:47 deltasquared    so whatever it is still occurs on the v5.2.0 hot off the presses
14:50 deltasquared    ... ok, how. so resetting my config fixed it. yet it was a valid config before the upgrade. I'm going to need to narrow down which line caused issues...
14:51 Krock           check the font paths first. the issue lies somewhere there
14:52 Krock           I also have custom fonts, even testing without TTF stuff which was horribly broken a while ago. Yet no such problem
14:54 deltasquared    Krock: the files it reference still exist
14:54 deltasquared    and I know in any case the monospace font wouldn't have been disturbed, it's one I generated myself from a pcf using a script, so it's not like it would be suddenly borked
14:55 deltasquared    gah. if only arch didn't make it such a pita to backpedal a version in a hurry
14:55 deltasquared    ... ah, it's hiding there in my cache. let's see what that does
14:55 deltasquared    (v5.0.1 that is. good thing I didn't purge it entirely)
14:56 deltasquared    ok, v5.0.1 definitely works
14:56 deltasquared    (with the crash-y config)
14:57 deltasquared    so what I'm going to do now is I'll start with a blank config file (thus letting MT populate with defaults) and add a line back one at a time from the "failing" one, to see which one breaks things
15:15 deltasquared    Krock: was there any code changed that affected how the effective font size is calculated?
15:15 deltasquared    I've noticed before gui_scaling also scaled the font_size variables
15:16 deltasquared    I've had problems before because this particular monospace font (it was that line after "bisecting" it) that was generated from a pcf has strictly one size to it
15:16 deltasquared    and that MT loading the wrong size would cause it to crash
15:17 deltasquared    (null pointer somewhere I guess?)
15:19 deltasquared    so basically if that effective value ever gets changed (either because of ui scaling change or something else)... KABOOM
15:20 deltasquared    less than graceful error handling but it seems there's not very much MT could do if that is the case again
15:21 deltasquared    except that now with taking all scaling settings out of minetest.conf I can no longer find a valid mono_font_size value :(
15:22 deltasquared    so yeah, anything that changed that affected font size lookup of any kind? (multiple sizes perhaps... in which case my use of a bitmap font is most definitely doomed anyway)
15:43 Krock           *reads*
15:45 Krock           gui_scaling -> size calculation https://github.com/minetest/minetest/blob/master/src/client/fontengine.cpp#L271-L272
15:46 Krock           the calculation there did not change. checking the prop.size orgins
15:47 Krock           deltasquared: what values did you use previously? gui_scaling and mono_font_size?
15:48 Krock           *what kind of values did
16:02 Krock           OH
16:03 Krock           deltasquared: I did some F6 profiler changes which now also use monospace fonts
16:04 Krock           thing is, it's not the full size but like * 0.85 to be not too disturbing
16:05 Krock           #8750
16:05 ShadowBot       https://github.com/minetest/minetest/issues/8750 -- Better F6 profiler by SmallJoker
16:05 Krock           https://github.com/minetest/minetest/blob/master/src/client/gameui.cpp#L83-L84
16:07 Krock           deltasquared: hence your font would also need support for "math.floor(font_size * 0.9)"
16:09 Krock           or if you're using the fallback font (due to missing characters for your language) "math.floor(fallback_font_size * 0.9)"
16:30 deltasquared    Krock: ok interesting. the different sizes in that case is slightly annoying
16:31 deltasquared    as for what values I used previously, one sec
16:32 deltasquared    ... ah, in this circumstance gui scale was 1.0 and mono_font_size = 12
16:32 deltasquared    so in other words due to the profiler I'd need a scalable font basically
16:32 deltasquared    shame I can't use one specifically for the chat and/or just force the font sizes individually.
16:33 MinetestBot     02[git] 04MuhdNurHidayat -> 03minetest/minetest_game: Add Malay translation 137106056 https://git.io/Je1HI (152019-11-30T16:31:40Z)
16:33 MinetestBot     02[git] 04Aresiel -> 03minetest/minetest_game: Add Swedish translation 1310dce79 https://git.io/Je1HL (152019-11-30T16:31:29Z)
16:33 deltasquared    (I confess the chat *when open*, i.e. mono font, is the main reason I use this problematic font)
16:34 deltasquared    /and/ this particular font even if it were pre-scaled inside the font kinda looks horrendous, as it's designed to be pixel sharp.
16:34 deltasquared    it's terminus in case you're wondering; I converted just the size I wanted to use (12px) to an otb font and plonked that in my /usr/share/fonts.
16:35 deltasquared    so in all fairness, it is a limitation in the font, in the same way that it can be seen as a web page's fault when it's not responsive to screen or font size scaling.
16:36 Krock           well, it's possible to create fonts with sharp edges that are scalable
16:36 Krock           although I must admit that I never designed fonts
16:36 deltasquared    Krock: idk how though from an existing pcf font. I don't even recall how I made /this/ one, it was a one-liner given to me in #archlinux IIRC
16:37 deltasquared    pixel fonts aren't really intended nor do they look good at non-integer scaling of their base size in any case
16:37 deltasquared    honestly my preference in this case would have been to keep the profiler using the same font size as I can live with that given I wanted to use this precise font size
16:38 deltasquared    (I'm not a fan of terminus's larger sizes even)
16:38 Krock           hmm. If you really want to us the very same font, please edit the scale multiplier from 0.9f to 1. This is somewhat a rarity and thus, an edge-case
16:38 Krock           sadly the profiler takes up a notable amount of space if run in singleplayer
16:38 deltasquared    this is true. guess I never was bothered that much by it
16:39 deltasquared    bearing in mind the other reason for that tiny font size is my current ongoing project with small hdmi screens, so I was /already/ dealing with various screen elements taking up a fair chunk of space.
16:39 Krock           maybe my screen is just a bit small, thus I also tried to keep the profiler readable but not too disturbing from the actual game. Just so you got a little insight in my thoughts
16:40 deltasquared    you'll be glad to know MT handles 800x480 just fine with a hud scaler applied (assuming my font selection was working) :)
16:40 Krock           heh :D
16:40 deltasquared    the idea was a switch-like handheld
16:40 deltasquared    the screen itself is a 7" newhaven module
16:40 deltasquared    (it even has controllable backlight! albeit via a PWM signal)
16:40 Krock           the latency of such screen can be quite horrible, though.
16:40 deltasquared    Krock: it has no latency at all
16:40 deltasquared    it has no scaler
16:41 * Krock         looks up
16:41 deltasquared    it's direct to scan-out from hdmi
16:41 deltasquared    newhaven display is what you're after
16:41 deltasquared    basically, it takes the hdmi signal from a fixed resolution only and throws that directly at the panel
16:42 deltasquared    it /only/ supports 800x480 and will display garbage if you try to do anything else (assuming your OS isn't smart enough to realise that from the EDID info)
16:42 Krock           miniature screen :D
16:42 deltasquared    and for some ineffable reason, it runs at ~66hz
16:42 Krock           66.6 Hz?
16:42 deltasquared    more like 65.89hz
16:42 deltasquared    hold on
16:42 deltasquared    oh, 65.40 my apologies
16:43 deltasquared    anyway, that produces a /lot/ of fun in apps that assume a nice 50, 59 or 60hz
16:43 Krock           the user guide shows 65.68 Hz but that's tolerances
16:43 deltasquared    what model you looking at?
16:43 deltasquared    just so I know it's the same
16:43 Krock           http://www.newhavendisplay.com/userguides/NHD-7.0-HDMI-N_User_Guide.pdf
16:43 deltasquared    ah, that's the one
16:43 deltasquared    non-touch model here so I can get those pixels in all their matte glory
16:44 deltasquared    I have an even smaller one as well, that one was some random rpi screen off amazon.
16:44 Krock           apps should expect any frame rate since 120 Hz and 144 Hz displays became a thing
16:44 deltasquared    Krock: tell that to crash n.sane trilogy
16:44 deltasquared    I have the smaller mounted on a frame that allows me to attach it to a harness around a steam controller.
16:45 deltasquared    essentially, I could play minetest on a handheld :D
16:45 deltasquared    albeit tethered atm
16:45 deltasquared    I have a compute stick I intend to fix that
16:46 deltasquared    so yeah, I was pleasantly surprised minetest held up on such an esoteric setup at least
16:46 deltasquared    given that other games had... issues
16:46 deltasquared    mostly tearing
16:46 deltasquared    apparently even modern games don't get what vsync actually means
16:46 deltasquared    *some modern games
16:46 nephele         oh hi deltasquared
16:47 Krock           PS: If you want to gain some frames per second but are willing to sacrifice colored text and text shadows: check out bitmap fonts
16:47 Krock           those are also pixel-sharp but ignore the font scaling setting
16:47 deltasquared    Krock: well MT stopped supporting pcf fonts before which lead me to using the otb translation
16:47 deltasquared    it just wouldn't launch after an update one day
16:47 deltasquared    oh wait, I think that was a freetype thing? I remember reading about it /somewhere/...
16:48 Krock           this was probably before I started developing. I never saw pcf fonts in use before
16:48 deltasquared    nephele: apologies, can you remind me where we have met before again
16:48 deltasquared    Krock: so what bitmap formats does MT do currently?
16:48 deltasquared    I can look into performing a separate conversion
16:48 Krock           no no, bitmap fonts are separate from freetype. they're basically used when you compile Minetest without freetype
16:48 deltasquared    right
16:48 deltasquared    so again what formats
16:49 Krock           https://github.com/kahrl/irrtum
16:49 deltasquared    does this solve that issue I've noticed where large amounts of chat takes an FPS hit by any chance?
16:49 nephele         deltasquared: the vanessae server chatroom thingy
16:49 Krock           well, it's a single image which contains the supported characters that are then rendered 1:1 to the screen
16:49 Krock           yeah
16:50 nephele         i used 'nepugia' there though
16:50 deltasquared    oooooooooh hi
16:50 deltasquared    Krock: so basically an old school UV glyph atlas? I approve :D
16:51 Krock           sorry, I don't know the format. Only thing I know is that there's a conversion tool, lack of colored text but faster drawing
16:51 Krock           looking for the comparison....
16:51 deltasquared    /home/user/dot/Downloads/irrtum/graybitmap.cpp:72:11: warning: control reaches end of non-void function
16:51 deltasquared    tut tut :P
16:52 Krock           https://github.com/minetest/minetest/pull/8641#issuecomment-507056996
16:52 nephele         so sad that valve doesnt make steam controllers anymore
16:53 Krock           so yeah.. on weak hardware you can really gain some frames by using bitmap fonts
16:53 deltasquared    nephele: get em while they're on fire sale I guess
16:53 nephele         They aren't
16:53 deltasquared    gone already? I heard they were being cleared out for $5 each
16:53 nephele         only found out that they were on sale after they were sold out
16:53 deltasquared    dang
16:53 nephele         and that was two days ago :)
16:53 deltasquared    /to be fair/, you're not missing too much
16:54 deltasquared    the touchpad is definitely useful, but it's not fantastic, just good
16:54 nephele         are you kidding? best controller i own
16:54 nephele         i hate analog sticks, the digital trackpads are soooo much better
16:54 deltasquared    I actually want to make my own SC-like controller at some point
16:54 deltasquared    I found where to buy that tiny trackpad
16:54 deltasquared    I'd just need to learn how to talk to it
16:54 Krock           Minetest compatibility?
16:54 deltasquared    Krock: of the steam controller?
16:54 nephele         deltasquared: sounds nice, want to do something similar
16:55 deltasquared    I wrote my own mapper
16:55 nephele         the board the SC uses is known too
16:55 nephele         i kind of want to remove the analog stick though
16:55 deltasquared    Krock: I wrote my own bindings in a program I wrote using lua and lua-evdev.
16:55 Krock           deltasquared: yeah, so you need a conversion script in order to use it like keyboard?
16:55 nephele         Krock: either you need a driver, or you need the steam software
16:55 Krock           ah
16:55 deltasquared    I picked the former
16:56 deltasquared    screw being tied to proprietaREEEEEEEEEEe software
16:56 nephele         The format is kind of known by now
16:56 deltasquared    nephele: it's supported by evdev
16:56 nephele         want to write a driver for haiku at some point :3
16:56 nephele         deltasquared: uhh? what?
16:56 nephele         it's "just" usb_hid
16:56 nephele         so any support would be interesting in userspace
16:56 deltasquared    modern linux kernels expose the SC like any other game controller, so no need to muck around with hid raw stuff
16:57 nephele         linux has a steam controller driver, yes
16:57 deltasquared    hid_steam
16:57 nephele         on haiku i can make a steam-like per-window binding, and without steam owo
16:57 deltasquared    anyway, I wrote a "driver" using lua-evdev. https://github.com/thetaepsilon/sc_lunar_mapper
16:58 deltasquared    per-window binding, man I would kill for that
16:58 deltasquared    without needing hacks like xorg snooping
16:58 deltasquared    I have to change bindings manually atm
16:58 nephele         It's a lot easier on haiku
16:58 nephele         applications have an application signature
16:58 nephele         so i can map to that
16:58 deltasquared    I wish it was possible to propogate up the stack such a signature on linux
16:58 deltasquared    there's no built-in concept of "what's the user looking at right now"
16:58 nephele         It's possible
16:59 nephele         Just write a display server that has that as a requirement to talk to it ;)
16:59 deltasquared    yeah no thanks, I have limited free time as it is :P
16:59 nephele         haha
16:59 deltasquared    Krock: my SC bindings for minetest look like the following
16:59 nephele         linux isn't a desktop OS anyhow
16:59 deltasquared    right pad -> mouse (duh)
16:59 deltasquared    LT and RT are attack and place
16:59 deltasquared    when they bottom out that is
17:00 nephele         deltasquared: does your thing do motion sensor -> mouse too?
17:00 deltasquared    nephele: not atm
17:00 nephele         That is my favorite layout usually for the controller
17:00 deltasquared    it's a separate evdev device see
17:00 deltasquared    also I think it may be turned off by default until you tickle it with some HID commands, not sure, was reading about that somewhere
17:00 deltasquared    anyway, item switch was two of the gamepad buttons
17:00 deltasquared    motion was the thumbstick
17:00 deltasquared    screenshot was the steam button.
17:01 deltasquared    ESC was the left "select" arrow
17:01 deltasquared    can't remember what I used the right "start" arrow for
17:01 deltasquared    one of the shoulder buttons was enable full render, other was fast move
17:01 deltasquared    left paddle sneak, right paddle jump
17:01 deltasquared    zoom was pressing the touchpad in (it clicks)
17:02 deltasquared    IIRC I found the left "dpad" touchpad annoying so I didn't bind it.
17:02 nephele         binding the digital pad to movement speed and direction would be fun
17:02 nephele         but that requires a way for minetest to accept such movements
17:02 nephele         instead of the generic wasd 100% in direction movement
17:02 deltasquared    ah, the "start" arrow was open inventory.
17:02 nephele         deltasquared: to use or to bind?
17:03 deltasquared    nephele: to use
17:03 deltasquared    especially when I was resting my thumb. didn't figure out a way to prevent false trigger.
17:03 deltasquared    and I didn't like the way it clicked
17:03 nephele         clicked?
17:03 nephele         why are you clicking it?
17:03 deltasquared    when you pressed the dpad it can click in four places
17:04 nephele         there is no dpad on the steam controller
17:04 deltasquared    nephele: I say "dpad" in quotes
17:04 deltasquared    I mean the left touchpad that has arrows on it
17:04 nephele         it has some markings, but they are irrelevant
17:04 nephele         it is a perfectly working touchpad
17:04 nephele         and i love to use it for movement controlls, used that for many many games :)
17:04 nephele         but with the steam software
17:05 deltasquared    it also has four "click" zones. I've seen the raw event stream, I know what it can do
17:05 nephele         uhh, it doesn't seem to have that in hw?
17:05 deltasquared    as in, you can press it in at up, down, left, right, and they are dedicated button events
17:05 deltasquared    mine does... I swear it does. hold on
17:05 deltasquared    gotta get it out :)
17:05 nephele         i can click anywhere i want
17:06 nephele         on steam it is just a "clicked" event usually
17:06 nephele         and i don't see a reason to use it different to that, seeing as i just use touch-> movement normally
17:08 deltasquared    nephele: it has dedicated hardware events for pressing in the four dpad directions. https://a.uguu.se/jVqLzNMrBZBn_dpad.png
17:08 deltasquared    I filtered out the touch events there so you just see buttons
17:08 deltasquared    notice the btn_thumb event at the start and end as I initially press then release
17:08 nephele         deltasquared: how do you know?
17:09 deltasquared    nephele: that's a live dump from the device
17:09 deltasquared    with no software intercepting it
17:09 nephele         no, that's evdev
17:09 deltasquared    yeah and?
17:09 nephele         as in linuxes kernel driver, no?
17:09 nephele         not the hid events
17:09 deltasquared    huh, never thought the kernel would be sugar coating it xD
17:09 deltasquared    idk then
17:09 Andrey01        hello, why does get_properties() method output entity def without custom properties?
17:09 deltasquared    nephele: could always just ignore the dpad emulation if that is the kernel I guess
17:10 nephele         deltasquared: what would the steam controller driver do other than that?
17:10 nephele         isn't it it's sole purpose to make it a bit easier to use?
17:10 deltasquared    nephele: couldn't say
17:10 Andrey01        I`ve added my own props and they`re not in that table
17:11 deltasquared    nephele: it seems strange though that the kernel would not expose a generic "touchpad pushed in" event in that case though :/
17:11 nephele         since there is no specs i can't tell you specifically
17:11 nephele         but that it would be a dpad seems strange
17:11 nephele         since i can push it on all positions
17:11 nephele         what would it be if i push the top right?
17:12 nephele         if it's just generic events it would just be the hid stuff no?
17:13 nephele         Krock: does any movement api exists that minetest would accept for input?
17:13 nephele         to specify how much speed the player should move in
17:13 deltasquared    that would be great, I could have true analog then
17:13 deltasquared    I have to coerce to keypresses from the analog stick currently
17:13 Krock           physics overriders
17:13 deltasquared    Krock: I think nephele meant in the client
17:14 deltasquared    to send input events to the client
17:14 Krock           key settings
17:14 deltasquared    more directly than key presses
17:14 deltasquared    Imma guess the answer is no
17:14 nephele         Krock: an api to input stuff, instead of sending "wasd"
17:14 nephele         To more directly control the player
17:14 nephele         like dinput or directx input do with the thumbsticks on windows
17:14 deltasquared    if MT doesn't support analog input though there's not a real distinction from sending buttons is there
17:15 Krock           not that I'd know of
17:15 nephele         deltasquared: should just specify an api and try to push that :P
17:15 deltasquared    nephele: that's annoying. you can't actually press the virtual dpad up and right at the same time. that sucks :/
17:15 deltasquared    just tried it
17:15 nephele         deltasquared: by the way, i want in on your like-steam controller controlled
17:15 deltasquared    nephele: ???
17:15 nephele         deltasquared: huh? but that works fine with steam
17:16 nephele         deltasquared: you said you wanted build a similar controller
17:16 nephele         :)
17:16 deltasquared    oh, do you want the part link for the touchpad at least?
17:16 deltasquared    I haven't really invested any time into it yet
17:16 nephele         That's why i said i want in :P
17:16 deltasquared    I was figuring out how to interface it last time I thought about it
17:16 nephele         We could use the same board the steam controller uses
17:17 deltasquared    Krock: btw, I got this from that tool... https://a.uguu.se/5EOwXQbavnQd_Terminus-Medium.otb.png
17:17 nephele         Add to that a 3d mould we specify to print and it should be produceable by consumers :)
17:17 deltasquared    Krock: is that correct? they don't look much like characters
17:17 Andrey01        and also minetest.registered_entities doesn`t save user properties, I dont understand why?
17:18 hecks           what the hell: ERROR[Server]: Server::ActiveObjectMgr::addActiveObjectRaw(): no free id available
17:18 nephele         deltasquared: do you have another comms channel other than minetest? to plot a controller :)
17:18 nephele         other than irc i ment...
17:18 Krock           deltasquared: that totally went wrong. Maybe it only works for scalable TTF
17:19 deltasquared    Krock: guess I'd have to try and convert it some other way (from the origin pcf instead of the otb I made)
17:19 Krock           hecks: congrats for filling up the pool with 2^32 objects
17:19 hecks           wow, I have like 3 objects actually
17:19 deltasquared    how the hell is that server not on fire
17:19 hecks           but I guess something is trying to recreate thme in a loop
17:19 Krock           deltasquared: it will look like this: https://krock-works.uk.to/u/DejaVuSansMono_15pt.png
17:19 deltasquared    nephele: one that I pay attention to on a reliable basis? none currently
17:19 deltasquared    suggestions are welcome... (PM?)
17:20 nephele         not like email or something?
17:20 deltasquared    erm
17:20 deltasquared    my email... betrays my real name xD
17:20 nephele         my git has my real name in plain view :P
17:20 deltasquared    yet it's the only one I can access from anywhere
17:20 deltasquared    on account of the diceware password being burned into memory by now
17:20 deltasquared    for reference, the touchpads that apparently the SC uses - https://eu.mouser.com/new/cirque/glidepoint-circle-trackpads/
17:21 deltasquared    I should make a video sometime of me playing minetest on my "gameboy"
17:21 deltasquared    (the screen on it's harness around the controller)
17:22 Krock           hecks: actually it's that 1) there are only 2^16-1 active objects possible (at this moment) and 2) overflows in the ID assignment are handled wrong
17:22 hecks           I'm doing my own persistence using entities without static_save, so, it's probably trying to "load" an entity infinite times
17:22 Krock           ignore 2
17:23 hecks           while I'm fiddling with active block management
17:23 Krock           in 1 of 2^16 entities you'll get this error once due to integer overflow
17:23 Krock           minetest bug
17:24 hecks           can't be the case with fresh worlds
17:24 Krock           actually no. the code also handles ID = 0 correctly as reserved
17:24 Krock           for now it appears to be a mod issue
17:25 hecks           it's my problem, it was just... puzzling
17:25 hecks           you'd think creating 1<<32 entities would take a measurable amount of time, guess not
17:29 Andrey01        does anybody know where custom properties of entity def table are being saved? They are really neither in minetest.registered_entities nor in self
17:29 hecks           getmetatable(self).__index ?
17:30 hecks           docs said something about metatables
17:30 rubenwardy      Andrey01: code please
17:32 hecks           oh now I see the part where it creates 0xFFFFFFFF entities, it does actually take a few seconds, you can see it by mapblocks not loading anymore :)
17:39 Andrey01        rubenwardy: https://pastebin.com/xSbSQ1da
17:45 Krock           * 2^16 objects. I don't know why I thought of 32 bits
17:51 Andrey01        rubenwardy, have you not reviewed my code yet?
17:51 rubenwardy      I'm busy with other things
17:58 hecks           minetest.forceload_block wants a "position" but what kind of position is this? in nodes? in blocks?
17:59 rubenwardy      in nodes
17:59 hecks           thank you
17:59 rubenwardy      the API tries not to expose mapblocks to the Lua API
18:00 sfan5           https://github.com/minetest/minetest/blob/master/src/script/lua_api/l_env.cpp#L1278
18:00 rubenwardy      really forceload_Block should be forceload_area, and take minp/maxp
18:00 sfan5           are you sure?
18:00 hecks           uh oh
18:00 rubenwardy      bad documentation then
18:02 rubenwardy      sfan5: the position is converted to blocks in Lua:   https://github.com/minetest/minetest/pull/1094/
18:02 rubenwardy      hecks: it's in nodes
18:02 sfan5           ah
18:02 sfan5           that's unexpected
18:02 hecks           alright
18:02 rubenwardy      I think that's bad practice
18:02 hecks           it's acceptable if there's no official "convert to block coords" function
18:03 hecks           also if this used minp/maxp, how would you deallocate forced blocks?
18:03 rubenwardy      it's bad practice for the Lua API to hijack the C++ code like this
18:03 hecks           you mean forceloading in general?
18:03 rubenwardy      no, that PR I linked
18:03 hecks           oh
18:03 rubenwardy      there should be a comment in the C++ at least
18:04 hecks           anyway I do need forceloading badly to ensure MT doesn't eat entities when I don't want it to, and the way it works now with individual blocks is ideal
18:10 Andrey01        so again question: does minetest.registered_entities save inside custom properties or not? And then where can they be saved?
18:11 hecks           Andrey01: I don't know the answer to this, but, if nothing else works you can wrap the registration in something that stashes the extra data somewhere
18:16 rubenwardy      Andrey01: custom properties will be saved in registered_entities and then accessable through self due to metatables
18:16 rubenwardy      oh
18:16 rubenwardy      you can't define custom object properties
18:16 rubenwardy      that's an engine API
18:16 rubenwardy      you can only store things in seld
18:16 rubenwardy      *self
18:16 rubenwardy      do:   self.head
18:17 rubenwardy      an object property in an engine API to set various different things like view appearance, it's not to store arbitrary data
18:18 Andrey01        ah, thanks
18:18 Krock           fun fact: linked lists are possible in Lua, even though it might be quite pointless
18:19 hecks           heresy
18:19 rubenwardy      Andrey01: be aware that head won't be copied for each new entity, as it's a metatable
18:19 Krock           and pointer-less (insert knee slapping here)
18:19 rubenwardy      you'll need to do something like       rawset(self, "head", table.copy(self.head))    if it's a mutable table  iirc
18:20 rubenwardy      *table.deepcopy()
18:20 rubenwardy      it's a string though, so is fine
18:20 rubenwardy      strings are immutable
18:20 Krock           Minetest's function is called table.copy() which does a deep copy
18:20 rubenwardy      ah
18:20 Krock           FYI
18:20 rubenwardy      I had doubts
18:21 Krock           I had grep "table." lua_api.txt
18:21 Krock           actually "table\."
18:21 hecks           deep copies considered harmful
18:21 rubenwardy      then again, I always end up typing "minetest.file_exists()" and then having to search the lua_api to find that it's just "file_exists"
18:22 Krock           file_exists does not exist in lua_api
18:23 Krock           minetest.get_dir_list is documented, though
18:23 rubenwardy      oh yeah, it's undocumented: https://github.com/minetest/minetest/blob/master/builtin/common/misc_helpers.lua#L208
18:24 Krock           lovely
18:24 rubenwardy      probably shouldn't be using that, then
18:25 Krock           this really should be a minetest namespace/table function
18:25 Krock           and get_last_folder cleanup_path should not even exist in the global env
18:38 hecks           yee https://a.uguu.se/rvqY81HYe8bS_blkloader.mp4
18:41 hecks           now for the diamond question, how do I make UI work on different DPIs
18:42 hecks           hud and item icons in particular
18:44 hecks           at default dpi, item graphics must be 16x16 or they'll scale horribly... at 96 dpi, they must be 24x24
18:45 hecks           I might have to screech at people to use 96 to play this which is a less than ideal thing to do
23:05 Copenhagen_Bram hello
23:19 hisforever      Hi is there a way to hide the tool bar while taking a screen shot?
23:24 nephele         one of the f(n) keys hide the hud, i forgot which one though
23:24 Calinou         hisforever, nephele: press F1 to toggle the HUD
23:24 Calinou         (and F2 to toggle the chat, if any)
23:26 hisforever      thanks my friends I like sharing screenshots you are great in here