Time  Nick          Message
00:33 hmmmm         i was going to but i got caught up in the main menu crap
00:33 hmmmm         if you do start on it, please do it in C instead of C++
01:09 hmmmm         ehhgh
01:10 hmmmm         how do you tell if someone changed the selection on a dropdown box for the main menu?
01:17 kahrl         you can with textlists (and tables) but I'm pretty sure not with dropdowns
01:19 hmmmm         that's pretty arbitrary.  i have no idea what's going on with formspec, what would I have to do to fix that?
01:21 kahrl         catch EGET_COMBO_BOX_CHANGED in GUIFormSpecMenu::OnEvent and generate an event for lua
01:21 hmmmm         thanks
01:29 hmmmm         adding an event is probably going to break lots of things
01:30 hmmmm         at least sapier's main menu, anyway.  there's a lot of code that's like:  "if <some element that generates events but isn't an OK button> then handle thing, tabbuilder.ignore_redraw = true; return; end"
01:30 hmmmm         so basically everything needs to be restructured due to some dopey design
01:31 hmmmm         "yes, let's treat ALL events as close the formspec unless something else is explicitly specified!"
01:31 kahrl         yeah it's great ;)
01:35 VanessaE      I believe RBA complained rather vocally about that some time back
01:36 VanessaE      when he was working on Unified Inventory's search function - we wanted the enter key to trigger the actual search button.  No good, enter = exit.
01:41 hmmmm         VanessaE, that was something different though
01:41 hmmmm         I think I recall that saga... the problem was that the client would need to predict what the server wants to do when input has been accepted
01:42 hmmmm         all of this bullshit will be fixed, come client-side modding
01:42 hmmmm         no more guesswork will be involved
01:45 VanessaE      but...but..
01:45 VanessaE      what would minetest be without at least some hassle and guesswork? :)
01:45 hmmmm         good
01:45 khonkhortisan not developed by a random group of *guesses*
01:46 khonkhortisan lunatics. looked it up
01:47 hmmmm         so uh what's with the first 40 lines of GUIFormSpecMenu::OnEvent
01:48 hmmmm         I don't get it.. why are there two adjacent break statements, why is there a default case with a comment that says "can't happen at all!" and then an assertion for it, ...
01:48 hmmmm         strikes me as very sloppy
01:51 kahrl         looks like template coding
01:51 kahrl         some button in an IDE that generates switch () { break; default: break; }
01:53 hmmmm         what's the rationale behind the checks for if the ID is > 257?
01:54 kahrl         IDs 0...257 are reserved for elements that don't correspond to a field
01:54 hmmmm         ah ok
05:14 celeron55     PilzAdam: can you write down the way how you can create and generate "pilztest" trees?
07:34 jin_xi        pilztest uses decorations and schematics for trees
15:55 BlockMen      any comments on #886, #928 or #930?
16:00 kahrl_        BlockMen: #886 seems fine
16:02 proller       897 895 892 882 wants too
16:03 kahrl         #930 could be made more generic, I believe
16:04 kahrl         as in allow images to take half the screen, for example
16:06 kahrl         #928 has buffer overflows
16:07 kahrl         you read a color using split() but then you don't check that there are actually 4 parts
16:08 BlockMen      kahrl, so like -0.5 == half screen ? (#930)
16:09 BlockMen      and you mean at #928 like here? -> https://github.com/minetest/minetest/pull/928/files#L1R1378 so checking if color.size() == 4 ?
16:11 kahrl         #930: either that or adding a bool scale_from_screen (or whatever) seems fine
16:11 kahrl         #928: exactly
16:13 BlockMen      ok, gonna change that. thanks.
16:22 kahrl         proller: #895 ummm... why do you set liquid_loop_max even lower
16:22 kahrl         it should be restored to 10000 if anything
16:22 proller       because transform calls more frequent
16:22 kahrl         also the default values should be tuned for normal liquids not finite liquids (because normal liquids are the default)
16:23 proller       its ok for normal too
16:24 proller       if queue big - transform may work 10 times per second = 5000 nodes per second
16:24 proller       its better than one looong call
16:26 proller       processing 10000 nodes cause stuck of server
16:26 kahrl         okay
16:32 kahrl         proller: I'd just get rid of the whole #if 0 block here: https://github.com/proller/minetest/commit/283799333cfacc981c45222f1f373fc4704655d5#L4R1217
16:33 BlockMen      kahrl, done
16:33 kahrl         m_modified_blocks should probably be renamed to something with "liquid" in the name
16:33 proller       it can be modifyen not only by liquids
16:34 proller       its just list for sending
16:34 kahrl         but the interval they are sent in is called liquid_send
16:35 proller       maybe better to rename it to..
16:37 proller       block_send_every ?
16:38 proller       block_send_minimal ?
16:39 kahrl         well currently it's only used by liquids so a generic name would be confusing imo
16:39 kahrl         abms don't use it for example
16:39 proller       they need to use it
16:40 BlockMen      proller, does #880 fix the liquid range for infinite liquids too?
16:40 proller       yes
16:40 BlockMen      ok :)
16:40 proller       and removes range
16:41 kahrl         880?
16:41 BlockMen      ups..#882
16:41 proller       https://github.com/minetest/minetest/pull/882
16:47 proller       880 fixes most issues with finite
16:48 proller       and some old
16:50 BlockMen      880 fixed the umlauts issue :P -> https://github.com/minetest/minetest/pull/880
16:51 proller       ups, 882
17:02 proller       modified block send is strange
17:02 proller       is here range lomit to player for sending?
17:02 proller       limit
17:20 hmmmm         you should change m_liquid_transform_every to m_liquid_transform_interval
17:22 proller       done
17:23 hmmmm         people might not "get" the former right away unless they say it out loud
17:24 BlockMen      hmmmm, any comments on #886, #928 or #930?
17:51 hmmmm         I wrote a comment for #930
17:51 hmmmm         as for #928, i have no idea what that does but it looks okay
17:52 hmmmm         #886, i guess..... would anybody find this useful?
17:53 hmmmm         for 886, lines 340-343 could use rangelim() instead
17:54 hmmmm         I would prefer if you didn't access the settings every single draw call
17:59 celeron55     the line where doing a binary tree string lookup matters really is beyond doing it a couple of times per frame
17:59 celeron55     if it's done in a loop that is done in a frame, that matters
18:01 celeron55     it hinders usability if new settings aren't taken into use automatically at runtime
18:02 hmmmm         some original HUD code was written like that and I optimized it by moving the calls into the ctor and saved quite a bit of time
18:02 hmmmm         that is NOT just a binary tree lookup
18:02 hmmmm         you forgot about the lock, all the string comparisons, then the conversion
18:03 hmmmm         settings is *SLOW*
18:04 BlockMen      hmmmm, #930 is a nice suggest. i will change to that
18:04 hmmmm         if code is being added/modified in some subsystem that i'm involved with, i'd like people to avoid unnecessary calls.  that's the key word here, unnecessary.  it's not making things more inconvenient or putting some burden on development, it's just a matter of being prudent
18:06 BlockMen      #886 hud.resize is called in every loop of game.cpp AFAIK. So moving the reading of the setting out of the function should do it, right?
18:31 hmmmm         yes, make it a member variable and set its value in the hud constructor like the other settings
18:40 BlockMen      ok. #930 is changed already
18:45 celeron55     and now you can't handily resize the hotbar in a singleplayer game by using /set?
18:50 BlockMen      celeron55, do you think that is needed?
18:51 celeron55     well, i probably won't use that hotbar resizing functionality at all
18:52 celeron55     but things should generally be made in some way that doesn't unnecessarily hinder functionality
18:55 BlockMen      in general you are right. but imo things like hotbar size, shaders(on/off), etc can be changed by gui (settings tab)
19:19 PilzAdam      BlockMen, especially HUD things need to be adjustable at runtime
19:19 PilzAdam      you dont want to restart Minetest over and over again to find the right setting
19:21 BlockMen      PilzAdam, well THEN we should add an ingame setting screen
19:21 celeron55     and what does that ingame setting screen do other than, like, set settings?
19:22 celeron55     well, that'd be some kind of a reloadable settings struct then i guess
19:22 celeron55     and that gets us to the core of C and C++: writing everything at least 5 times
19:23 celeron55     good thing i'm not doing that but hmm is
19:23 celeron55     hmmmm*
19:25 BlockMen      celeron55, nothing else. but that would be the right way to let user change settings, not /set ...
19:33 BlockMen      PilzAdam, any comment on #930?
19:51 hmmmm         celeron55, if you're going to add *that* which requires changing code anyway, you might as well just add a setHudSize() or whatever
19:51 hmmmm         not hard..
19:56 celeron55     yeah, nice spaghetti for each setting 8)
19:56 celeron55     spaghetti is always better than a bit slow code
20:16 Jordach       it'll go better with some new sauce
20:32 hmmmm         you have literally no basis for saying that it's spaghetti until you actually see the end result
20:38 proller       897 895 892 882 !
20:40 kaeza         42
20:41 hmmmm         no idea what the wingsuit thing is about
20:41 hmmmm         i've heard some people who don't like it
20:42 hmmmm         i suppose 895 is good, i'm not entirely sure of the ramifications though
20:42 hmmmm         you tested it i trust, and it doesn't break anything?
20:43 proller       you can feel winsuit only when jumping from 100+
20:43 sapier        tztz complaining about mainmenu event handling when there haven't even been events when mainmenu was created ;-)
20:43 hmmmm         sapier
20:43 hmmmm         hey man i've got some questions for you
20:43 proller       all is 1 month testing on sky server
20:43 sapier        ok I'm here ask ;-)
20:43 hmmmm         i'm trying to add an event for changing a dropdown item but i'm not having much luck
20:44 sapier        what's wrong ?
20:44 hmmmm         i have it all wired up and it's executing gotText with the CHG:itemindexhere
20:44 hmmmm         and I'm not seeing anything from the Lua side
20:44 hmmmm         here's my code
20:45 hmmmm         hold on
20:46 hmmmm         https://github.com/kwolekr/minetest/commit/2738e2cd53e07a1e04e1555a6580576ebd039b3e
20:46 hmmmm         I have no idea what's going on btw
20:47 proller       commit 895? next step - make one send block queue with optimizations
20:47 sapier        ok your problem is at c++ side did I understanc correct?
20:48 hmmmm         sapier, I think so, yeah
20:48 sapier        and the problem is no event is generated?
20:48 hmmmm         yup
20:48 sapier        why did you comment s.send?
20:49 hmmmm         s.send is already set in parseDropDown
20:49 sapier        ahh yes I remember dropdown is interpreted as text input field not as button
20:49 hmmmm         oh great... what does this mean?
20:50 hmmmm         where's it set? how do I change that?
20:50 kahrl         hmmmm: have you tried uncommenting the first line of engine.button_handler?
20:50 sapier        for buttons you can decide if a button is pressed by checking if it's present in the fields table
20:50 sapier        if it's not there it's not been pressed
20:50 hmmmm         i haven't, kahrl
20:50 sapier        text fields are always sent
20:52 sapier        so if you change it that way you need to ad some other mechanism to decide if it's an event issued by dropdown change or a button press ... but thats a different issue
20:52 hmmmm         kahrl, the problem is that button_handler isn't getting called though
20:53 hmmmm         sapier, if the element is a dropdown then it's obviously not a button press
20:53 hmmmm         so i would need to reorganize the code that assumes a dropdown won't create a button press event
20:53 sapier        if you want to send an event it is
20:53 hmmmm         but i already understand that
20:53 sapier        everything triggering a send event is a "button press" ;-)
20:54 sapier        just because there wasn't another mechanism except button press but yes we could rename it to "event" too
20:54 hmmmm         right.... oh wait a minute
20:55 hmmmm         gotText() taking the map calls handleMainMenuButtons
20:55 sapier        but there's nothing obvious wrong I assume you already tried to uncomment first line in button handler to see if it's called?
20:55 hmmmm         which is always called if m_text_dst (which it obviously is, since i'm getting that printf)
20:55 hmmmm         not the first commented line, a different one
20:55 hmmmm         hold on a minute
20:56 hmmmm         see, i don't understand why I wouldn't be getting it because there's no break-out-of-function return in button_handler
20:56 sapier        you added "CHG" in front of the string so you renamed the field to "CHG:<yourdropdownname>"
20:56 hmmmm         unless i am totally missing something
20:57 hmmmm         oh...
20:57 hmmmm         yup, I'm getting button_handler it turns out
20:57 hmmmm         alright, thanks..
20:57 hmmmm         i feel pretty stupid now
20:57 sapier        but you're getting this EVERYTIME  >>ANY<< button is called!
20:58 hmmmm         alright
20:58 hmmmm         i understand what i need to d onow, explode_textlist_event(fields["dd_createworld_preset"]) and then all that nonsense
20:59 sapier        actualy the explode is done on the value NOT the name
21:00 kahrl         I recommend adding 1 to the result of getSelected for consistency with textlist
21:00 hmmmm         yeah, gonna do that
21:01 hmmmm         what i have there is clearly sloppy and unfinished
21:01 sapier        getSelected is only usefull in mainmen but yes it'd be usefull
21:01 sapier        you have to find a way to do this in game to ... atm dropdown is designed as a special textedit field fhile textlist is more like a buttonlist
21:01 hmmmm         sapier, this whole thing is only useful for the main menu.. mods with dropdown boxes are going to start closing when they select an item
21:02 hmmmm         if i understand the system correctly
21:02 sapier        what do you mean with "start closing"?
21:02 hmmmm         it's going to accept input as if it were a button press
21:02 hmmmm         and formspecs exit when input is accepted
21:02 sapier        yes but a button press is not closing a formspec
21:02 sapier        no
21:02 hmmmm         oh sorry
21:02 hmmmm         it's the enter key that does that
21:02 sapier        it's just sending a fields event
21:03 sapier        enter or escape yes
21:03 hmmmm         anyway
21:03 hmmmm         it seems my actual problem was the if-elseif chain
21:03 hmmmm         I didn't realize that *all* of the fields are sent
21:03 hmmmm         thanks though
21:04 sapier        there aren't all fields sent
21:04 sapier        button and similar fields aren't sent
21:04 kahrl         this should be documented in lua_api.txt
21:04 sapier        that's formspec behaviour as long as it's there ;-) but yes it could be made more clear
21:12 hmmmm         btw i didn't look at the documentation at all until now and i agree it's crap
21:30 proller       all wingsuit sense is here https://github.com/minetest/minetest/pull/897/files#L2R564
21:31 proller       if many peoples dislike it - it can be simply removed
21:32 proller       rest of pull - falling optimizing and bugfixing
21:34 proller       to start disliking it you must build something 150+ nodes high and jump
21:39 kahrl         I explained what I disliked and it wasn't related to jumping off a pole
21:40 proller       it was something with fast move, but what exactly?
21:41 kahrl         it makes it slower and disallows changing direction midair
21:42 kahrl         (this is fast_move without free_move)
21:44 proller       ok. its bug, fixing..
22:10 proller       fixed
22:18 proller       now it disabled when e pressed
22:23 kahrl         ah that's much better
22:23 kahrl         still kicks in if you fall very far with fast_move but that's probably intended
22:25 kahrl         environment.cpp: In member function ‘virtual void ClientEnvironment::step(float)’: 2125:9: warning: unused variable ‘speedl’ [-Wunused-variable]
22:32 proller       speedl fixed
22:33 proller       what if falling very far?
22:34 proller       i tried - pressing e allow you to move at +20nps, but it much lower than fall speed
22:35 kahrl         well it's the same as wingsuit without fast_move
22:36 kahrl         but it doesn't come up during my normal usage of fast_move so I'm okay with it
22:41 proller       its because now air have viscosity
22:42 proller       and not very big speed loss
22:42 proller       commit whole?
22:44 kahrl         let's see what the others think
22:48 celeron55     i continue to fail to see the point
22:51 proller       why?
22:59 proller       try to jump with it and without
23:03 celeron55     isn't one of the points of falling that you fall and most probably die if you haven't planned on what you are standing?
23:06 proller       you have small chance to hit water
23:06 proller       most likely you die
23:07 proller       but it will be fun
23:08 proller       but somebody can make parachute mod
23:13 celeron55     whatever, i can't really say no or yes and can't bother to check the code
23:14 celeron55     the problem with this once again is that this is completely random coming out of nowhere
23:15 celeron55     and there's no clue whatsoever into what kind of coherent whole it belongs
23:16 celeron55     have you thought at all what does it do to the balance of the game?
23:17 proller       it add semi-hidden unique feature
23:17 celeron55     if so, can you note down your thougths and also if there's something else you're about to do to get things more to the same direction
23:17 proller       and make falling not boring
23:18 proller       its good addition to floatlands and any sky-filling mod
23:19 hmmmm         falling is boring?
23:19 proller       in this direction - cant think anything except parachute mod
23:19 celeron55     is it compatible with everything you can imagine?
23:20 kahrl         what would a parachute mod do? currently the wingsuit is hardcoded
23:20 proller       hmmmm, yes, now you can slowly move 4nps and cant fly around something
23:21 proller       kahrl, open parachute -> stop falling
23:21 celeron55     i don't have anything against adding something that allows more useful movement in floatlands, but it's not like this is the only possible implementation on that
23:22 celeron55     assuming the code is perfect, i think i'll give a yes for this altough i have no idea what i'm doing; you'll need an another yes still though
23:23 celeron55     or, well, not really, but at least you need no 'no's 8)
23:25 proller       it doesnt touch usual gameplay wha can said 'no'?
23:26 proller       except PilzAdam of course
23:32 proller       maybe in future make changeable fly control value, for something like real wingsuit mod, but now now too few players to use it
23:56 hmmmm         now why is it that a mod can't have the same effect
23:56 hmmmm         oh, this is movement speed control for the player
23:56 PilzAdam      mods can control the speed, but not the velocity itself
23:57 hmmmm         see, it certainly feels like proller is coding a lot of "specific" things that would be better as client-side mods
23:57 PilzAdam      mods can also limit the movement speed of the players, by setting gravity to 0 when the player reaches it
23:57 hmmmm         i feel a lot of what he does goes against the idea of minetest as no more than an engine