Time  Nick          Message
00:43 VanessaE      it would be nice if #849 could be merged before 0.4.8, as well as pilzadam's default inventory/furnace/chest formspec textures
00:45 VanessaE      (the latter refers to this:  https://github.com/PilzAdam/minetest_game/commit/b395f297759f5d1a4b532b92c7deccbbae2d288f )
00:47 PilzAdam      no, thats glitchy
00:49 VanessaE      works fine in Unified Inventory
00:53 VanessaE      http://digitalaudioconcepts.com/vanessa/hobbies/minetest/screenshots/Screenshot%20-%2008032013%20-%2008:55:59%20PM.png
00:54 VanessaE      it took a few tries to get the images sized and aligned to fit well over the default translucent background, but they work fine in practice and those couple of pixels at each edge are not noticeable in practice.
01:05 iqualfragile  this viewingrange fix realy helps a lot
01:06 VanessaE      indeed so!
01:09 BlockMen      VanessaE, im currently reworking the formspecs to allow properly costum backgrounds -> https://dl.dropboxusercontent.com/s/2x1iyfx5vjaqoq7/inv_rework-dev.png
01:09 VanessaE      oh good
01:09 BlockMen      but i doubt i will be finished until 0.4.8
01:10 VanessaE      nothing says that 0.4.8 can't be delayed until then :)
01:11 BlockMen      i thought 0.4.8 should be released for this magazine?
01:11 VanessaE      well maybe
01:12 VanessaE      I don't know much about that though
01:12 BlockMen      me neither..i just scrolled over the log and remembered sth like that^^
01:14 VanessaE      I think the deadline for it is Aug. 14
01:48 NakedFury     then it can be the center feature for 0.4.9
04:22 celeron55     there's no hurry with formspec texturing; and indeed it's worth waiting for BlockMens's stuff to be ready
04:23 celeron55     -s
04:28 celeron55     hmm
04:28 celeron55     i wonder what kind of action would be hard for a DoS client to fake
04:34 celeron55     probably everything
04:56 celeron55     (i mean, nothing)
05:33 celeron55     i'm running this thing under valgrind and digging up CPU consumption under DoS
05:35 celeron55     there was a thing in the network stack that used a huge lot and i fixed it, and did some other thing; now i'm seeing that a huge auth.txt file is a huge processing hog at each player join
05:36 celeron55     it's basically what allows a server to be "cogged up" with DoS now
05:39 celeron55     or, well
05:39 celeron55     i'm not too sure where the bottleneck tends to be on an actual setup
05:42 celeron55     actually, i'm pretty sure it's this
05:45 celeron55     at least after a while of DoSing, this is what makes it just plain impossible to gain any control anymore
05:53 celeron55     uhm... now the whole thing got stuck
05:54 celeron55     anyway, basically what i want to achieve first is to make network bandwidth the bottleneck for new users, not CPU
05:54 celeron55     (or disk)
06:23 VanessaE      need a horribly-network-laggy server to test with? ;)
06:27 celeron55     also a large problem was that the server put away the task of removing denied clients and got clogged up of them too, trying to send each of them blocks until timeout
06:27 celeron55     (which ended up in practucally an infinite loop until the timeout happens...)
06:27 celeron55     practically*
07:41 nore          I rebased #655 against current master
07:42 nore          It should be merged soon
07:47 celeron55     http://paste.dy.fi/Whf
07:47 celeron55     that makes it so in practice that a server won't be affected by DoS for any time further than the duration of the DoS
07:48 celeron55     and even then, it should lessen the effect by a lot (whether that makes the server usable during the DoS is an another question though)
07:53 celeron55     (it's actually still possible to get it behave pretty badly (not even close as badly as before), but i won't tell how 8)
07:53 celeron55     )
08:01 celeron55     one way to fight the filling up of the maximum player count would be to create a play time timer for each player and add a privilege to them (eg. "regular_user") when it reaches some value and add a "max_nonregular_users" setting to the engine
08:01 celeron55     i'd like to hear some comments from server admins about that
08:02 celeron55     of that all other than the checking of privilege at connect time can be done by a mod
08:02 VanessaE      wouldn't affect my server, as interact isn't granted by default anyway
08:03 celeron55     what does interact have to do with maximum user count filling up
08:03 VanessaE      well I would assume one would want to upgrade a user from "non-regular" to "regular" given enough time, and then interact, if given by default, becomes useful.
08:04 VanessaE      but I get your point
08:04 celeron55     i don't care about interact or anything
08:04 celeron55     it's a thing on a completely different level and completely irrelevant for DoS proection
08:04 celeron55     +t
08:04 VanessaE      true.
08:05 VanessaE      as for the max player count filling up (or not), yours is a good idea for sure.
08:05 celeron55     now i hope some people will still read what i said up there and ignore this
08:09 VanessaE      I'm sure they'll read it, jeez
08:10 celeron55     8)
08:14 * VanessaE    goes back to her TV Tropes clickfest
08:19 celeron55     the max_users setting should be changed to max_players i think
08:19 celeron55     we don't use the term "user" anywhere else
08:19 celeron55     and i need to add a related setting to which i'd hate to use "user" too
08:21 iqualfragile  blockmen mentioned something about a magizine?
08:22 celeron55     probably this http://irc.minetest.ru/minetest-dev/2013-08-01#i_3230599
08:34 iqualfragile  minetest.ru is kind of dieing right now, but thanks for the link
08:37 iqualfragile  thexyz: something is wrong with minetest.ru
08:46 thexyz        iqualfragile: guess there were some connection issues with hetzner
08:48 thexyz        also, don't you think we need a polish translation finished?
08:48 iqualfragile  i allso get a security exception when connecting to forum.minetest.net via https
08:48 iqualfragile  yes, that would be needed when publishing minetest in a polish magazine
08:49 celeron55     this is a diff for the nonregular player limit http://paste.dy.fi/etS
08:49 thexyz        iqualfragile: what exception?
08:50 iqualfragile  ssl_error_bad_cert_domain; tells me that the certificate is only valid for minetest.ru and users.minetest.ru
08:50 iqualfragile  sadly firefox does not even give me the option to ignore that warning
08:50 thexyz        your browser probably doesn't support SNI
08:50 iqualfragile  "firefox"
08:50 thexyz        are you using firefox < 2.0?
08:51 iqualfragile  i am using archlinux so yes
08:51 thexyz        okay
08:52 iqualfragile  hmm… strange, it works in konqueror
08:53 iqualfragile  there has been another user reporting the same problems
08:54 iqualfragile  and i can not remember changing any settings
08:54 celeron55     i use firefox and arch and have no problems
08:55 celeron55     (firefox 21.0)
08:57 iqualfragile  hmm… werent you using linuux mint?
08:57 celeron55     i don't use the same distro on every computer
08:58 celeron55     (i have more archlinux than anything else installed)
09:01 thexyz        iqualfragile: try now
09:01 iqualfragile  does not work
09:02 thexyz        and now?
09:02 celeron55     humm
09:03 celeron55     i do get the same error if i use https on that URL
09:03 celeron55     but the url has http, why would i
09:03 iqualfragile  not yet
09:04 thexyz        celeron55: well, iqualfragile gets error when connecting to forum.minetest.net
09:04 thexyz        that's rather odd
09:05 celeron55     now that i think of it, i might have gotten an error on https://forum.minetest.net
09:05 celeron55     i generally don't care about SSL so i don't remember 8)
09:05 iqualfragile  when i use https for the irc.minetest.net link i get the same error but here i have the option to add an exception
09:07 celeron55     you don't have an option to add an exception on forum.minetest.net?
09:08 celeron55     that has definitely not happened to me; even if it failed to verify it, it did allow adding an exception
09:08 celeron55     i don't see an exception in ff's exception list
09:09 celeron55     so none of it has happened to me
11:05 celeron55     is it so that there is no further performance improvement to come for fancier trees in the near future?
11:06 celeron55     if it is, then work should be started to add some of them to minetest_game in a lag-aware way
11:09 celeron55     as far as i can see, the decoration stuff isn't really moving forward at the moment and as for what i have heard, it won't affect speed much
11:09 celeron55     am i wrong?
11:13 PilzAdam      celeron55, I think the best way to add fancier trees to minetest_game are schematics
11:13 jin_xi        lol
11:14 PilzAdam      the decoration doesnt even support L-system currently
11:16 PilzAdam      and in my other game I already use schematics to place trees, and huge jungles are no problem at all
11:17 celeron55     how do you generate the schematics?
11:17 celeron55     by copying results of l-system trees generation? 8)
11:17 PilzAdam      I have created a normal and a jungletree schematic by hand
11:18 PilzAdam      then do some hacky code to get a probability table and you are done
11:18 celeron55     i'd like to see a result of that
11:19 PilzAdam      https://github.com/PilzAdam/pilztest
11:19 PilzAdam      http://mg.viewskew.com/u/pilzadam/m/jungle-in-the-pilztest-game/
11:20 jin_xi        one has to wonder why ltrees are so slow even when consistion of only a few nodes (9 in my roads test)
11:20 PilzAdam      the schematics are in mods/base/schematics/ and the relevant decoration definitions are in https://github.com/PilzAdam/pilztest/blob/master/mods/base/mapgen.lua#L90
11:21 Exio4         jin_xi: i think it is because the lua overhead, mainly
11:22 PilzAdam      jin_xi, http://irc.minetest.ru/minetest-dev/2013-08-03#i_3235686
11:22 jin_xi        what lua overhead? i think it has to do with them loading fixed size of blocks regardless of tree size
11:22 jin_xi        PilzAdam: aha
11:22 Exio4         how do you 'spawn' something with ltrees?
11:23 Exio4         ah
11:23 celeron55     the speed in pilztest is pretty much perfect; this'd just need more variation and it'd compete directly with moretrees
11:25 celeron55     it's just a matter of adding more leaf and trunk types schematics i guess
11:25 celeron55     +and
11:25 celeron55     and some kind of logic for where to place what
11:29 celeron55     i'd just copy stuff directly from what moretrees generates 8)
11:29 PilzAdam      the trees can be easily placed by perlin noise, though I used "fill_ratio" because Im not very familiar with noise stuff
11:30 jin_xi        so, would it be feasible to make ltrees just generate schematics?
11:32 iqualfragile_ it would feel quite wrong to use static schematics when you could use l-trees instead
11:33 celeron55     while it'd be neat, i really don't care
11:33 Exio4         i think ltrees should be used for "super random" things like trees, or so :P
11:35 celeron55     when making a game it's better to change attitude to taking what works and just producing the wanted end result; all the pedanticism can be left for when working on the engine 8)
11:45 jin_xi        about the ltrees and engine features: i still think it would be good to split the functionality into two parts: one generic turtle to actually place stuff and then an l-system rules parser to generate turtle programs
11:45 jin_xi        then use schematics (or whatever else fast method) to place the results
11:46 jin_xi        hmmmm hinted at something here: http://irc.minetest.ru/minetest-dev/2013-08-03#i_3235702 and i am gonna give this a try soonish
11:47 jin_xi        but then my coding is hardly up to mt standards... so i might need a lot of help
11:48 PilzAdam      so, you want that the L-System interpreter in the engine creates schematics?
11:48 jin_xi        here is the turtle system i have so far: https://github.com/obneq/minetest/tree/turtle
11:49 jin_xi        PilzAdam: if it is so much faster to place them, why not, but see irc log link above, idk what hmm meant by that
11:49 PilzAdam      it would actually slow it down
11:50 jin_xi        oh :( well
11:50 PilzAdam      hmmmm said that the lighting needs to be omptimized
11:51 celeron55     why is lighting fast for schematics?
11:52 jin_xi        ^
11:52 Exio4         because it uses the 'new' lighting algorithm?
11:52 celeron55     i don't understand why these have so much speed difference
11:52 celeron55     there's no rational reason for it
11:52 Exio4         if i understand correctly
11:52 Exio4         the old one is slower because it is less glitchy
11:52 PilzAdam      well, hmmm said they have; 1) the lsystem uses the "old" lighting code and 2) it takes every corner case into account
11:55 PilzAdam      also the lsystem is not the only reason why plants_lib is so slow
11:56 celeron55     that's why i said "as for what i have heard" 8)
11:56 celeron55     i do guess that the lua placing thing in it is not fast at all, but i have no proof of that
11:58 celeron55     does anyone have anything to say about these? http://paste.dy.fi/Whf http://paste.dy.fi/etS
11:59 celeron55     the first one is kind of only a fix, but the second one has some design in to which someone might want to suggest something
12:01 nore          I have rebase #655, do you have anything do say against it?
12:01 nore          s/rebase/rebased
12:02 PilzAdam      celeron55, how do the DoS scripts work? do they send INIT and INIT2?
12:02 Exio4         celeron55 should say why it was 49 nore
12:02 nore          yes, we still don't know that...
12:03 celeron55     PilzAdam: that's what i tested
12:03 RealBadAngel  as some1 said two last digits of phone number of girl c55 dated night before ;)
12:03 celeron55     flooding that is the most effective way to kill the server currently
12:03 celeron55     and it's very effective, the server quickly goes unresponsive for an infinite time even if the attack is stopped
12:04 celeron55     that fixes the after-effect completely
12:04 RealBadAngel  i am fine with this limit being configurable, so i think #655 is needed
12:05 RealBadAngel  49 limit is causing heavy damage to workin tubes setups
12:06 celeron55     Exio4: nore: it was set to 49 because ot was 50 before that, because when it was 50 many mapblocks were flooded with 50 monsters so the obvious solution was to delete everything from those that contain more than 49 and limit it to 49 from then
12:06 nore          yes I had that (with teleport tubes)
12:06 celeron55     that was sometime before 0.3
12:06 celeron55     and some limit exists for obvious reasons
12:06 nore          so now it should be best that it could be configured
12:08 nore          and if one wants to delete everything he can set the limit to 0...
12:08 celeron55     the name of the setting is completely bad
12:08 nore          what should it be then?
12:09 Exio4         wait, why is it raised up to 500? what's up with entity duplication?
12:09 celeron55     the best that comes to mind: max_objects_per_block
12:09 celeron55     also don't raise it, we have too much bugs currently to afford that
12:09 nore          ok, changing that now...
12:12 nore          done
12:13 celeron55     you forgot the .example 8)
12:14 nore          done
12:15 nore          and anything new about #852?
12:17 RealBadAngel  i will test it now for a while
12:18 RealBadAngel  atm im fighting with wrench mod, dunno why but it seems to not workin
12:18 RealBadAngel  have anybody got problems with it?
12:18 nore          dunno, not tested for now
12:18 Exio4         RealBadAngel: do you have any online/public branch for your shaders?
12:19 RealBadAngel  it seems like when placing the node itemstacks meta is not accesible
12:20 RealBadAngel  were there any changes made lately to itemstacks or its meta?
12:20 RealBadAngel  Exio4, i will make the repo today. i were offline and busy for a while, just moved to new flat
12:21 Exio4         ah
12:30 celeron55     so i guess https://github.com/minetest/minetest/pull/852 is to be merged now?
12:32 celeron55     has sapier commented on it?
12:45 PilzAdam      I really dont like how all/ is called "Default" in #852
12:47 sfan5         #856 should be merged, shouldn't it?
12:47 celeron55     i too think it's not very logical
12:47 celeron55     maybe "None" would be better
12:47 PilzAdam      yep
12:47 PilzAdam      and there is no reason to keep the old all/ pack
12:48 celeron55     sfan5: has somebody tried to use it and seen that it works?
12:48 sfan5         IIRC not
12:48 sfan5         I'll try it
12:50 celeron55     but i see at least one problem in it
12:50 celeron55     it calls minetest.check_player_privs by default while the engine already does that
12:51 sfan5         it should return true by default
12:52 celeron55     it's probably there for automatically handling some kind of custom cases, but checking multiple times for the general case isn't good
12:52 PilzAdam      maybe pass pointed_thing instead and add it for entities too
12:53 PilzAdam      then the priv checking can be completly done in Lua
12:54 celeron55     i guess i'll merge #852 and change "efault" to "None" (and fix a thing in a header)
12:55 celeron55     +D
12:56 sfan5         856 works
12:57 sfan5         may I merge it after I removed the check for interact privs?
12:57 PilzAdam      are you against my suggestion?
12:57 celeron55     sfan5: don't
12:57 sfan5         no
12:58 celeron55     sfan5: it needs further refinement
12:58 sfan5         refinement = for entities too?
12:59 celeron55     dunno
12:59 celeron55     (maybe)
13:00 celeron55     lol, the texture pack menu lists the texture_packs_here.txt file in the UI
13:00 sfan5         so refinement only refers to the duplicate interact check
13:01 sfan5         or am I wrong?
13:02 celeron55     sfan5: that should be attempted to be taken care of, and entities too, and maybe something else too or instead if it turns out so
13:02 sfan5         I already did the first thing and squashed my commit into the other
13:02 sfan5         (locally)
13:02 PilzAdam      the engine checks the interact priv for moving stuff out of the inventory and crafting too
13:02 PilzAdam      that could be passed to Lua too
13:04 celeron55     what is the goal here in the first place
13:04 Exio4         i don't think that is needed PilzAdam
13:04 celeron55     we could pass every node access through a privilege function but that would be pointless
13:05 celeron55     is the point position-dependent interactions?
13:06 celeron55     it would seems so
13:06 PilzAdam      the player should not be able to drop items in protected areas
13:06 celeron55     -s
13:06 Calinou       PilzAdam: what if you want to trade items?
13:06 PilzAdam      while crafting could be allowed there
13:06 PilzAdam      Calinou, then youd have to go to unprotected areas or install a direct trade mod
13:07 PilzAdam      bascically the point is to leave it up to the protection mod to decide such things
13:07 PilzAdam      thats why its reasonable to pass such things to Lua
13:08 celeron55     by the way, the interact privilege is built-in to the client too; it can't be bypassed in certain cases
13:08 celeron55     that's why the check must be done by the engine
13:09 celeron55     so really the correct way mihgt be to just return true in the default lua function
13:09 celeron55     might*
13:10 celeron55     simply no interaction should be possible without interact; that's the point of that privilege
13:10 nore          celeron55: what should I change in #852?
13:10 Exio4         in a dream world; the client should be able to "pre-known" those places and predict it
13:10 celeron55     Exio4: no, that's wrong and poisonous to the original design
13:11 celeron55     or, well, it could do that for additional privileges
13:11 Exio4         well, with client-side lua in the "on_place" that could be possible
13:13 celeron55     nore: fix it
13:13 celeron55     8)
13:13 celeron55     texture_packs_here.txt shows up in the gui for me
13:13 nore          ok for changing "Default" to "None", but what is that header thing?
13:13 celeron55     i already did the change of the name
13:14 celeron55     and the header thing
13:14 celeron55     just figure out how to hide files
13:14 nore          it is already done to ignore base/
13:15 celeron55     what?
13:15 nore          in fact, I need to change the calls to get_dirlist, to use second parameter=true (ignore files
13:16 nore          nvm about ignoring base/ , this is for a specific folder/file
13:16 celeron55     oh it works now
13:16 celeron55     good
13:16 celeron55     now i just need a texture pack... 8)
13:17 nore          what I did was to copy a single stone texture, and to use it as screenshot too...
13:17 PilzAdam      celeron55, could you send me the new patch? I have a ton of texture packs
13:18 celeron55     http://paste.dy.fi/ekF
13:18 celeron55     nice binary garble
13:20 sfan5         hows this http://hastebin.com/divebetavi for #856 ?
13:20 PilzAdam      sfan5, an you use a pastebin that doesnt require JS?
13:21 sfan5         i just typed "git show | haste" because it was easier
13:21 Exio4         example in lua_api.txt?
13:21 Exio4         hmm
13:22 sfan5         PilzAdam: http://pastie.org/8205341
13:22 PilzAdam      celeron55, the all/ folder can be simply removed now
13:23 RealBadAngel  852 has a few whitespace errors
13:23 celeron55     there are too few opinions on removing it
13:23 PilzAdam      the screenshot shows moretrees and the description still says "default"
13:23 celeron55     oh by the way
13:23 celeron55     how about server-side texture packs
13:24 celeron55     the server always uses textures/all for those
13:24 nore          what for?
13:24 PilzAdam      the server should always send the textures that are in the mods
13:25 celeron55     what for? ehm... well maybe for setting default textures for a server? i mean, that functionality already exists and has existed for long
13:25 PilzAdam      otherwise people might accidentally send their HD textures
13:25 celeron55     eh
13:25 celeron55     people want to set textures for their servers without going to copy each texture to the correct mod
13:25 nore          if server uses textures/all, it is the same as no TP, so no long loading time
13:25 celeron55     removing that functionality would be utterly stupid
13:26 nore          because textures are in their own TP folders
13:27 celeron55     but of course the server shouldn't send the textures in the pack that the local player is using
13:27 PilzAdam      maybe use a "server" folder in textures/ ?
13:28 celeron55     maybe
13:28 celeron55     that doesn't tell what happens to 'all' though
13:28 PilzAdam      all just gets removed
13:28 PilzAdam      there is no reason to keep it
13:28 celeron55     but what does removing it mean
13:29 celeron55     the game sure doesn't go and delete the folder
13:29 PilzAdam      we remove the folder from the repo
13:29 nore          with TP selection, no textures are to be put in textures/all for client, and those in it are for server
13:29 PilzAdam      so if people have textures in all/ then it will show up as a texture pack
13:29 nore          so it is the same as a server/ folder
13:30 celeron55     PilzAdam: i'm fine with that
13:30 celeron55     however, it's just us three arguing
13:30 nore          what should then be the folder selected for "No texture pack"?
13:30 celeron55     no folder
13:31 celeron55     but... yes, that'd need to be as a special case in the gui
13:31 celeron55     which makes sense
13:31 nore          the problem is that with the "all" folder, the player can change texture_path in minetest.conf to get a different folder in which texture packs are stored
13:31 PilzAdam      also dont try to find a screenshot or info.txt for it
13:31 celeron55     just insert to the top of the list
13:32 nore          celeron55: it is always at the top for now, because of the renaming
13:33 celeron55     anyway you can see my header addition at line 242 here: http://paste.dy.fi/ekF
13:34 celeron55     i think PilzAdam is right in how it should work; remove "all" from the filesystem and make it add a top entry which is not an actual directory that says "No texture pack", with no image
13:34 celeron55     in other words, consider "all" a regular texture pack
13:35 PilzAdam      and texture_path is set to "" if none is selected
13:35 nore          ok, but how can the user change the path in which texture packs are stored then?
13:35 PilzAdam      texture_pack_path in minetest.conf?
13:35 nore          because with the current way, he can change it to a new folder by editing minetest.conf
13:35 celeron55     what?
13:36 celeron55     but aren't you modifying the setting in the UI?
13:36 nore          if he changes texture_path to some_random_folder/all
13:36 PilzAdam      default for texture_pack_path would be $path_user/textures/
13:36 nore          the texture packs looked for will be in some_random_folder
13:36 celeron55     no no no no
13:36 celeron55     there shouldn't be such a path
13:36 nore          why?
13:37 celeron55     there's no reason the lookup path shouldn't be strictly based on $path_user
13:37 celeron55     because so isn't games and so isn't mods' path
13:37 nore          for now the default is games/../textures/
13:38 PilzAdam      and what is "games"?
13:38 celeron55     nore: do you mean $path_user/textures/`
13:38 celeron55     ?*
13:39 nore          well, the code asks where are the games stored (remove_last_folder(engine.get_gamepath())..DIR_DELIM.."textures")
13:39 celeron55     i wonder what engine.get_gamepath() does
13:40 nore          engine.get_gamepath()
13:40 nore          ^ returns path to global gamepath
13:40 celeron55     "gamepath" isn't an official name for any path
13:40 celeron55     sapier...
13:40 nore          from menu_lu_api.txt
13:40 celeron55     sapier oh sapier
13:40 nore          it is the games/ folder
13:40 PilzAdam      porting::path_user + DIR_DELIM + "games" + DIR_DELIM
13:41 PilzAdam      ^ thats what get_gamepath() returns
13:41 celeron55     that's ridiculously arbitrary
13:41 nore          in a RUN_IN_PLACE=0 install, I reckon
13:41 celeron55     also adding DIR_DELIM to the end if a path variable should never be done
13:42 celeron55     i hate it when a base for a code is bad
13:43 nore          remove_last_folder works whenever the last character is DIR_DELIM or not
13:43 celeron55     we need to first fix this menu api
13:43 celeron55     before we consider nore's stuff
13:44 nore          I reckon BlockMen did a pull to fix some of the menu stuff too.
13:46 celeron55     anyway throw that "From tile.cpp:" in the texture_cache.h comment
13:47 nore          that: A cache from texture name to texture path
13:47 nore          ?
13:47 PilzAdam      tile.cpp:140 should be removed if the new stuff goes in
13:48 celeron55     nore: like on line 242 here: http://paste.dy.fi/ekF
13:48 nore          I do not have such comment...
13:48 nore          strange
13:48 celeron55     it's a header with no corresponding .cpp file so nobody knows what it's doing otherwise
13:48 celeron55     nore: yes, add it
13:48 celeron55     that's the point
13:49 nore          ah, ok
13:49 celeron55     anyway, i need a break
13:49 PilzAdam      nore, and remove the whole "Check from $user/textures/all" block from tile.cpp:138
13:49 celeron55     but overally the menu api must be fixed first by adding some kind of get_userpath or other sane function
13:50 nore          OK, I will do both
13:51 nore          I commented the block for textures/all, and no screenshot nor information for the "None" texture pack
13:52 celeron55     or alternatively get_texturepath
13:52 PilzAdam      dont comment code out, just remove it
13:52 celeron55     something that makes sense rather than these tricks
13:52 nore          OK
13:52 kahrl         why is there a texture_cache.h? can't this be in tile.h?
13:53 nore          IIRC, it was to fix a multiple variable problem
13:53 nore          but I had to change it to extern, so I can put it again in tile.h
13:53 kahrl         yeah that sounds like the extern was missing
13:57 PilzAdam      nore, https://gist.github.com/PilzAdam/6150399
13:58 PilzAdam      you can use that for now, the whole thing has to be fixed later on anyways (as celeron55 said)
13:59 nore          I had just finished to do it
13:59 nore          but thanks anyway
14:07 nore          I updated the pull, I did not squash the commits yet since I removed a few things, but tell me if it is OK
14:08 PilzAdam      you dont need to add /textures/texture_packs_here.txt to gitignore, just  git add -f textures/texture_packs_here.txt
14:08 PilzAdam      also always at a newline at the end of files
14:09 nore          yes, but it still should be here, like textures_here.txt
14:09 nore          and what file did I add?
14:09 PilzAdam      https://github.com/minetest/minetest/pull/852/files#L8R1
14:10 nore          yes, of course ;)
14:10 PilzAdam      and  git rm -r textures/all
14:10 nore          ok, done
14:11 nore          PilzAdam: no textures/all is still needed (for no TP, and for server)
14:12 kahrl         oh, this includes util/container.h from tile.h, that might be bad for compile times
14:12 kahrl         tile.h is included from a lot of places and util/container.h is template heavy
14:13 nore          kahrl: so I should create again my file texture_cache.h?
14:13 kahrl         no, maybe remove the extern declaration and instead add a function clear_texturename_to_path_cache
14:14 nore          the problem is that if I remove extern, I get multiple variable declaration error
14:14 kahrl         sorry, I didn't see it when I asked about texture_cache.h earlier
14:14 kahrl         no, I mean remove the entire declaration
14:14 nore          but then, g_texture... is out of scope in game.cpp
14:15 kahrl         that's what the clear... function is for
14:15 nore          where should I put the function then?
14:15 PilzAdam      nore, textures/all is supposed to be removed completly, and textures/server should be used in a server
14:15 kahrl         tile.cpp and declared in tile.h
14:15 nore          PilzAdam: and for no TP?
14:15 nore          kahrl: but where should the function be declared then?
14:15 PilzAdam      what for no TP?
14:16 kahrl         hm?
14:16 nore          it needs a folder for TP = None
14:16 kahrl         below getTexturePath
14:16 nore          karhl: because if it is in tile.h, g_texture... will be out of scope
14:17 kahrl         declare it in tile.h and define it in tile.cpp
14:17 nore          and in tile.cpp, it will be out of scope in game.cpp
14:17 nore          karhl, Ok, I do not master enough C right now
14:17 nore          I understand
14:17 nore          do you think it is better?
14:17 PilzAdam      nore, no, no folder for "None", it should be hardcoded in the list
14:17 PilzAdam      its a special case
14:18 nore          yes, but how could the user change the folder in which he stores TPs?
14:18 kahrl         it's good to limit usage of a global variable to a single cpp file, too
14:19 PilzAdam      nore, he doesnt, just like he cant change the folder where mods are installed
14:19 nore          ok, doing it (again...)
14:19 kahrl         PilzAdam: symlinks ;)
14:19 PilzAdam      kahrl, in windows
14:20 kahrl         doesn't windows get symlinks a while ago?
14:20 kahrl         didn't*
14:21 celeron55     maybe some of us should just do it 8)
14:23 nore          kahrl: is it OK like that?
14:24 kahrl         the name clear_texture_cache is a bit unspecific (does it remove all texture files in cache/media?)
14:24 kahrl         i'd call it clearTextureNameToPathCache
14:25 celeron55     why not just "texturename cache"
14:25 celeron55     the fact that it maps them to paths is somewhat irrelevant
14:25 kahrl         yeah, clearTextureNameCache is fine
14:25 kahrl         or clearTexturePathCache?
14:26 celeron55     hmm, that's actually best
14:26 PilzAdam      that reminds me to the story with nuclear power plants and bikes :-)
14:27 nore          ok
14:31 celeron55     PilzAdam: but it's important to get at least something right"
14:31 celeron55     !*
14:33 nore          PilzAdam: could you tell me what you think of the last commit?
14:36 PilzAdam      what needs to be changed to let the server grab the texture from $path_user/textures/server/?
14:36 PilzAdam      +s
14:38 celeron55     server.cpp:4277? 8)
14:38 nore          it is changed, just look at the commit again ;)
14:38 celeron55     that one should use $path_user though, now it relies on the working directory being $path_user
14:39 nore          just a question: in a RUN_IN_PLACE=1 install, path_user = Minetest directory?
14:39 nore          s/=/==
14:39 celeron55     yes
14:39 nore          ok
14:40 nore          tell me if everything works correctly for you, I will squash the commits.
14:40 celeron55     anyway change the two lines to paths.push_back(porting::path_user + DIR_DELIM + "textures" + DIR_DELIM + "all");
14:40 celeron55     (remove the std::string path_all = "textures"; line, it's useless)
14:41 celeron55     i mean, "server" there of course
14:41 nore          done
14:42 nore          does it work now?, I did not test it yet...
14:43 PilzAdam      nore, why have you moved #include "util/container.h" down in tile.cpp? it doesnt build if its included after util/thread.h
14:44 nore          that it because I removed it and added it again...
14:44 nore          I will move it again
14:44 celeron55     testing
14:45 kahrl         PilzAdam: huh, that shouldn't break the build
14:45 nore          PilzAdam: how do you move a commit?
14:45 nore          to squash it with the one before last
14:46 kahrl         just add another commit and squash all of them when everything is done
14:47 celeron55     well one can freely move commits around in git rebase -i; it's probably not wise though if you're aiming for a single squashed commit as a result though
14:47 nore          It is what I did for now
14:49 PilzAdam      https://github.com/PilzAdam/minetest/commits/texture_packs2
14:49 PilzAdam      squashed and fixed
14:49 PilzAdam      is anything missing in there?
14:50 celeron55     i'm too lazy to test anymore :P if it works for you, let's call it done
14:50 PilzAdam      oh, I have moved #include "util/container.h"  too
14:51 nore          Me too
14:51 celeron55     oh
14:51 celeron55     one thing
14:51 celeron55     don't put two lines in texture_packs_here.txt
14:51 PilzAdam      why?
14:51 nore          I put only one line, I reckon
14:51 celeron55     there's a reason for it: when minetest is built on linux for windows, you end up with \n newlines in there, and if you look at that with notepad, there's no newline
14:52 celeron55     or, well, i have used that criterion for those short files
14:52 nore          One last question: should I add my name in "Contributors"?
14:53 celeron55     people on that list have contributed quite a lot, so maybe not yet 8)
14:54 nore          is there a minimum number of contributions? ;)
14:54 nore          that makes 2 commits for me (perhaps not enough)
14:54 celeron55     yes, the minimum number is "quite a lot"!
14:55 celeron55     refer to the stupidly vague numerals handbook
14:55 * nore        tries to hide...
14:55 PilzAdam      the one with the least amount of commits there is 0gb.us with 5 commits
14:55 nore          of course, trying to compete with sapier is impossible...
14:56 Exio4         that is impossible
14:56 Exio4         you would need to write 10k of lines in two days
14:56 PilzAdam      according to the "12 Month Commits" list of oholoh Im the first
14:57 PilzAdam      and thats mainly merging other peoples stuff :-p
14:57 celeron55     commit stats isn't a reliable metric
14:57 Exio4         https://github.com/minetest/minetest/graphs/contributors
14:57 celeron55     it measures how wastefully one uses commits :P
14:58 Exio4         i only have 9 commits vs the 11 of jeija
14:58 Exio4         he has 11 commits / 2,133 ++ / 153 -- ..
14:58 Exio4         (vs 9 commits / 80 ++ / 17 --)
14:58 PilzAdam      pushed that thing now
14:58 Exio4         3 commits and like 2100 lines of dif!)
14:58 Exio4         er, that is a spare )
14:59 celeron55     nore: it's a good first contribution anyhow though (you can tell that from not being put in the trash)
15:00 nore          so I need my other pulls being merged...
15:00 nore          that is force_load, and tool callback
15:01 Exio4         i like the feeling of looking at some code and say "ah! that is how this is working!" :P
15:01 PilzAdam      everyone agrees on that force_load thing but nobody dares to merge it
15:01 nore          celeron55: it is the second, the first was the one you merged a few hours ago
15:01 PilzAdam      chances that it breaks everything are too high
15:02 celeron55     nore: you might want to refer to the numbers if you want anyone to even maybe look up something
15:02 nore          force_load will perhaps be the pull request that will have waited the most time before merging... ;)
15:02 nore          it is almost 1 year old now
15:02 PilzAdam      1 year is nothing in Minetest...
15:03 nore          for a pull request?
15:03 PilzAdam      currently "Sensible require system" is the oldest open pull request with 11 months
15:04 celeron55     that isn't ever going to be merged, we have decide to use a different kind of requirement system
15:04 celeron55     might be a good idea to close it :P
15:04 celeron55     +d
15:04 PilzAdam      I quote you there :-p
15:05 celeron55     you didn't even write my name there, bah
15:05 PilzAdam      now its "Possible to not type the password every time, and use strong passwords."
15:05 PilzAdam      an "and" in pull request is generally a bad sign
15:06 PilzAdam      *pull requests and commit messages
15:06 celeron55     "add new block and rework everything"
15:06 nore          I have as many commits and more line diffs than VanessaE and 0gb.us...
15:06 nore          but nvm, I will make more pull requests
15:06 PilzAdam      I dont know why 0gb.us is in that list
15:06 celeron55     no don't, we hate handling these!
15:07 celeron55     hey but
15:07 celeron55     how about the two of mine
15:07 Exio4         V is the 'testing girl' here :P
15:07 Exio4         something to try? tell her!
15:07 Exio4         it is like trying the worst-performance-case that is possible when using minetest
15:07 celeron55     i'll merge the DoS fix now but let's finalize this one: http://paste.dy.fi/etS
15:08 celeron55     one of the things was that i would like to change max_users to max_players
15:08 celeron55     because we don't call players users anywhere else
15:08 PilzAdam      and add fallback code to max_users?
15:08 PilzAdam      (since we dont want to break peoples minetest.confs)
15:09 nore          minetest.register_setting_alias?
15:09 Exio4         the server admin should look at the changelog
15:09 celeron55     well it's a server setting, maybe we can assume they follow some news
15:11 celeron55     (aactually, i seem to have merged that previous one already)
15:12 celeron55     so is it necessary to be able to fall back to max_users?
15:12 PilzAdam      maybe not
15:12 celeron55     it requires removing the old and new setting from defaultsettings.cpp and handling them specially everywhere where they are used
15:13 celeron55     (in two places)
15:13 PilzAdam      it would mess up the code too much
15:13 celeron55     maybe i'll just make the server log a warning at startup if it sees max_users
15:15 nore          I will make a post on the forum about the TP selection.
15:15 celeron55     is the "regular" privilege name any good? and the max_nonregular_users/players setting?
15:16 celeron55     it's supposed to mean that the player is confirmed to be actually a human playing the game and not just visiting or possibly being some bot or script
15:16 PilzAdam      nore, why?
15:17 nore          for texture pack creators
15:17 PilzAdam      nore, ok, I think its better if I write it in "News"
15:17 nore          I am writing the post, could you move it?
15:17 celeron55     maybe nore writes it and somebody moves it there
15:17 PilzAdam      that works too
15:18 PilzAdam      but somebody else has to move, Im not a moderator
15:18 celeron55     i am :P
15:18 PilzAdam      nore, also include an example structure of the textures/ folder
15:18 Jordach       i can move it if the post IS WRITTEN.
15:19 Jordach       (we are allowed bold here right?)
15:19 celeron55     no
15:19 Jordach       oh
15:19 sfan5         thats caps not bold
15:20 Jordach       sfan5, i don't know if +c is on this channel
15:20 PilzAdam      the +C flag is set
15:20 Jordach       thank you.
15:20 PilzAdam      your IRC client should show that
15:20 nore          https://forum.minetest.net/viewtopic.php?pid=103752#p103752
15:20 Jordach       https://forum.minetest.net/viewforum.php?id=18
15:20 Jordach       nore, ^
15:21 sfan5         Jordach: oh you meant, color bold
15:21 Jordach       yus
15:21 Calinou       "- there should be a "screenshot.png" that will be a screenshot of the TP. Its format must be 4:3." which size in pixels?
15:21 Jordach       he means 128^2
15:22 Calinou       that's not 4:3
15:22 Calinou       4:3 would be 512×384
15:22 PilzAdam      somone update this: https://forum.minetest.net/viewtopic.php?id=1592
15:22 celeron55     i will
15:22 nore          there is no size, the image will be resized (so if you put a 128x128 image, it will be squashed)
15:23 Calinou       I mean, so that it looks good
15:23 Calinou       and not pixelized
15:23 sfan5         what has do be done for #856 (Add basic protection support to builtin) aside from the doublecheck and adding support for dropping items too?
15:23 nore          I don't know... (I think it is displayed 160x120, or at least VanessaE told me so)
15:24 nore          Calinou: edited the post
15:24 Calinou       ok
15:25 PilzAdam      nore, add this as example structure: https://gist.github.com/PilzAdam/6150673
15:26 nore          PilzAdam: done
15:27 nore          that should be added in celeron55's post about texture pack installation too.
15:27 PilzAdam      yea
15:27 PilzAdam      and in the wiki page
15:27 PilzAdam      (is there one about texture packs?)
15:29 nore          dunno
15:30 nore          http://wiki.minetest.net/Texture_Packs
15:30 PilzAdam      someone should update that
15:32 celeron55     remember to not remove the instructions for 0.4.7 and earlier though
15:32 PilzAdam      I just copied the "Installing Mods" from the dev wiki to wiki.minetest.net
15:35 nore          celeron55: two space are missing in the last lines (indent not correct)
16:14 celeron55     i think that deprecation warnings that only happen once per server startup should use log level "error"
16:15 celeron55     or actually... i think it's just up to the importance of the particular warning
16:15 celeron55     (currently they use level "info")
16:31 celeron55     hmm, a texture pack selection menu bug: why does it show the game icons?
16:31 celeron55     it doesn't do anything game-specific
16:33 nore          what do you think about that sounds pack idea?
16:33 nore          https://forum.minetest.net/viewtopic.php?pid=103762#p103762
16:33 celeron55     19:32:03 < celeron55> hmm, a texture pack selection menu bug: why does it show  the game icons?
16:34 celeron55     19:32:07 < celeron55> it doesn't do anything game-specific
16:34 nore          game icons are always shown, I reckon
16:34 celeron55     no they aren't
16:34 celeron55     they are now shown in singelplayer tab and texture pack tab
16:34 nore          in fact not
16:35 nore          it is because I copied code from singleplayer tab
16:35 celeron55     well that is obvious
16:35 nore          doing a commit to fix that
16:36 PilzAdam      just removing ..menubar.formspec works
16:36 nore          just need to remove line 1010 of mainmenu.lua
16:36 PilzAdam      and 1038
16:36 VanessaE      regarding texture packs,
16:36 nore          yes
16:37 VanessaE      since I missed out on the whole conversation...  I'm okay with the current thinking, as long as there is a user-level folder in which one can place files that override (or supplement) the selected texture pack
16:37 VanessaE      user-level as in works in a system-wide install e.g. a folder somewhere under ~/.minetest/
16:38 nore          textures/base/pack should work
16:38 nore          or textures/server
16:38 PilzAdam      nore, that is in $path_share
16:38 celeron55     VanessaE: there isn't anything like that
16:38 VanessaE      these are not server-side textures I'm thinking of, nore
16:38 VanessaE      celeron55: there used to be.
16:38 celeron55     why do you need such?
16:38 VanessaE      PilzAdam just broke it by pushing that commit through
16:38 celeron55     why nobody else needs such?
16:38 VanessaE      celeron55: I use custom fonts.
16:38 celeron55     if you are the only one requesting it, you won't get it
16:39 VanessaE      as do a number of others.
16:39 nore          I have been asked to remove it
16:39 VanessaE      those by definition must be independent of a user's texture pack
16:39 PilzAdam      VanessaE, no, there used to be only 1 texture pack
16:39 VanessaE      PilzAdam: wrong.
16:39 celeron55     VanessaE: find someone else who wants the same thing, bring them here and we can talk
16:39 VanessaE      PilzAdam: before, you could set texture_path=/foo/blah  and also put a few files in ~/.minetest/textures/all  and *both* paths get used exactly as expected.
16:40 celeron55     i know that functionality existed before, but it was not advertised nor specified as being available, and also almost nobody used the functionality as such
16:40 VanessaE      celeron55: \https://forum.minetest.net/viewtopic.php?id=4436
16:40 VanessaE      https://forum.minetest.net/viewtopic.php?id=4436  *
16:40 VanessaE      several people at least there.
16:40 PilzAdam      why not use TTF?
16:41 PilzAdam      I am sure there is a proper font for you somewhere
16:41 VanessaE      PilzAdam: <offtopic> because TTF can't be shadowed. </offtopic>
16:41 PilzAdam      you can even set the size in minetest.conf
16:43 VanessaE      you could have just left the check for textures/all and made the TP selector ignore the presence of that folder and it would have been fine
16:44 VanessaE      there is no reason to remove functionality like that if it doesn't cause a performance problem or confuses people.
16:44 VanessaE      or doesn't confuse*
16:44 celeron55     well now that there is an actual example of it's valid usage, the request seems a bit more justifiable
16:44 PilzAdam      if its incosistent with the new system then there is a reason
16:44 celeron55     if it's added back, it's going to be called textures/override
16:45 celeron55     because that's what it is
16:45 VanessaE      celeron55: also, as an aside, note the date on that post... :)
16:45 nore          one question: should the "server" folder be hidden by the TP selector?
16:45 Calinou       +1 for selecting "all" pack and hiding it, for backwards compatibility
16:45 PilzAdam      nore, no
16:45 VanessaE      I'm fine with almost any path you want to use, as long as it doesn't require admin rights to populat eit.
16:45 celeron55     VanessaE: by the way, none of the users who you have told to put the font in textures/all have their texture pack in an another location via texture_pack
16:45 VanessaE      populate it*
16:45 celeron55     VanessaE: i'm pretty sure
16:45 Exio4         at least it isn't like https://xkcd.com/1172/
16:46 celeron55     texture_path*
16:46 VanessaE      celeron55: true, but I didn't want to make a specific recommendation to that end.
16:46 PilzAdam      Calinou, wut? backwards compatibility is is possibel, since you can simply select "all" in the GUI
16:46 Calinou       nevermind then
16:46 Calinou       don't know what's VanessaE about then
16:47 celeron55     VanessaE: but this brings the problem that you are probably the only one in the world using it 8)
16:47 PilzAdam      Calinou, she wants another level on top of texture packs
16:47 VanessaE      Calinou: because if textures/all is selected, and all it has is a couple of fonts, then say textures/sphax won't be used
16:47 VanessaE      celeron55: I've seen screenshots here and there that would disagree with you :)
16:47 celeron55     VanessaE: screenshots of folder contents?
16:48 VanessaE      celeron55: screenshots showing various texture packs with my fonts riding on top.
16:48 VanessaE      but your point stands.
16:48 VanessaE      I guess in the past most people just dumped everything into textures/all
16:49 celeron55     most builds of minetest also use freetype so that font isn't applicable to those at all
16:49 VanessaE      PilzAdam: "she wants to retain the second level above texture packs" you mean :P
16:49 VanessaE      celeron55: yeah I know, because putting freetype support in breaks the use of bitmap fonts entirely
16:49 VanessaE      which is, frankly, stupid
16:49 celeron55     it doesn't break it
16:49 celeron55     it replaces it
16:49 VanessaE      yes, it does.
16:50 VanessaE      same difference.
16:50 VanessaE      it worked before, that feature gets added, it stops working.
16:50 VanessaE      it makes for fonts that are less readable, or fancy, or whatever, than they were before.
16:50 celeron55     yeah and supports hundreds of times more characters
16:51 VanessaE      true
16:51 celeron55     of course you are happy as you use the language ascii was made for
16:51 celeron55     but it's quite selfish
16:51 VanessaE      I dare say 80% of the minetest community uses English :P
16:51 Calinou       french works fine with bitmap fonts too, but freetype still supports more chars :P
16:52 celeron55     i would be happy if it did support both though; but i don't consider that even worth the effort of looking whether it would be possible
16:52 celeron55     (even more as i did nothing of the ttf thing)
16:53 celeron55     (you'd have to bug thexyz, who won't do anything)
16:53 VanessaE      not bugging anyone anyways, I just want the alternate level put back :P
16:54 celeron55     we need to sum up the time of every minetest user wondering why textures load so slowly and compare it to the time that VanessaE uses to patch her version
16:54 celeron55     8)
16:54 VanessaE      heh
16:54 VanessaE      gotta run
16:54 VanessaE      back in a few hours
16:55 nore          I noticed that I get now textures loading far faster with an hi-res TP...
16:55 nore          since they aren't sent
16:55 PilzAdam      nore, wut?
16:56 nore          before, with textures in textures/all, using an hi-res TP caused a very looooong load time
16:56 nore          but now, since the server does not send them, it is far faster
16:56 celeron55     clear the cache directory and try again
16:57 celeron55     and enjoy the load time for the second time
16:57 nore          it is not because of the cache directory, since I just copied HavenTP from a folder outside Minetest folder to my textures/ folder
16:57 nore          before that, nothing
16:58 celeron55     in any case there hasn't been anything made that would affect that
16:58 celeron55     as far as i know
16:58 PilzAdam      celeron55, we could change main.cpp::1396 to read font_path from minetest.conf too, then fonts would be more consistent
16:58 nore          in fact, it would work same before if we only changed texture_path instead of putting textures in textures/all
16:59 nore          but I only discovered that today
16:59 nore          PilzAdam: looks a good idea
17:00 celeron55     PilzAdam: well it's done for TTF; why not
17:00 celeron55     that'll get VanessaE the font
17:00 BlockMen      why we have a news topic for tp over menu but none for the new menu itself??
17:01 BlockMen      *-?
17:01 nore          BlockMen: no idea
17:01 BlockMen      PilzAdam, you should make a topic on that :P
17:02 PilzAdam      celeron55, should we move fontlucida.png to fonts/ too?
17:03 celeron55     maybe the font image should be renamed at the same time
17:03 thexyz        we should just drop bitmap fonts support
17:03 Calinou       thexyz: TTF fonts cannot have an outline right now
17:03 Calinou       they are usable, but hard to read behind a bright background
17:03 celeron55     thexyz: can you get VanessaE the outline with TTF?
17:04 celeron55     that's really the only problem with it, as far as i can see
17:05 thexyz        i think we should have a shadow, not outline
17:05 thexyz        just duplicate and make it black
17:05 celeron55     well a shadow is a million times easier to make at least
17:05 PilzAdam      https://gist.github.com/PilzAdam/6151023
17:05 thexyz        yeah
17:05 celeron55     altough you can make a relatively good outline by drawing shadows in four directions
17:05 thexyz        but what's the point of outline when we can have shadows?
17:06 Calinou       some people prefer outline, some prefer shadows
17:06 celeron55     what's the point of fonts on television having outlines and not shadows?
17:06 Calinou       generally, outline is better for readability
17:06 celeron55     the point is that it's even more readable
17:06 Calinou       that's why memes use outlines, celeron55 ;)
17:08 thexyz        PilzAdam: how about we make it a run time setting, not compile time?
17:10 celeron55     can the TTF/bitmap selection be made runtime?
17:11 celeron55     PilzAdam: you still left the other bitmap font to be in textures though
17:29 BlockMen      is #858 ready to merge?
17:29 BlockMen      *-ready + good
17:31 sfan5         yes
17:35 BlockMen      and why isnt it merged then? :P
17:36 sfan5         dunno
17:38 BlockMen      other devs? ^
17:39 PilzAdam      celeron55, thexyz, should I leave the option to compile without freetype?
17:40 thexyz        I vote for no
17:41 thexyz        because why would you compile without freetype?
17:44 BlockMen      thexyz -> https://forum.minetest.net/viewtopic.php?pid=103074#p103074
17:44 BlockMen      could you fix that?
17:45 BlockMen      its impossible to open http://wiki.minetest.net/FAQ
17:45 celeron55     well two reasons: 1) fonts provided by freetype aren't good, or 2) you don't have freetype
17:46 thexyz        for now we don't plan to drop bitmap fonts if I got it correctly
17:47 thexyz        and 2) well, you're fucked
17:47 PilzAdam      https://gist.github.com/PilzAdam/76c048704250a0dd919d
17:48 thexyz        BlockMen: fixed
17:48 thexyz        clear cache if it doesn't work
17:50 thexyz        PilzAdam: it's probably not a very good idea to expose private mail, don't you think?
17:50 sfan5         PilzAdam: he also created a forum topic about that
17:50 BlockMen      thexyz, thanks. you answer casimir?
17:50 celeron55     PilzAdam: set those to be collected by a spam filter
17:51 thexyz        uh, sure, it's too hard to point them to wiki/docs
17:51 celeron55     thexyz: there was a time when we cared about exposing the "private" mails of those people; not anymore
17:53 thexyz        okay
17:53 thexyz        I still think we shouldn't but I cannot force it
17:53 sfan5         celeron55: either disable https for minetest.net or do the certificate stuff right
17:53 sfan5         wait
17:53 celeron55     sfan5: what
17:53 sfan5         dep
17:54 sfan5         forget what I just said
17:54 thexyz        oh there's https @ minetest.net
17:54 thexyz        cool
17:54 thexyz        too bad it doesn't work
17:57 celeron55     could your certificate be used there?
17:58 thexyz        yeah, but i'm not going to give you private key :P
17:58 PilzAdam      https://gist.github.com/PilzAdam/6151207
18:00 BlockMen      PilzAdam, celeron55, thexyz #858 ?
18:03 celeron55     PilzAdam: that will cause memory leak or segfault if freetype is changed when running 8)
18:03 celeron55     PilzAdam: if(font) font->drop();
18:03 celeron55     and font = NULL;
18:03 celeron55     (in the beginning)
18:03 celeron55     umm
18:03 PilzAdam      BlockMen, are there other files that could profit from the wstrgettext() cleanup?
18:03 celeron55     actually no; that'll not work
18:03 celeron55     well you know what the problem is anyway
18:04 PilzAdam      celeron55, I wonder how many people try to change that at runtime
18:04 celeron55     collect the setting at the beginning of that file
18:04 celeron55     into a boolean
18:04 celeron55     it's simply stupid to not do it properly
18:05 BlockMen      PilzAdam, AFAIK not. every else it is given directly as wchar_t* to irrlicht
18:06 celeron55     PilzAdam: alternatively you could ->grab() the font gotten from getFont() :P
18:06 celeron55     then you can drop it
18:06 celeron55     (at least should be able to, unless irrlicht plays some tricks)
18:06 PilzAdam      I just save it in a boolean
18:09 BlockMen      is it ok when the inventory looks like this when nothing is defined(like background, etc)
18:09 BlockMen      or should i remove in that case the full screen overlay that it looks like now?
18:10 BlockMen      oups..forgot the img link -> http://i.imgur.com/4PiQKHu.png
18:10 celeron55     well, sounds like a sane default
18:11 celeron55     is there some written docs about this?
18:11 PilzAdam      celeron55, https://gist.github.com/PilzAdam/6151207
18:11 celeron55     i'm not knowledgeable at all what you're doing
18:12 celeron55     for example are the styles per-formspec or per-game?
18:14 BlockMen      no, every formspec can be set different (if wanted)
18:14 BlockMen      by extending (and adding) the defining strings like list[....]
18:14 celeron55     PilzAdam: i think it's good now
18:19 PilzAdam      https://github.com/PilzAdam/minetest/commit/bf48133be9b8ec117d47bb31d207f6292d411c03 are the comments in minetest.conf.example good like that?
18:20 Exio4         you should add examples for the "bitmap fonts" too, i think?
18:21 PilzAdam      why?
18:22 Exio4         for completeness?
18:22 Exio4         it is more like why not?, it is a two lines changed for "completing examples"?
18:22 thexyz        why don't we drop freetype cmake params and make it required?
18:23 PilzAdam      well, dunno
18:23 PilzAdam      celeron55 seems to be against it
18:23 celeron55     why do that? it's just about hating users and doesn't bring any convenience to anyone
18:24 celeron55     just hinders some who want to throw a quick build on windows
18:24 thexyz        those #ifdefs are everywhere
18:25 Calinou       still requires -DENABLE_FREETYPE=1 (if it isn't default already) on linux
18:26 celeron55     freetype should be changed to be used by default if it's available
18:26 PilzAdam      btw, is anyone else annoyed by all these warnings from src/cguittfont/irrUString.h?
18:26 Calinou       +1, same for gettext
18:27 celeron55     i'm not annoyed because... apparently i have been too lazy to set -DENABLE_FREETYPE=1 again after some cmake reset
18:27 PilzAdam      its like 80 lines of warnings from this one file
18:31 PilzAdam      anyway, should I push that freetype commit now?
18:32 celeron55     there's a bug in it actually
18:33 celeron55     you don't initialize m_use_freetype at all
18:33 celeron55     in the case of no freetype being built in
18:34 PilzAdam      yea, I just declare it #if USE_FREETYPE
18:34 celeron55     thexyz: i consider these #ifdefs to be bad encapsulation; it wouldn't be hard to make a small wrapper header that would get the font in an opaque way
18:35 celeron55     humm... well you declare it always, but apparently you don't touch it always
18:37 celeron55     anyway, there's the almost exact same code copypasta'ed in two places
18:37 celeron55     this isn't good code by any standards 8)
18:38 celeron55     anyway just commit it now, i'll maybe rework it a bit
18:40 PilzAdam      done
18:42 celeron55     this is the latest version of the max_visitor_players thing... not sure if i like it http://paste.dy.fi/HcT
18:43 celeron55     it's not functional as-is because the only way it can work in practice is with a mod that tracks player's playtime and automatically sets the privilege after a while
18:43 celeron55     maybe that functionality should be put in builtin
18:43 celeron55     it's like 10 lines of code anyway
18:44 celeron55     but the whole usefulness of this isn't validated
18:44 celeron55     where's the #minetest-serveradmins channel :P
18:46 PilzAdam      its joined together with #minetest-mods
18:46 PilzAdam      #minetest-bots #minetest and #minetest-offtopic in #minetest
18:50 PilzAdam      858 seems good, anyone against merging it?
18:51 PilzAdam      "seems good" as in "doesnt break anything in Linux"
19:01 BlockMen      what is better for setting <color> in formspecs definition? as argb (255,255,255,255) or as combination of alpha and hexadecimal (20,#303030) ?
19:05 RealBadAngel  as i am digging with shaders vector4 representation is better for me
19:07 BlockMen      so eg "size[3,5;128,0,0,0]" ?
19:07 PilzAdam      other formspec parts use hexadecimal
19:07 RealBadAngel  im not pushing it, just mentioned that in shaders theres rbga form
19:08 RealBadAngel  *rgba
19:09 RealBadAngel  all the color calculations are made on vector3 (rgb) or 4 (rgba)
19:09 BlockMen      hmm, ic. but since the definition of video::Scolor (a,r,g,b) is i would prefer argb (easier to read then)
19:10 PilzAdam      hexadecimal is perfectly readable
19:10 RealBadAngel  no problem with hex for me
19:11 BlockMen      ok, then "size[3,5;128,#000000]" ?
19:11 RealBadAngel  shaders code is pretty well isolated, it can have different rules
19:11 celeron55     umm
19:12 celeron55     i wouldn't mix an integer value and a "web color" like that
19:13 celeron55     i'd use "size[3,5;128,0,0,0]"
19:13 PilzAdam      why not add alpha to the hex value?
19:13 BlockMen      i vote for argb too
19:14 celeron55     PilzAdam: nobody does that
19:14 RealBadAngel  8 chars hex?
19:15 PilzAdam      everything else in formspec and HUD uses hex
19:15 celeron55     in CSS stylesheets colors are either #ffffff or if one needs alpha, they are rgba(255,255,255,0.5) 8)
19:15 PilzAdam      I dont like the inconsistency
19:16 celeron55     i guess we could use a 4-byte hex
19:16 RealBadAngel  "size[3,5;#80000000]" then?
19:17 celeron55     uhg... no, it's not readable at all
19:17 celeron55     we can't do that
19:17 BlockMen      argb has another pro, everyone can read it easily from every graphic program
19:18 RealBadAngel  thats a 4 byte long hex...
19:18 BlockMen      without having any experience with hexcolors
19:18 celeron55     i'm voting primarily for "size[3,5;128,0,0,0]", secondarily for "size[3,5;128,#000000]"
19:19 RealBadAngel  so i do vote for all the colors and alpha separated with comas
19:19 celeron55     (altough i don't really like it that a size statement contains a color; i think there should be a separate color statement for setting colors)
19:19 Jordach       ARGB is easy to read, GIMP even puts colour in it (provided you look in the colour picker)
19:19 PilzAdam      size[3,5;128,#000000] is at least consistant with other stuff
19:19 RealBadAngel  It is?
19:19 proller       faster freeze-melt fo top water layer https://github.com/proller/minetest/commit/868df0eedd092aa511c034cdf31eecc4411628e4
19:20 BlockMen      celeron55, i wasnt sure yet what would be best, because the colors[] would get larger and larger
19:20 BlockMen      so setting a few things on the elements itself isnt that bad
19:20 RealBadAngel  if one of those propositions is already used in code, then yes
19:20 PilzAdam      proller, thats exactly why you should do that in a fork
19:20 PilzAdam      tweak things until they are useable and the commit all at once
19:21 proller       they is useable
19:21 proller       but i making it better
19:21 PilzAdam      you basically repeating the mistake of finite liquids
19:21 celeron55     BlockMen: do you have an example formspec to show what it looks like currently?
19:21 celeron55     BlockMen: the text
19:21 RealBadAngel  PilzAdam, nobody ever commited code that was perfect from at the start
19:22 proller       PilzAdam, finite liquids is mistake only for you
19:22 RealBadAngel  every code can be changed
19:22 BlockMen      sure, one sec
19:22 RealBadAngel  technic gets rewrites all the time
19:23 celeron55     it's a different thing to work on an engine than to work on a game or am od
19:23 celeron55     mod*
19:24 celeron55     you need different priorities
19:24 celeron55     BlockMen: this second is rather long 8)
19:25 PilzAdam      he is probably moving really fast compared to you
19:25 BlockMen      i have to make it readable, the code is horrible atm
19:25 BlockMen      you guys would cancel the whole thing :D
19:25 celeron55     doesn't matter
19:25 celeron55     just throw it somewhere, we know how to format code
19:26 PilzAdam      BlockMen, protip(tm): write readable code from the first line on, it helps yourself to structure your mind
19:26 celeron55     also i know you can't format code from your PHP stuff
19:26 celeron55     8D
19:27 BlockMen      fine, there http://pastie.org/private/8yrssewmukledwbo11h9ew
19:27 BlockMen      but there are now missing the (for testing) hard coded things
19:28 BlockMen      this are most parts of what planned -> http://pastie.org/private/8ea1nup9c3iqwgj55hma
19:29 celeron55     hmm so yeah
19:29 celeron55     don't put a color in size[]
19:29 BlockMen      celeron55, why?
19:29 celeron55     because it's a size
19:29 celeron55     what color is that? background color?
19:29 PilzAdam      I really like that background[] element, I requested RBA to do it like this from the beginning on
19:29 celeron55     just call it what it is
19:29 BlockMen      c55, yes bgcolor
19:30 celeron55     bgcolor[20,255,8,8] size[8,8.5] itembg[...] and so on
19:30 celeron55     don't use camelcase in there
19:31 celeron55     just lowercase
19:31 BlockMen      ok, gonna change that
19:32 BlockMen      but as i said to VanessaE already, i dont think it will be done for 0.4.8
19:32 celeron55     most likely no; don't aim for that
19:32 PilzAdam      dont care about releases
19:32 celeron55     the only thing that is important is that it is well thought out
19:33 celeron55     (and that it works)
19:33 BlockMen      ok
19:34 celeron55     one thing i'm not sure about is whether the "variable setter"-like elements should always be global or whether they could affect only the element coming after them or something
19:35 BlockMen      but is it ok to add "<itemrect_hover>;]" to list for setting hover =0||1 ?
19:35 celeron55     how are you going to do that when there are two possible lengths for it already
19:36 celeron55     oh well, just make an even longer version i guess
19:36 BlockMen      yes, like at pressed_button img support for img_button ()
19:36 celeron55     (make sure that old formspecs work reasonably still)
19:37 celeron55     what would <itemrect_hover> do though?
19:37 BlockMen      is set to true you have hover effect on slot
19:37 celeron55     is it something to distinguish clickable items and non-clickable items
19:38 PilzAdam      celeron55, it brightens the rect if hoovering
19:38 BlockMen      PA explained better :P
19:38 RealBadAngel  highlight effect, yes?
19:38 celeron55     but the question is why it shouldn't be for every item
19:38 BlockMen      its like at mc, if your mouse pointer is hitting the rect its highlighted
19:39 PilzAdam      I think it should be always on
19:39 BlockMen      like now already included when draging item
19:39 celeron55     why are you adding a parameter for it then
19:39 celeron55     oh that was pilzadam
19:39 celeron55     oh whatever, it can be tweaked afterwards
19:40 BlockMen      if you say it should be default i will kick param for
19:40 BlockMen      i prefer always on, but i had in mind that many want keep possibility of having old elements
19:42 RealBadAngel  i dont think that highlighting is really needed for each and every element
19:42 RealBadAngel  but thats rather matter of taste
19:42 BlockMen      RBA, just for item slots
19:43 RealBadAngel  but on the other hand UI uses item image buttons, that could be highlighted
19:44 celeron55     just keep in mind that not everything needs an option; that might need
19:45 BlockMen      ok, i will rebase my work and then see what is needed to do then
19:46 RealBadAngel  btw, #807 can be merged by now
19:46 celeron55     (in UI it's kind of a fight between consistency and giving room for clever tricks)
19:47 BlockMen      celeron55, and i might giving to much room now because we have that less atm
19:47 PilzAdam      RealBadAngel, #807 is an issue, not a pull request
19:47 PilzAdam      you probably mean #854
19:48 RealBadAngel  ooops, yes 854 that fixes the issue 807
19:48 RealBadAngel  i merged that change into my rewrite, but theres no reason that it cannot go into right now
19:50 BlockMen      works fine for me too
19:52 PilzAdam      proller, nice that you just push things without any other dev agreeing on it
19:52 RealBadAngel  i was fighting today with wrench mod and it seems that itemstack:get_metadata() stopped to work
19:53 proller       PilzAdam, everyone must agree fixing broken js?
19:53 PilzAdam      no that one: https://github.com/minetest/minetest/commit/714a8cef38ed46ddd6ef7c2689c93db2704969eb
19:53 PilzAdam      +,
19:53 celeron55     should we revoke proller's push rights
19:53 celeron55     this is quite far from what it should be
19:53 RealBadAngel  have some1 was workinh lately on anything related to itemstacks or deserialize?
19:54 celeron55     proller: no, but especially if there are people on this channel, you should pastebin the patch first and link here
19:54 celeron55     and not just go and push it
19:54 RealBadAngel  proller, you shall just ask anybody
19:54 celeron55     or at least note that you are going to push something
19:55 celeron55     then wait for 10 minutes and then push, or if someone asks for details, provide them
19:55 celeron55     if you need studying material for that, there's plenty in the previous days' irc log
19:55 proller       i post link to commit at :19 min, and this is small and trivial change
19:56 RealBadAngel  proller, we need to follow the rules
19:56 celeron55     proller: you can't merge based on the fact that nobody says "don't merge"
19:56 RealBadAngel  no matter how many bits we change
19:56 celeron55     you need to beg until someone says "do merge"
19:56 celeron55     that's what i do too
19:57 proller       ok, next time will wait
19:57 celeron55     unless there really is nobody giving a shit
19:57 celeron55     (in a day or so)
19:57 celeron55     of course that masterserver fix doesn't require waiting a day, but it requires noting here minutes before pushing
19:58 RealBadAngel  proller, i was so impatient as you before, but i learned that sometimes its even better to wait and test the change for a while
19:59 RealBadAngel  or even let it wait on the shelf for a moment
19:59 PilzAdam      RealBadAngel, whats the progress on animated bumpmaps?
19:59 RealBadAngel  im preparing now texture packs that support it
20:00 RealBadAngel  i will let folks test the changes for a week or two, then will pull rewrite against mainstream
20:01 celeron55     "sometimes its even better to wait"
20:01 RealBadAngel  maybe meanwhile i will find less hackish way to pass tangent vectors to shaders
20:01 celeron55     more like "always"
20:02 RealBadAngel  hehe, yup
20:02 sfan5         CMake Error at /home/stefan/mt-build/build/c55/minetest-minetest-44eb016/build/cmake_install.cmake:67 (FILE):
20:02 sfan5         file INSTALL cannot find
20:02 sfan5         "/home/stefan/mt-build/build/c55/minetest-minetest-44eb016/textures/all/textures_here.txt".
20:02 sfan5         ^ latest git
20:02 PilzAdam      oh
20:03 RealBadAngel  ^^ one of the cases ;)
20:05 celeron55     also if nobody is interested of a patch, one should wait for a silent moment and explain what it does; how much it sucks currently and how little it sucks after the patch; what kind of performance measurements one made; tell people what to try to do to see a problem
20:06 celeron55     it's really some kind of an interestocracy
20:06 PilzAdam      sfan5, https://gist.github.com/PilzAdam/6151733
20:07 celeron55     (there are plenty of silent moments in a day)
20:13 hmmmm         ugh man
20:13 hmmmm         i really know what people were saying about the lua main menu being slow now
20:13 hmmmm         there's a very noticable delay
20:13 celeron55     when it can't connect to public server?
20:13 hmmmm         no, i'm talking about from the time the window for minetest is created and the gui starts rendering
20:14 celeron55     oh, is that because of the menu?
20:14 hmmmm         is it?
20:14 celeron55     i have no idea
20:14 hmmmm         dunno, but it's two seconds long for me
20:14 PilzAdam      sfan5, does that patch work?
20:14 hmmmm         actually come to think of it, it wasn't like this the last time i loaded minetest up
20:14 hmmmm         this is clearly a regression
20:15 celeron55     what if you load it up again
20:15 PilzAdam      it directly downloads all the stuff for the serverlist and the modmanager, no matter if its needed or not
20:15 celeron55     is it just because you disk cache wasn't caching stuff
20:15 hmmmm         nope, has nothing to do with caching
20:15 celeron55     +r
20:15 hmmmm         I suppose pilzadam is right
20:16 celeron55     (the OS cache i mean; that could be it if it was loading textures or something)
20:16 hmmmm         why does it do this?  whose idea was it?
20:16 celeron55     sapier's obviously
20:16 celeron55     you can blame him for everything in the menu
20:16 celeron55     (almost)
20:16 hmmmm         are you *sure*?
20:16 PilzAdam      I know now why he doesnt show up in this channel anymore ;-)
20:17 hmmmm         what if he's busy at work
20:17 PilzAdam      hmmmm, previously it downloaded the serverlist if you actually click on "Public list"
20:17 hmmmm         what was the rationale for this change?  why don't we just change it back?
20:17 hmmmm         i'm sorry but a 2 second delay is simply unacceptable
20:17 celeron55     previously it was gloriously made by proller i think
20:17 hmmmm         don't really understand why this wasn't fixed within the hour it was committed
20:18 hmmmm         pf there's nothing glorious about waiting 2 seconds for the gui to start rendering
20:18 PilzAdam      many (including me) already complained about it, but sapier doesnt seem to care about it
20:18 hmmmm         are your fingers broken!?
20:18 hmmmm         you have commit access right?
20:18 celeron55     hmmmm: proller made it so that it didn't fetch stuff if it didn't need them
20:18 hmmmm         i don't get how anybody could've let this go
20:19 hmmmm         celeron55, why didn't he fix it then?
20:19 celeron55     what?
20:19 hmmmm         he should've fixed it
20:19 celeron55     do you understand there have been two versions of the menu
20:19 celeron55     the C++ one, and sapier's Lua one
20:19 hmmmm         ?? yes, and I used the lua menu plenty of times
20:19 hmmmm         it started doing this now
20:19 celeron55     the original server list was implemented in the C++ one
20:19 PilzAdam      hmmmm, you were the one who merged it, its there since the mainmenu was merged
20:20 celeron55     in such a way that didn't piss you off
20:20 hmmmm         what a
20:20 hmmmm         am i not making sense or something?
20:20 hmmmm         I said I just noticed this happening right now
20:20 celeron55     go get coffee or something
20:20 hmmmm         this isn't caused by the lua main menu or anything
20:20 RealBadAngel  if it is noticed, there shall be issue posted
20:20 hmmmm         and definitely not sapier's fault
20:21 Jordach       then what caused it then
20:21 hmmmm         the problem is that now the serverlist is always fetched on startup
20:21 hmmmm         has nothing to do with a C++ or lua version of the menu
20:21 hmmmm         so the last person to dick with this is at fault
20:21 RealBadAngel  btw, since there are more folks online atm, who was messing with itemstacks or deserialize things? :)
20:22 RealBadAngel  itemstack:get_metadata() is not working anymore
20:22 RealBadAngel  i had to use workarounds to get it workin, convert item to table then access meta
20:22 kaeza         hmmmm, you have a good net connection for a 2 sec delay; I have 10 secs or so with my crappy inet
20:23 kaeza         and with no net connection, it waits until timeout (30 or 60 secs, dunno)
20:24 hmmmm         and celeron stopped talking
20:24 celeron55     i have nothing to talk
20:24 celeron55     anyway, measurements: i have a 1s delay until the window pops up, and then 1s delay until it starts rendering
20:24 celeron55     quite precisely
20:25 hmmmm         kaeza, this is definitely broken behavior and it is 100% unacceptable, what i just don't understand is why nobody fixed this after they first started noticing it
20:25 celeron55     except that now it was slower
20:25 PilzAdam      RealBadAngel, how about looking at the commit log?
20:25 kaeza         hmmmm, I started noticing this once I decided to pull after the lua menu changes
20:26 kaeza         I guessed it would be slow Lua initialization or something
20:26 RealBadAngel  PilzAdam, i will have to eventually, since its really broken
20:27 PilzAdam      hmmmm, apparently my first thought was wrong; I thought that the list is passed to the menu once and the cached there so there is no delay when clicking on the Public Serverlist
20:28 PilzAdam      so its not a design issue by sapier
20:28 hmmmm         I knew that
20:28 hmmmm         i have no clue why you were blaming him in the first place
20:28 PilzAdam      I just explained why
20:29 sfan5         PilzAdam: it should work..
20:29 hmmmm         that doesn't explain why you'd think it was sapier's doing, that just explains how you thought it worked
20:29 hmmmm         but we had the lua main menu for actually quite a while now
20:30 hmmmm         you knew this wasn't acting up before
20:38 PilzAdam      comment out line 35 of builtin/modstore.lua
20:39 hmmmm         well hold on i want to find what commit it was added in
20:39 hmmmm         well, i guess it really was sapier's fault
20:40 hmmmm         it's a little faster but there's still a delay
20:40 PilzAdam      its in the first Lua menu commit
20:40 PilzAdam      it doesnt download stuff now, though
20:40 PilzAdam      and the modmanger is not usable anymore
20:41 hmmmm         that doesn't fix it totally
20:41 hmmmm         wasn't there some commit that made modmanager and something else enabled by default?
20:42 PilzAdam      no, its also called if the modmanager is disabled
20:42 PilzAdam      moving modstore.update_modlist() to modmgr.lua:554 works fine
20:42 hmmmm         it doesn't help too much
20:43 sfan5         PilzAdam: th
20:43 sfan5         *the patch does work
20:43 hmmmm         i clearly recall the lua main menu not doing this before i took a break and i rebase to upstream and then this shit
20:44 PilzAdam      sfan5, pushed
20:45 PilzAdam      hmmmm, but it definitely speeds it up for me
20:46 PilzAdam      do you use LuaJIT?
20:46 hmmmm         yes
20:46 hmmmm         well, hrmm, it happened before 32001d1e2b6659a77eab024015da85079b2252c1
20:46 hmmmm         still looking
20:47 hmmmm         it doesn't help that i need to recompile everything when i do a hard reset
20:53 hmmmm         okay this is odd
20:53 hmmmm         have i just become more sensitive to delays like that?
20:53 hmmmm         i guess i remembered wrong
20:59 hmmmm         hrm, the delay has been in there for quite some time, but it seems to have gotten worse at some point.
21:14 celeron55     http://dev.minetest.net/Git_Guidelines#Rule_1_in_practice
21:15 celeron55     not sure if that is just useless blah blah but maybe it will help someone
21:15 celeron55     i've learned to know it's impossible know what is useful 8)
21:23 RealBadAngel  http://i.imgur.com/Lfjjd4Y.png
21:23 RealBadAngel  cool, i like the feature :)
21:28 celeron55     proller: see this http://dev.minetest.net/Git_Guidelines#Rule_1_in_practice
21:28 RealBadAngel  ^^ agree with that
21:30 proller       ok
21:38 proller       using rule 2: better masterserver: https://github.com/proller/minetest/commit/6ba8e9dcd04d5a279641a76f9740477b9e7d8869
21:41 celeron55     what does that even do
21:42 celeron55     let's code some perl: {$_->[~~:'start'
21:43 celeron55     if you think you might have more of similarly sized commits coming in the next day or two, you should combine them, but if that's the only one, just go ahead
21:44 celeron55     (nobody understands the master server code itself anyway so people will only care if it clutters the commit log)
21:47 proller       some not important params sended to masterserver only on server start
21:48 proller       this change gets this params from old saved announce
21:48 proller       ~~: <-- syntax error
21:49 proller       no commits on next few weeks, if no bugs will found
21:57 ShadowNinja   Does anyone here have IPv6?
21:58 PilzAdam      I use it in singleplayer
21:58 ShadowNinja   proller: Exactly every five minutes I get this, but my server stays on the list: ERROR[ServerThread]: Serverlist at url servers.minetest.net error (Timeout was reached)
21:59 proller       if you want announce from ipv6-only host - use serverlist_url = v6.servers.minetest.net
22:00 ShadowNinja   It is a little old, updating it... And I an on a IPv4-only host, my IPv6 question is unrelated.
22:00 proller       ShadowNinja, seems you dont recieve responce because timeout is small (1s), but server list recieves your message
22:01 PilzAdam      proller, does it use the curl_timeout setting from minetest.conf?
22:01 ShadowNinja   Hmmm, maybe the timeout should be 5-10s.
22:01 proller       PilzAdam, no
22:02 proller       minimal timeout for not blocking anything
22:02 proller       and server does not need in responce
22:03 ShadowNinja   proller: It shoudn't block for even a second. Can't you ask for a response and check it later?
22:03 PilzAdam      why writing to errorstream then?
22:04 proller       need to read curl doc about async
22:04 proller       PilzAdam, because curl gets error
22:05 kahrl         how about adding a background thread for HTTP requests
22:05 kahrl         it would be used by the main menu as well as the server announce thing, and perhaps the media loader
22:06 ShadowNinja   Jeija's Lua curl pull used something like that.
22:08 ShadowNinja   Comments? https://github.com/ShadowNinja/minetest/tree/bind_address
22:08 proller       kahrl, best solution, but not trivial
22:09 kahrl         proller: basically it's MediaFetchThread but can be used for other things
22:11 proller       in announce no need to parse and do anything with responce -> can be much simpler
22:12 proller       ... yet another TODO
22:12 kahrl         without the thread you still have to wait for the server to accept your connection (or timeout)
22:14 proller       now its ~0.2s every 5 minutes
22:16 PilzAdam      thats one minute per day
22:19 ShadowNinja   The configure button seems broken, can somebody else confirm this?
22:19 PilzAdam      no
22:22 ShadowNinja   http://pastebin.ubuntu.com/5949136/
22:24 PilzAdam      yea, that happens if world.mt doesnt contain any mods
22:29 ShadowNinja   Hmmm, a newline between the gameid and the mods caused it. Either way it is a bug.
22:30 * VanessaE    is back
22:35 ShadowNinja   BBL. Take a look at this and maybe test it with IPv6 while I am away. https://github.com/ShadowNinja/minetest/tree/bind_address
22:56 VanessaE      so, how do I get my custom font working again?
22:56 VanessaE      I'm not sure I quite understood how that's been made to work
22:57 PilzAdam      VanessaE, point font_path and mono_font_path in minetest.conf to it
22:57 PilzAdam      and freetype = false
22:58 VanessaE      PilzAdam: oh real nice
22:58 VanessaE      http://pastebin.com/4gG9tyxg
22:58 PilzAdam      and its recommended to put the bitmaps in fonts/, where the truetype fonts are too
22:59 VanessaE      THERE we go
22:59 VanessaE      that crash ^^^^ happens if you point the font PATH to a PATH and not a file :P
23:00 VanessaE      point them to the actual files and freetype=false and it works properly.
23:00 PilzAdam      you actually dont need freetype=false if you havent compiled it in
23:00 VanessaE      true
23:00 VanessaE      well it's there whenever freetype goes "always on"
23:00 VanessaE      er, always-in
23:37 BlockMen      PilzAdam, https://github.com/minetest/minetest/pull/859
23:37 VanessaE      also, regardless of the selected texture pack, game overlays in the pack aren't being used anymore (only what comes with a game)
23:39 VanessaE      the overlay for the "client" tab works, though.  just not the ones that should display in the singleplayer tab
23:39 kahrl         VanessaE: sapier warned about that... did nobody fix it?
23:39 VanessaE      kahrl: apparently not.
23:39 VanessaE      I could swear it was working before today, but maybe not
23:44 kahrl         preliminary design for an async http fetch interface: https://gist.github.com/kahrl/6152525
23:44 VanessaE      kahrl: did you see sapier's attempt at that?
23:44 kahrl         I did not
23:44 VanessaE      https://github.com/sapier/minetest/commit/c900051f68fa01cc8e58034f934eb009abbe2885
23:45 VanessaE      you might find that ^^^ of some use
23:46 kahrl         ah, what I'm trying to do is intended to be more general
23:46 kahrl         but that will definitely be useful when it comes to the main menu integration
23:47 PilzAdam      kahrl, btw, what about your animation crack fix?
23:47 VanessaE      good.  save a bit of potential duplicated work :)
23:47 kahrl         my thinking is: why replicate the basic structure of MediaFetchThread a couple times
23:47 kahrl         PilzAdam: you can merge it if it works for you
23:48 PilzAdam      will do that
23:49 kahrl         oh, that also reminds me of that jthread discussion from a couple days ago
23:49 kahrl         somehow nothing resulted out of it
23:50 Exio4         what happened with it?
23:50 PilzAdam      tile.cpp:1517:31: warning: comparison between signed and unsigned integer expressions [-Wsign-compare]
23:50 kahrl         what is that line again? I can't apply the patch right now to check
23:50 PilzAdam      Ill make crack_count an u32
23:51 kahrl         sounds good
23:51 PilzAdam      it basically compares u32 progression with s32 crack_count
23:51 PilzAdam      but crack count is u32/u32
23:51 PilzAdam      so I guess its safe to make in an u32 too
23:51 kahrl         yep
23:54 kahrl         I wonder if I should make httpfetch_async use a thread pool or a single background thread with curl_multi