Minetest logo

IRC log for #minetest-dev, 2014-06-19

| Channels | #minetest-dev index | Today | | Google Search | Plaintext

All times shown according to UTC.

Time Nick Message
00:10 hmmmm [06:08 PM] <ShadowNinja> (((parts.size() >= 5) && (parts.size() <= 9)) && (parts.size() != 6)) < Unnecessary grouping.
00:10 hmmmm may be unnecessary but that doesn't make it wrong
00:10 hmmmm in fact grouping all operands of && and || is a rule in MISRA
00:11 OldCoder joined #minetest-dev
00:12 hmmmm you know paramat, I think it might be a neat idea to have mod-defined ridges like that in v7
00:13 hmmmm if the noise is between a certain range, make the resulting height constant, and anything above that range is equal to the noise value minus the range of the ridge
00:13 hmmmm but that doesn't work for cutting off only portions of map to be flat
00:13 hmmmm hmmm
00:14 VanessaE ~tell sapier you fixed a fairly large portion of my memory usage issue with #1359 - Creative server is now using right about 9GB at the client with HDX512 and all of the settings enabled as before, versus 13GB before under the same conditions.  FPS is still shit but that's my video card unable to handle the load.
00:14 VanessaE fuck.
00:14 VanessaE no shadowbot, no minetestbot.
00:15 VanessaE well make the above 9-10 GB now, versus 13+
00:16 VanessaE still an improvement.
00:17 hmmmm the intersection of two perlin maps should fix that problem so you can have unconnected blobs of flat land
00:17 hmmmm however i can't figure out how you'd make them appear at a random height
00:18 * paramat reads (was cooking)
00:25 paramat actually noisegrid does occasionally make isolated blobs of flat land, but rare
00:26 hmmmm perhaps it does, but this way you can have consistent blobs of flat land
00:31 paramat yes. i have a new unlimited tunnel generator https://forum.minetest.net/viewtopic.php?f=9&amp;t=9521 tunnel is the volume of intersection of 2 narrow bands of 3D noise
00:50 hmmmm wow, nice
00:50 hmmmm that's how caves should be IMO
00:50 hmmmm maybe with a smaller spread factor though idk
00:57 ShadowNinja hmmmm: Yes, but it didn't seem to help readability either.  That's why I mentioned it.
00:57 ShadowNinja But *shrug*.  It doesn't really matter.
01:13 paramat thanks thats cool you like it. now my laptop needs to cool ...
01:26 RealBadAngel ShadowNinja, putting extra code per texel in world for sake of pause is not a wise idea
01:28 RealBadAngel but when entering pause, we can just disable shaders in settings
01:29 RealBadAngel ofc, visual effect with some options enabled will look not so good
01:30 VanessaE why not just pass a pause flag to the shader?
01:30 VanessaE if not set, allow the shader timer to run so the animations move
01:30 RealBadAngel because thats extra code per each texel in the world
01:30 RealBadAngel it will slow down shaders
01:30 VanessaE damn
01:31 VanessaE can't stop the timer itself somehow?
01:31 VanessaE I mean, the shaders refer to a timer somewhere to run the animations
01:31 RealBadAngel im trying hard to eliminate any unnecesary code in shaders and now i shall add some weirdo? no way :)
01:31 VanessaE why not shut it off at its source?
01:32 VanessaE nono I mean before the shaders get the timer's value
01:32 RealBadAngel just disable it in settings
01:32 VanessaE wherever the value comes from to begin with, shut it off there
01:32 RealBadAngel on next meshupdate shaders will be disabled
01:32 Miner_48er joined #minetest-dev
01:37 RealBadAngel what for disable timers?
01:37 RealBadAngel to stop things from waving?
01:37 VanessaE if you disable the timers, the waving shaders have nothing to refer to
01:37 VanessaE so the animations should stop, no?
01:37 VanessaE it doesn't stop the GPU usage
01:37 VanessaE but it stops the visual movement
01:38 VanessaE which gives the impression that the game is paused.
01:38 RealBadAngel and nothin more but impression
01:38 VanessaE I think that's all ShadowNinja really wanted :P
01:38 VanessaE besides, better to give the impression of stopped-shaders than to let them wave when everything else has been paused, I think
01:39 VanessaE if you can't truly stop them without adding a performance hit when they are running.
01:39 RealBadAngel i can disable them at all temporarily
01:40 RealBadAngel just set the config setting and restore it when unpause
01:40 ShadowNinja RealBadAngel: That would cause a significant graphics difference though, and waving things would snap back to their nermal positions.
01:40 VanessaE yeah but that turns off a number of other effects
01:40 RealBadAngel yes
01:41 VanessaE just stop the timer at its source.  fake it.
01:41 RealBadAngel thats not what issue is about
01:41 RealBadAngel guy wants to calm down his pc
01:41 RealBadAngel lol
01:41 ShadowNinja Things should simply freeze.  How do moving things dicide where they should be?  A timer?
01:42 RealBadAngel a sec
01:42 RealBadAngel game.cpp circa #800
01:43 RealBadAngel wow, its #803, i have good memory ;)
01:44 RealBadAngel https://github.com/minetest/minetest/blob/master/src/game.cpp#L803
01:45 ShadowNinja RealBadAngel: Just set that to 0, or, even better, the last value.
01:45 RealBadAngel something like if (pause) animation_timer = 0; should do the trick
01:45 VanessaE didn't I suggest that 15 mins ago? :)
01:48 RealBadAngel how is the pause applied at all? some global variable?
01:52 ShadowNinja RealBadAngel: I think you have to check currentForm->pausesGame()
01:53 ShadowNinja Where currentForm is whatever formspec/window/menu is being currently displayed.
01:55 ShadowNinja RealBadAngel: Can't this https://github.com/minetest/minetest/blob/master/src/game.cpp#L813-L826 be replaced by simply the three calls with 0, 1, and 2 directly in the parameters and C++ type auto-conversion for whatever's available?
02:00 kaeza joined #minetest-dev
02:16 Megaf This still broken, it wont compile https://github.com/minetest/minetest/pull/1359
02:18 Megaf trying again
02:20 us`0gb joined #minetest-dev
02:20 Megaf compiled
02:20 Megaf Hey us`0gb
02:22 us`0gb Hey Megaf!
02:34 hmmmm hmmm
02:34 hmmmm what is sapier doing with the VOXELFLAGS?
02:34 kaeza joined #minetest-dev
02:35 hmmmm I don't remember exactly but I added in one of those not for performance reasons, but to solve the problem of that portion of the map being updated while another voxelmanipulator has that area loaded
02:36 hmmmm if he screws that up then no doubt you'll get the walled-in-cave problem again with more than one emerge thread
02:55 us`0gb joined #minetest-dev
04:20 diemartin joined #minetest-dev
04:34 Eater4 Does anyone here have control over the public server list?
04:34 Megaf xyz does
04:40 Eater4 Ok
04:51 kaeza joined #minetest-dev
05:39 Hunterz joined #minetest-dev
06:15 LemonLake joined #minetest-dev
06:24 grrk-bzzt joined #minetest-dev
06:50 Miner_48er joined #minetest-dev
07:02 alexxs joined #minetest-dev
07:29 ShadowNinja joined #minetest-dev
07:30 deltib joined #minetest-dev
07:51 RealBadAngel joined #minetest-dev
07:57 RealBadAngel well, https://github.com/minetest/minetest/commit/65a4630f3176feb0f4dd4db6ce12cd6d32e2b369
07:57 RealBadAngel it has broken all the labels
07:57 RealBadAngel sapier...
08:37 restcoser joined #minetest-dev
09:04 celeron55 03:31:47 < paramat> yes. i have a new unlimited tunnel generator
09:04 celeron55 03:50:52 <+hmmmm> that's how caves should be IMO
09:04 celeron55 this is how mapgen v5 does caves
09:07 Eater4 Good night
09:43 sapier joined #minetest-dev
09:48 sapier hmmm if I understand the code correct you added voxeflags to not have to copy the whole voxelmanip, BUT in order to use that feature you have to do a 3d loop through memory copying elements one by one. Of course there may be situations where not copying may be faster but for all typical situations I profiled this resulted in doing a double loop over all elements. In almost all situations there hasn't been a single not loaded element whe
09:49 sapier moreover memcpy is usually highly optimized by compiler for each architecture and way more fast then single asignment
09:51 sapier most likely if you have a voxelmanip with two parst far from each other old variant might be faster. Yet optimizing a hardly ever used usecase by adding penalty of more then 100% to common case isn't very usefull (imho)
09:51 VanessaE ...loaded element whe...?
09:51 VanessaE you got cut off.
09:51 sapier argh
09:51 sapier ... single not loaded element where actually time could've been saved
09:53 sapier Still those performance changes are suggestions only I can't be sure I didn't miss some case causing major issues so anyone having similar concerns please tell
10:36 proller joined #minetest-dev
10:38 vifino joined #minetest-dev
10:48 CraigyDavi_ joined #minetest-dev
10:59 celeron55 testing multithreaded map generation is kind of not viable because it's broken anyway
10:59 celeron55 for regressions in the flags, i mean
11:09 iqualfragile joined #minetest-dev
11:12 ImQ009 joined #minetest-dev
11:43 sapier how is multithreading related to it?
11:49 alexxss joined #minetest-dev
11:59 PenguinDad joined #minetest-dev
12:13 celeron55 sapier: because then it is generating multiple areas at the same time and has to merge them
12:13 sapier yes but it'd use different voxelmanips to do so wouldn't it?
12:14 sapier wait .. ok if one thread would generate (0,0,0) and (1000,1000,1000) within same voxelmanip that might be an issue
12:16 sapier but still considering the 100%+ overhead for regular case I'd suggest to fix the mapgen not to generate things randomly distributed around map in a single voxelmanip
12:16 Jordach joined #minetest-dev
12:21 Exio4 joined #minetest-dev
12:23 Anchakor_ joined #minetest-dev
12:31 SmugLeaf joined #minetest-dev
12:31 SmugLeaf joined #minetest-dev
12:36 sapier https://github.com/minetest/minetest/pull/1319 I'm gonna push this small fix in a few minutes
12:36 celeron55 sapier: it's not about performance; it's about consistency of results
12:37 celeron55 but i'm not very knowledgeable about this, you have to ask hmmmm
12:41 sapier I'll talk to him to make sure I didn't miss something
12:59 jin_xi joined #minetest-dev
13:13 smoke_fumus joined #minetest-dev
13:30 hmmmm joined #minetest-dev
13:49 SmugLeaf joined #minetest-dev
14:14 Eater4 joined #minetest-dev
14:32 RealBadAngel joined #minetest-dev
14:32 RealBadAngel hi sapier
14:33 lovehao joined #minetest-dev
14:33 RealBadAngel there are problems with bounding boxes for labels
14:33 RealBadAngel and propably irrlicht
14:33 RealBadAngel for me it is broken, all the labels are shifted down
14:34 RealBadAngel i do have irrlicht 1.7.2, vanessa has 1.8 and havent noticed any problems
14:34 zat joined #minetest-dev
14:34 kaeza joined #minetest-dev
14:35 RealBadAngel http://i.imgur.com/2OneR68.png
14:35 RealBadAngel it affects all the labels, ingame and in mainmenu
14:36 RealBadAngel getDimension propably returns something weird, when i revert it as it was before (+300) everything goes back to normal
14:39 Evolykane joined #minetest-dev
14:40 Megaf joined #minetest-dev
14:44 CraigyDavi Is this a known bug? https://cdn.mediacru.sh/PMlKepRZrd03.png My server seems to sometimes do this now, it did not happen on 0.4.9 stable
14:45 CraigyDavi There was no high cpu, high ram or anything at the time. It looks like it happened when someone timed out
15:11 sapier RealBadAngel: I've got 1.7.3 and don't have that issue ther ehas to be something different
15:12 sapier that change doesn't do anything to y position of labels and checkboxes it's very very strange those are affected for you
15:12 sapier the only thing where ypos is even touched is for vertlabels
15:13 RealBadAngel vertlabels and labels are both affected
15:13 sapier CraigyDavi: hard to tell that screenshot doesn't provide any information but it's beeing crashed
15:13 sapier RealBadAngel: as I said thats strange because ypos of those isn't even touched
15:13 RealBadAngel and im very suprised too, how the x change can make it shift down
15:14 sapier is it same for mainmenu?
15:14 RealBadAngel yes
15:14 sapier even more strange I doubt difference between irrlicht 1.7.2 and 1.7.1 can do this
15:14 RealBadAngel going to add now some debug code to see those values
15:14 sapier 1.7.3
15:15 sapier getDimension isn't used for y position too so any bug there couldn't affect labels and checkboxes
15:16 CraigyDavi Well the debug.txt gives no information either, but yeah. it crashes frequently like that
15:17 sapier what build do you use?
15:22 RealBadAngel looks like width values calculated are too small, and text is wrapped
15:22 RealBadAngel thats why it looks like shifted down
15:22 sapier text wrapping in labels?
15:22 sapier since when does this work?
15:24 sapier I never saw a label to be wrapped by now
15:24 RealBadAngel m_font->getDimension(wlabel.c_str()).Width + 2
15:24 RealBadAngel adding 2 is enough for texts to be correctly displayed
15:24 sapier maybe use the kerningwidth?
15:25 sapier then it'd be font specific
15:25 RealBadAngel i tried both default and freetype
15:25 RealBadAngel same behaviour
15:25 sapier freetype isn't a font ;-)
15:25 RealBadAngel i mean compiling with and without freetype
15:26 sapier what's causing labels to be wrapped for you?
15:26 sapier they're not supposed to autowrapp
15:26 RealBadAngel idk if theyre wrapped or whatever
15:26 RealBadAngel theyre shifted down
15:27 RealBadAngel adding at least 2 to width solves the issue
15:27 sapier that's strange they've been just cut for me ... no matter if kerningwidth doesn't work the 2 will be fine too
15:27 sapier just add a comment like "add some safety addon to avoid line wrapping on some irrlicht versions"
15:27 RealBadAngel also i can see vertlabels incorectly shown
15:28 sapier well vertlabels are a little bit changed as their positioning wasn't actually predictable before ... at least not by position
15:28 sapier in case that's an issue I'll just revert it back to 300 ... they will continue to be broken then
15:28 RealBadAngel http://i.imgur.com/zBp1gOu.png
15:29 sapier that's your autowrapping too
15:29 sapier there's no space between SIN and GLE for sure ;-)
15:29 RealBadAngel that i could see for ages already
15:29 RealBadAngel it was always broken
15:29 sapier I wonder why a N fits in there but a G doesn't
15:30 sapier other way round
15:30 sapier are you on windows or linux?
15:31 sapier hmm that's not a windows window deco
15:31 RealBadAngel adding 2 to height also caused vertlabels to be positioned correctly
15:31 sapier height?
15:32 sapier that's crazy
15:32 RealBadAngel yup
15:32 sapier it should help increasing the width not the height
15:32 RealBadAngel im on xubuntu 12.04
15:32 sapier I guess spending 2 additional pixels is still better then before, yet you should add a comment what those 2 are
15:33 ShadowNinja Python server list with Flask: http://sprunge.us/OHed?python  Much more readable, and two lines shorter.  :-)
15:33 RealBadAngel http://i.imgur.com/bUnlmEq.png
15:33 sapier can you try if adding 2 to width helps too?
15:33 RealBadAngel ofc
15:33 sapier maybe that'd fix the SIN GLE issue too
15:33 Megaf Hi all
15:33 Megaf how bad would it be if Minetest used TCP/
15:33 Megaf ?
15:34 sapier you'd have to open two ports instead of one
15:34 sapier non reliable communication still has to be sent by udp
15:34 Megaf TCP would reduce timeouts I think
15:34 sapier is there a pressing need to switch to tcp right now?
15:34 sapier no wouldn't
15:35 RealBadAngel lol, guess what
15:35 RealBadAngel http://i.imgur.com/OPMCR2v.png
15:35 sapier I'd expect it to help but as you ask it doesn't?
15:35 RealBadAngel it helped
15:36 sapier ok so don't change the height but the width only
15:36 RealBadAngel no, you have to change both
15:36 RealBadAngel but lemme try for sure
15:37 sapier that doesn't make sense :-) ... well actually none of this really makes sense ... do you have some special irrlicht version doing line breaks automaticaly ? ;-)
15:37 sapier or some patch none of us has?
15:37 PilzAdam joined #minetest-dev
15:38 RealBadAngel its clean build
15:38 RealBadAngel nothing is modified
15:38 sapier just crazy :-)
15:39 RealBadAngel i was right, on both x and y +2 has to be added
15:39 sapier I don't understand why you have to add it to y
15:39 sapier x ok to avoid that strange autowrapping but why y?
15:39 RealBadAngel because otherwise theyre shifted down
15:40 RealBadAngel i mean labels, vertlabels are shifted up
15:40 Calinou joined #minetest-dev
15:41 RealBadAngel notice that i had that SIN GLE issue for long time already
15:41 sapier I haven't seen that SIN GLE issue by now on debian
15:42 RealBadAngel all vertlabels were affected this way for me
15:42 sapier at least it's consistent ;-)
15:42 Calinou is using ABM registration on several nodes much more expensive than only one node?
15:42 RealBadAngel ive changed pos.X + 15 to pos.X + 17 and got rid of it now
15:42 Calinou what is the best interval for having low global load? 1, 2, something else?
15:43 sapier please don't add them, I'm gonna replace the 15 by a variable value in my dpi/gui_scaling fix
15:43 RealBadAngel also whenever width or height is calculated i have to add 2 to result (at least 2)
15:43 sapier write it 15 + 2
15:44 sapier do you have a special font?
15:44 RealBadAngel no
15:44 sapier some special default settings for freetype?
15:44 PenguinDad ShadowNinja: looks much better than the old perl server list
15:44 RealBadAngel no, its clean and just built copy
15:44 sapier no I meant system wide default settings
15:45 RealBadAngel i dont have any system wide install of the game
15:45 RealBadAngel all are made with drun_in_place=1
15:45 sapier yes but maybe freetype lib uses them automaticaly?
15:45 sapier I'm just guessing right now
15:46 RealBadAngel it behaves the same with freetype enabled and disabled
15:46 RealBadAngel it doesnt matter
15:46 sapier something has to be different because you're the only one telling/having it
15:46 RealBadAngel idk whats different
15:47 sapier ok guess 2 pixels aren't worth spending more time on it if it's fixed by adding that additional room I agree to do it ... at least if you use the +2 style and add a comment
15:48 RealBadAngel what comment should i put? "im adding here 2 because idk wtf is going on." ? :)
15:48 sapier maybe not exactly this words but something like that ;-)
15:49 sapier just make sure it's clear what that 2 is from ;-)
15:49 Megaf sapier; would it be possiblet to use dynamic cache on clients, acording to the amount of RAM?
15:49 sapier cache?
15:50 Megaf we already have the unlound unused timeout
15:50 Megaf that timeout could be dynamic
15:50 sapier what cache
15:50 sapier textures ... no
15:51 sapier engine has to have all textures in memory
15:51 RealBadAngel also for vert labels i suggest using getDimension for x too
15:51 sapier map ... maybe don't know that subsystem that much
15:52 RealBadAngel fixed 15 or 17 may be too few for bigger fonts
15:52 sapier hmm I thing I didn't use getDimension there because I'd have to do a getDimension on each single char to find the one with tallest width
15:52 sapier -g+k
15:52 sapier quite a lot of cpu operations to save only a couple of pixels
15:53 RealBadAngel ok, so 15 +2
15:53 sapier tallest witdth is denglish isn't it?
15:55 jin_xi i know it not
15:55 RealBadAngel well, theres another problem
15:55 RealBadAngel by adding 2 to height it fixed label position
15:56 RealBadAngel but fucked up tooltips :)
15:56 RealBadAngel multiline tooltips are not correctly displayed anymore
15:56 werwerwer joined #minetest-dev
16:02 RealBadAngel well, its not caused by that
16:02 sapier lol
16:02 sapier maybe you should try updating to irrlicht 1.7.3? ;-)
16:02 RealBadAngel \n doesnt work anylonger for tooltips
16:03 sapier ok guess we need to find the real reason for why it's failing for you
16:03 RealBadAngel its something related to word wrapping for sure
16:04 RealBadAngel suddenly \n is not workin
16:09 Hunterz joined #minetest-dev
16:16 NakedFury joined #minetest-dev
16:21 RealBadAngel ok, when i compile with freetype, /n is not working
16:47 Jordach joined #minetest-dev
16:59 sapier comments? https://github.com/minetest/minetest/pull/1391
17:03 kaeza joined #minetest-dev
17:06 jin_xi sapier: is this like a fader element or strictly scrolling related?
17:07 sapier https://github.com/minetest/minetest/pull/1392 actually just a rebased #1206
17:07 sapier it doesn't do scrolling itself
17:07 sapier it's just a scrollbar
17:08 sapier I wanted to do a trackbar but irrlicht doesn't support those (without arrow buttons at end)
17:08 sapier scrollbar will do too
17:10 diemartin joined #minetest-dev
17:12 Megaf is minetest.net offline?
17:20 CraigyDavi Megaf, not for me
17:21 celeron55 joined #minetest-dev
17:21 Megaf Ok, its just a little slow
18:04 sapier pushing #1392 now
18:05 kaeza can we remove the automatic "Guest" naming before .10 ?
18:06 sapier do enough ppl agree to it?
18:06 kaeza most server owners block those usernames
18:10 celeron55 joined #minetest-dev
18:10 PilzAdam joined #minetest-dev
18:10 Megaf joined #minetest-dev
18:10 SmugLeaf joined #minetest-dev
18:10 alexxss joined #minetest-dev
18:10 restcoser joined #minetest-dev
18:10 Piggybear87 joined #minetest-dev
18:10 harrison joined #minetest-dev
18:10 lanxu joined #minetest-dev
18:10 robmyers joined #minetest-dev
18:10 dzho joined #minetest-dev
18:10 kaeza I'd dare say the servers that willingly let "Guest" users in, can be counted on the fingers of one hand
18:11 deltib joined #minetest-dev
18:13 sapier maybe but is this a reason to prevent it client side?
18:14 celeron55_ joined #minetest-dev
18:15 kahrl joined #minetest-dev
18:15 kaeza is there a reason to have it at all? I mean, it's simply typing a name into a textbox, plus many of those users don't even know what they are called
18:16 celeron55 joined #minetest-dev
18:16 PilzAdam joined #minetest-dev
18:16 Megaf joined #minetest-dev
18:16 SmugLeaf joined #minetest-dev
18:16 alexxss joined #minetest-dev
18:16 restcoser joined #minetest-dev
18:16 Piggybear87 joined #minetest-dev
18:16 harrison joined #minetest-dev
18:16 lanxu joined #minetest-dev
18:16 robmyers joined #minetest-dev
18:16 dzho joined #minetest-dev
18:25 kahrl sapier: although it's legal C++, it seems a bit weird to use struct curl_slist* when ENABLE_CURL is 0; can we make fetchJsonValue's second parameter std::vector<std::string> extra_headers?
18:28 kaeza sapier, also: https://forum.minetest.net/viewtopic.php?id=6213
18:28 sapier just provide a pull request to remove it kaeza ;)
18:29 sapier kahrl I'll try to do so
18:29 VanessaE Sapfor G*d sakes, in the name of all that is sane and good in the world, get the fruck rid of Guest usernames.
18:30 VanessaE fruck?  no, I mean FUCK. :)
18:30 VanessaE sapier ^^^^
18:30 VanessaE (note to self:  preview my screen before hitting enter)
18:30 kahrl sapier: if it works, you should be able to remove every #include <curl/curl.h> outside httpfetch.cpp (i.e. in mods.cpp and guiEngine.cpp)
18:31 sapier I don't care about those names ;-) if someone provides a patch and volonteers to take the punches I'll merge it ;-)
18:31 VanessaE no, really.  If you lack a mod that blocks them, they're a maintenance nightmare.
18:32 kahrl mods.h I mean
18:33 restcoser joined #minetest-dev
18:33 sapier I think it can be removed
18:33 sapier it == curl struct
18:34 harrison joined #minetest-dev
18:34 Megaf joined #minetest-dev
18:35 ShadowNinja Comments? (Yes, sapier, I'll make a PR later)  http://sprunge.us/gATB?diff
18:35 ShadowNinja It converts the server list server to Python, which is much more readable.
18:36 ShadowNinja The new server also checks the json data more strictly.
18:36 ShadowNinja (Lots of type checks)
18:37 sapier I don't know much about server list
18:37 ShadowNinja It also has a length limit, so you can't add crazy long pieces of data.
18:37 sapier python instead of perl?
18:38 kahrl wait, doesn't node.js mean javascript?
18:38 sapier hmm well I don't need the server list, so I don't care about it beeing python .. but isn't perl way more common then python? (at least as default script)
18:39 sapier guess it's already using way to much uncommon things another one wont be a difference
18:41 sapier I know python isn't exactly "uncommon" but it's usually not in default install too ;-)
18:41 ShadowNinja kahrl: Yes.  Node.js is used for the template compiler.
18:42 sapier but that's been this way before or not?
18:42 ShadowNinja sapier: Actually, it comes pre-installed on bigger distros like Ubuntu.  Neither Python or Perl was pre-installed on my system (Arch).
18:42 sapier ubunut? are you sure?
18:43 kahrl the only place where python could be a problem is shared hosting
18:43 sapier well I'm away from it for about 2 years now, by that time it wasn't default (irc)
18:43 kahrl and you don't want that for the masterserver anyway
18:43 ShadowNinja Python is comparable to perl in porularity.  And yes, at least last I tried it, in fact it cane with both python 2 and 3.
18:43 sapier well python doesn't have perl's history yet :-)
18:44 sapier sometimes i believe perl isn't one language but more 3 or 4 in same interpreter
18:44 ShadowNinja VanessaE: Can you apply this patch on one of your bigger servers?  http://sprunge.us/ELFH?diff
18:45 ShadowNinja Regardless, Python is much more readable, and it might be more performant since it's only started once.
18:45 ShadowNinja The perl script uses CGI and is reloaded on every request.
18:49 VanessaE ShadowNinja: applied globally (easier that way).  which server do you want me to run it with?
18:55 ShadowNinja sfan5: Is this O.K?  We can easily apply it from their patch.  https://github.com/minetest/minetest/pull/670
18:55 proller ShadowNinja, no, perl masterserver uses fcgiwrap and started once
18:55 ShadowNinja proller: Oh, O.K.  Better.
18:56 ShadowNinja VanessaE: It doesn't matter, as long as it's got a lot of mods and other data to send to the serverlist.
18:56 proller and dont forget use  git mv
18:56 VanessaE ShadowNinja: okay
18:57 ShadowNinja proller: git mv is just mv a b; git rm a; git rm b.
18:57 ShadowNinja Er, git add b.
18:57 VanessaE ShadowNinja: watch for Creative to restart.
18:57 proller ShadowNinja, no.
18:58 ShadowNinja VanessaE: What did it print?
18:58 VanessaE ShadowNinja: nothing as yet.
18:59 VanessaE wait, there it is
18:59 VanessaE 14:57:49: ERROR[ServerThread]: Announcement length: 2954
19:00 VanessaE (the message wasn't initially visible in tail -f output, not enough history)
19:00 ShadowNinja Wow, very close to the 3000 limit I set.  I'll set it to 5000 or so.
19:01 ShadowNinja VanessaE: You can remove that now.
19:01 VanessaE ok
19:02 ShadowNinja Also, sending as POST will make it more efficient, since we can use multipart/form-data which doesn't require url quoting.
19:03 sfan5 ShadowNinja: yes
19:04 sapier I suggest using at least a factor 10 to what we know to be legit
19:08 Jordach joined #minetest-dev
19:19 sapier kahrl better this way https://github.com/minetest/minetest/pull/1393 ?
19:21 kahrl sapier: looks good. Can #include <curl/curl.h> be removed from mods.h and guiEngine.cpp now?
19:25 sapier yes ... already updated the pull request
19:28 RealBadAngel sapier, got a real reason for fails with labels, wordwrap was turned on
19:28 sapier where was it turned on?
19:28 RealBadAngel http://irrlicht.sourceforge.net/docu/classirr_1_1gui_1_1_i_g_u_i_environment.html#adb56652b23932a391b08f710a9546ef3
19:28 sapier did they change the default behaviour?
19:29 RealBadAngel we had set it on
19:29 sapier that doesn't explain why it wordwraps for you only ;-)
19:30 RealBadAngel im not sure, maybe something with resolution?
19:30 sapier do you have some special one?
19:30 kahrl sapier: I think it's good to merge now
19:30 VanessaE sapier: remember, he's one of those blasphemous users with a widescreen monitor
19:30 sapier ok I'm merging it now
19:30 VanessaE ;)
19:31 RealBadAngel 1360x768
19:31 sapier hmm I've got three monitors next  to each other ... cumulated about 4200x1200 ... well I never did expand minetest to all screens
19:33 RealBadAngel also seems like we are missing one argument to addStaticText
19:33 RealBadAngel idk if thats important
19:33 sapier there's most likely a default value
19:34 RealBadAngel not important. so just turnin wordwrap off does the trick for labels
19:35 RealBadAngel no need to add anything
19:35 sapier ok
19:36 RealBadAngel vertlabels are worse, there have to be wordwrap enabled
19:37 sapier are you sure? they're done by using \n
19:37 RealBadAngel ive turned it off, and all whats visible is just first letter
19:37 sapier strange I guess we're still missing some piece of the puzzle
19:38 RealBadAngel workin on it
19:43 sapier additional comments to this one? https://github.com/minetest/minetest/pull/1379
19:44 ShadowNinja sfan5: Go ahead and apply it then.  :-)  I'm working on the serverlist and HTTPFetch.
19:44 sfan5 ShadowNinja: can you do it, please?
19:45 sapier ShadowNinja: hope those curl header removals don't conflict to your changes
19:47 TomasBrod joined #minetest-dev
19:47 sapier hmm I wonder if out mod mechanism might be an issue for google appstore
19:48 VanessaE why?  playstarz got away with it
19:49 RealBadAngel sapier, as far +2 is the only solution for vertlabels work for me correctly
19:49 EvergreenTree joined #minetest-dev
19:50 sapier hmm can you do a pull so we can check it works for others too?
19:50 RealBadAngel btw, one real issue is as i mentioned already, when compiled with freetype \n doesnt work
19:51 sapier that's even more strange as it works for me too
19:51 sapier at least for vertlabels
19:51 RealBadAngel it doesnt work for labels, tooltips and vertlabels
19:52 RealBadAngel in vertlabels text is just wordwrapped
19:52 RealBadAngel its not made using \n
19:53 RealBadAngel ive replaced \n with spaces
19:53 RealBadAngel and it still works, but theres no space between words just
19:54 ShadowNinja sfan5: No, not now.  I'm working on adding multipart/form-data support to HTTPFetch and finishing the serverlist.
19:54 sapier you're wrong RBA vertlabels are done by adding \n after each character guiFormspecMenu L1143
19:54 sfan5 ShadowNinja: I'll do it later then
19:55 RealBadAngel ofc theyre coded so
19:55 RealBadAngel but for me \n doesnt work with freetype enabled
19:56 sapier maybe we should have a look at your freetype version
19:57 RealBadAngel idk what ver i have ;)
20:00 RealBadAngel ok, ive disabled wordwrap for vertlabels too, without freetype now everything is ok
20:00 RealBadAngel tooltips too
20:01 RealBadAngel only problem left is that damn newline not workin when freetype enabled
20:05 sapier quite strange
20:07 Miner_48er joined #minetest-dev
20:09 sapier ok I can confirm newlines don't work for tooltips
20:11 TomasBrod left #minetest-dev
20:13 RealBadAngel do you have freetype enabled?
20:14 sapier yes I'm trying without freetype now ... and I already found a location that may be involved
20:14 RealBadAngel for me without freetype it works
20:14 sapier ok for me too ... so at least this one is reproduceable ... good
20:15 proller joined #minetest-dev
20:15 RealBadAngel this means vertlabels should be displayed different with and without freetype
20:15 RealBadAngel because \n doesnt work in general with freetype
20:16 RealBadAngel i mean for you too, you shall notice a difference
20:17 sapier ok but that proofes vertlabels work due to word wrapping only
20:17 RealBadAngel yes
20:17 RealBadAngel but i do have for both already disabled wordwrapping
20:18 RealBadAngel so, what about "the location" u mentioned?
20:19 sapier CGUITTFont.cpp there's code calculation positions
20:26 ShadowNinja https://github.com/minetest/minetest/pull/1394
20:26 ShadowNinja kahrl: ^ Comments on the second commit?
20:27 ShadowNinja Should only post_data be used for application/x-www-form-urlencoded like before?  I like the auto-escaping of post_fields though.
20:29 ShadowNinja It could also be changed to simple "fields" and encode app/urlenc GET requests.
20:34 kahrl ShadowNinja: I think the commit is good
20:35 kahrl although I've never worked with curl_formadd and its friends so I can't really say too much
20:35 kahrl about the auto-escaping feature, I think we should keep it, we don't want to reimplement at every call site
20:35 kahrl it at*
20:36 RealBadAngel sapier, ive changed this line:
20:36 sapier it's not detecting linebreaks at all
20:36 RealBadAngel #580 else if (currentChar == (uchar32_t)'\n') // Unix breaks
20:36 sapier yes that's the line
20:36 RealBadAngel replaced \n with N
20:36 sapier N?
20:36 RealBadAngel now N works as newline
20:36 sapier loh
20:36 sapier lol
20:37 sapier let's find out why '\n' doesn't work
20:38 sapier obviously '\n' is wrong and should be L'\n' but thats still not enough
20:50 sapier that's great
20:51 sapier got it
20:53 RealBadAngel yes?
20:53 sapier yes
20:53 RealBadAngel whats wrong?
20:54 sapier newline is considered to be a invisible char ... and newline handling is only done for visible chars ... quite useless
20:54 sapier I'm pushing the fix soon
20:56 RealBadAngel ok
20:56 RealBadAngel shall i push fixes to GUIFormSpecMenu ?
21:01 sapier please let someone check it first
21:01 sapier those things couldn't be reproduced by anyone by now
21:03 RealBadAngel you know that wordwrap shall be disabled
21:03 sapier ok pushed the fix now newline should work
21:04 sapier it's not disabled by now so not sure about that
21:04 sapier I'd like to check that change doesn't break it for everyone else ;-)
21:04 RealBadAngel so check it out yourself
21:05 sapier just post your patch to gist prior pushing it
21:05 RealBadAngel #1109 Environment->addStaticText(spec.flabel.c_str(), rect, false, false, this, spec.fid);
21:05 RealBadAngel cmon, just edit the file
21:05 sapier the only change?
21:05 RealBadAngel for labels
21:06 RealBadAngel #1155 Environment->addStaticText(spec.flabel.c_str(), rect, false, false, this, spec.fid);
21:06 RealBadAngel for vertlabels
21:06 RealBadAngel one true -> false , for both
21:08 sapier hmm not sure about that change for vertlabels
21:08 RealBadAngel im sure
21:08 sapier well looks crap for me
21:08 RealBadAngel screenie?
21:09 sapier give me a second
21:14 Jordach joined #minetest-dev
21:14 sapier http://i61.tinypic.com/2ro2kk0.jpg
21:17 RealBadAngel i see, same here with freetype and your patch
21:18 RealBadAngel space between letters is smaller
21:19 RealBadAngel it applies also to tooltips
21:19 sapier http://i61.tinypic.com/2ry2blh.png this is how it looks like with word wrap enabled
21:21 RealBadAngel http://i.imgur.com/tlXK40A.png vs http://i.imgur.com/cGIZmnW.png
21:22 sapier ...W is word wrap enabled isn't it?
21:23 RealBadAngel on both worwrap is disabled
21:23 sapier do you have that +2 in there?
21:23 RealBadAngel left is with freetype
21:23 RealBadAngel no
21:23 RealBadAngel its without +2
21:23 sapier strange
21:23 RealBadAngel just wordwrap disabled
21:24 sapier wait left is freetype and no wordwrap?
21:24 sapier and right is?
21:24 RealBadAngel both are no wordwrap, left is freetype
21:24 sapier but right is non freetype?
21:24 RealBadAngel yes
21:24 sapier can you do a screenshot with freetype and wordwrap enabled?
21:25 RealBadAngel sure
21:27 RealBadAngel http://i.imgur.com/ORdgGjL.png
21:27 RealBadAngel also all labels are fucked up
21:28 werwerwer joined #minetest-dev
21:31 sapier so it's either make them ugly for everyone or break them for you ... thats a stupid choice
21:31 RealBadAngel with freetype \n makes too small spaces between lines just
21:32 RealBadAngel and thats all
21:32 sapier no they aren't to small they're just plain line distance
21:33 RealBadAngel how do they call that distance between lines of text?
21:33 sapier I guess that'd be this code "offset.Y += font_metrics.ascender / 64;"
21:34 sapier still the vert labels and normal text have different requirements
21:34 sapier what happens for freetype if you do a width +2?
21:34 RealBadAngel lemme try
21:35 RealBadAngel btw, but tooltips will look wrong for all
21:35 RealBadAngel same too small distance
21:35 sapier well I didn't expect it to change but what about the vertlabels?
21:35 RealBadAngel checkin
21:35 sapier and enabled wordwrap of course
21:39 RealBadAngel looks ok
21:39 RealBadAngel +2 on X is enough
21:39 sapier so we're back to why do you need "+2" :-/
21:40 RealBadAngel whats "kerning" ?
21:42 RealBadAngel ok, i know
21:42 sapier don't exactly know that code was first time to read it for me too ... for what I figured out it's the delta based uppon previous char
21:42 sapier but that could be wrong
21:42 RealBadAngel http://en.wikipedia.org/wiki/Kerning
21:45 sapier ok seems I'm correct
21:50 RealBadAngel hmm, looks like height is get as result of a hack
21:50 sapier maybe that hack is wrong?
21:50 RealBadAngel see line #653
21:50 RealBadAngel 3 characters are tested
21:50 RealBadAngel and ONE pixel added
21:51 sapier quite strange
22:02 RealBadAngel its used only to get dimensions
22:03 sapier not for printing?
22:03 RealBadAngel i think problem with too small spaces are related to that visible/invisible stuff
22:03 RealBadAngel wordwrapping makes correct ones
22:04 sapier interesting
22:08 VanessaE older OS, older version of the font in use, so maybe that tweaks his setup relative to ours?
22:08 VanessaE (re: kerning)
22:37 RealBadAngel VanessaE, how texts on signs are made visible?
22:37 VanessaE textured entities.
22:37 RealBadAngel great
22:37 sapier no crap
22:37 sapier :-)
22:37 VanessaE nope, there's no crap here :)
22:37 VanessaE it's quite good imho, for what it is.
22:37 RealBadAngel we do have function to add text scene nodes
22:38 sapier do we?
22:38 RealBadAngel i just found it
22:38 sapier I know irrlicht has but does minetest have?
22:38 RealBadAngel #999 core::array<scene::ISceneNode*> CGUITTFont::addTextSceneNode(const wchar_t* text, scene::ISceneManager* smgr, scene::ISceneNode* parent, const video::SColor& color, bool center)
22:38 sapier ok so we dont
22:39 sapier I already suggested to use this to replace what vanessae uses
22:39 sapier http://www.freetype.org/freetype2/docs/tutorial/step2.html can we fix the height using this information?
22:39 VanessaE but only if it can replace the existing signs in a feature-complete manner.
22:45 EvergreenTree joined #minetest-dev
22:45 RealBadAngel i tried to play with this: offset.Y += font_metrics.ascender / 64;
22:45 RealBadAngel 52 seems to work fine
22:45 Gethiox joined #minetest-dev
22:48 RealBadAngel sorry, 48
22:48 Eater4 joined #minetest-dev
22:48 RealBadAngel that seems to work just fine
22:49 sapier I don't think 64 is a number
22:49 RealBadAngel line #586 ofc
22:49 sapier seems to be a shift value
22:50 RealBadAngel idk whats that but it works ;)
22:50 sapier someone already "fixed" it without knowing ... we shouldn't do it this way a second time ;-)
22:51 RealBadAngel without "fixing" it again, freetype will stay useless (almost) for new features
22:52 sapier I wasn't suggesting not fixing it but fixing it right ;-)
22:52 asl joined #minetest-dev
22:52 RealBadAngel ascender, descender -> They are expressed in 26.6 pixel format.
22:54 RealBadAngel height
22:54 RealBadAngel The scaled version of the original design text height (the vertical distance from one baseline to the next). This is probably the only field you should really use in this structure.
22:54 sapier there's something about this beeing implementation specific
22:54 sapier :/
22:55 RealBadAngel so, according to that, using font_metrics.ascender is just wrong ?
22:57 RealBadAngel yup
22:57 RealBadAngel replacing "ascender" with "height" works
22:58 RealBadAngel at least for newlines
22:58 RealBadAngel replace line #586 with offset.Y += font_metrics.height / 64;
22:59 sapier ok trying
23:00 sapier better
23:03 smoke_fumus joined #minetest-dev
23:04 iqualfragile joined #minetest-dev
23:07 iqualfragile minetest.net/mods contains outdated links
23:10 grrk-bzzt joined #minetest-dev
23:10 VanessaE well links to outdated mods to be more specific.
23:10 VanessaE but they all still work
23:10 VanessaE the links still work rather
23:13 RealBadAngel sapier, line #599, same replacement needed
23:14 RealBadAngel multiline text will be properly centered thx to it
23:14 sapier ok
23:14 RealBadAngel vertically, ofc
23:16 RealBadAngel well, not really
23:17 RealBadAngel thx to it its moved down slightly
23:20 RealBadAngel so just the 1st one
23:20 RealBadAngel im goin to take a nap, bbl
23:34 grrk-bzzt joined #minetest-dev
23:40 sapier left #minetest-dev

| Channels | #minetest-dev index | Today | | Google Search | Plaintext