Time Nick Message 00:00 Exio lemme do it and then we'll see :P 00:00 Exio i'm not even sure if it'll work hehe 00:04 ch98 compiling in mac ppc error 00:04 ch98 error: no match for 'operator!=' in 'i != std::list<_Tp, _Alloc>::rend() [with _Tp = RollbackAction, _Alloc = std::allocator]()' 00:04 ch98 on c++ code 00:04 ch98 for(std::list::const_reverse_iterator 00:04 ch98 i = m_action_latest_buffer.rbegin(); 00:04 ch98 i != m_action_latest_buffer.rend(); i++) 00:05 ShadowNinja ch98: File? Line? 00:05 kahrl use a pastebin (and you posted this in #minetest already) 00:05 ShadowNinja Happens with == too it seems. 00:05 ch98 well, there may be more experienced people here 00:06 PilzAdam ch98, all people that are here are in #minetest too; and posting this in many channels seems selfish 00:06 ch98 oh ok 00:08 ch98 I will be using this channel for error messages now 00:09 ch98 Users/net/Desktop/minetest-0.4.4/src/rollback.cpp:117: is the path 00:09 Exio what the actual fuck? 00:09 Exio 0.4.4? 00:09 Exio try with the lastest git 00:10 ch98 4.6 had more errors and asked for something that can't be installed on ppc mac 00:11 ch98 couldn't even get past the cmake 00:12 PilzAdam I dont know if this is the right channel to help you with compiling on mac 00:12 Exio well, i think fixing them by yourself would be faster 00:12 Exio as we don't own a mac 00:14 ch98 well, more reason for learning c and c++ 00:17 Exio https://github.com/EXio4/minetest/commit/087581084b90ef83f632a42b8e5506dfe690f44b 00:18 Exio any things to "fix" before throwing the pull request? 00:19 Exio like, names wrong or so 00:19 PilzAdam what does that do exactly? 00:19 Exio when you press T or / (the default keybinds), it opens a console what closes after pressing return/enter 00:33 Exio k - #751 00:39 kahrl Exio: haven't tested it, but what happens if you press '/' while the (closed) console contains a nonempty prompt? 00:40 Exio it adds the / 00:40 Exio to the end, i need to fix that, i guess 00:40 kahrl I'm not entirely sure what should happen 00:41 Exio this needs a clearInput, no? 00:41 Exio clearInput; add("/" 00:41 Exio ); :P 00:41 kahrl would be nice if the partially completed line went into the history 00:42 Exio making the clearInput throw the stuff to the history, then (i guess?) 00:43 kahrl add a method replaceAndAddToHistory to ChatPrompt 00:44 kahrl (which only adds to the history if the prompt is nonempty) 00:44 kahrl I don't see a problem in calling it from the_game 00:45 kahrl for T: replaceAndAddToHistory(L""), for /: replaceAndAddToHistory(L"/") 00:46 Exio ok 00:46 kahrl see ChatPrompt::submit for how to add something to the history 00:47 Exio yep 00:52 Exio wait, is actually ChatPrompt::submit just an "addToHistory" + "clear"? 00:52 kahrl actually yeah 00:52 Exio ah ok 00:54 kahrl guess you could replace the one submit call in the entire codebase with a replaceAndAddToHistory call 00:55 Exio i actually added an "addToHistory" function 00:55 kahrl that's good too 00:56 Exio http://dpaste.com/1209018/plain/ :P 00:57 Exio well, what you said is better 00:58 kahrl I don't think clear() is even called up to now 00:58 kahrl having addToHistory() and replace(std::wstring) would be enough 00:58 kahrl and replace returns the original content 01:00 Exio just a random thing, should addToHistory public or i just keep it private? 01:00 Exio even if it isn't used anywhere 01:01 kahrl public and then do this in guiChatConsole:423: prompt.addToHistory(); std::wstring line = prompt.replace(L""); 01:02 Exio and then i remove submit, no? 01:02 kahrl yeah 01:02 Exio ok 01:02 kahrl s/line/text 01:03 Exio void ChatPrompt::replace(std::wstring line) 01:03 Exio it actually doesn't return anything, i'll change that 01:03 kahrl right, so it returns what was there before 01:04 kahrl alternatively add a getLine method 01:04 kahrl whatever just like best :) 01:04 kahrl s/just/you 01:05 Exio getLine would be just return m_line; 01:05 Exio (i guess) 01:05 kahrl yup 01:24 Exio kahrl: i wonder, as you know more minetest code, would be making a 'wrapper' (replaceAndAddToHistory in guiChatConsole to m_chat_backend->getPrompt().addToHistory/replace) ok, or what should i do? :P 01:25 kahrl doesn't really matter I guess 01:25 Exio ok 01:25 kahrl you could call it "submit" ;) 01:28 kahrl ^ only partly in jest, but I admit "submit" is a bit misleading since it doesn't submit it to Client 01:28 Exio "jest"? 01:29 kahrl replaceAndAddToHistory says more clearly what the function does 01:29 Exio hehe 01:29 Exio replaceAndAddToHistoryWhenThereIsAnActualLine :D 01:30 kahrl replaceAndAddToHistoryWhenThereIsAnActualLineAndRemoveTopOfHistoryOnOverflow 01:31 Exio (tm) 01:31 Exio ;P 01:31 Exio hell, yeah 01:31 Exio works :D 01:32 kahrl grats :) 01:35 Exio other thing, should escape open the menu after the console is closed? 01:35 Exio i don't think it should 01:36 kahrl the pause menu? it does currently but it shouldn't, I think 01:36 Exio exactly 01:48 Exio kahrl: https://github.com/EXio4/minetest/commit/af917d5db920811bdbc7b410c702c6957f19a1f2 01:49 Exio escape now doesn't throw you to "the pause menu" 01:49 Exio (if you are using the console), just closes it 01:49 Exio plus the change about "" and "/" (they get added to the history), and the stuff from before 01:50 Exio about triggering a bug "more frequently", isn't that other "reason" for trying to fix it somewhen? 01:50 Exio i mean, if it affects more frequently, more people will try to fix it ;P 01:50 Exio "affects more frequently", whatever 01:52 Exio i actually just like that because the history and tab completion 01:53 kahrl the bad thing about the F10 bug is that it is just about impossible to track down 01:53 kahrl I'm a bit pessimistic about it getting fixed 01:53 VanessaE I haven't had an F10 crah in...forever 01:53 VanessaE crash* 01:54 kahrl well maybe it has been fixed :P 01:54 kaeza same here 01:55 Exio i didn't get it for a long ago, too, maybe it got fixed by a side effect of other bugfix/feature :P 02:00 Exio since a long ago, well, whatever 02:01 Exio you can understand it 02:04 kahrl I tried the branch and I like it 02:07 VanessaE which branch? 02:08 Exio https://github.com/minetest/minetest/pull/751 02:08 VanessaE ah 02:09 VanessaE ok, I'll bite. 02:11 VanessaE oh. 02:11 Exio kahrl: https://github.com/EXio4/minetest/commit/1acaf13735ee256932fdad024331974ca8ea14c9 02:11 VanessaE not as fancy as I expected, but I like the change 02:11 Exio what fancy did thing did you want? :P 02:12 VanessaE I expected something like the old command dialog but with several lines' worth of history 02:12 VanessaE but this is good, too 02:15 Exio the "old" command dialog is pretty basic and doesn't support lots of stuff what the console does, using the console "with other behavior" (the close_on_enter for example) seems better for avoiding repeating even more code) 02:15 VanessaE idea: 02:15 VanessaE when the chat console closes, put a brief dark region behind the chat text at the top of the screen. Make it fade away after a couple of seconds. 02:16 Exio wasn't that "done" by other pull request around? 02:16 VanessaE simple bit of visual feedback that'll help alleviate the impression that whatever the user typed was ignored. 02:16 VanessaE dunno, 02:17 Exio https://github.com/minetest/minetest/pull/608 02:17 VanessaE but this should only be visible for a few seconds, and would be a simple dark region identical to what's done to create the console 02:17 VanessaE yes, something like that. 02:18 VanessaE but darker and only momentarily visible. 02:18 VanessaE (I am not in favor of a permanent, darkened region like that, though) 02:19 VanessaE (I've got a decent, shadowed font that I use, so such would be redundant in my setup) 02:19 Exio i actually don't understand what *exactly* you want what that pull request doesn't "do" 02:20 VanessaE simple: 02:20 VanessaE when you open the main console, you've got the darkened background, right? 02:20 VanessaE if you opened it with / or T it auto-closes when you hit enter. 02:21 VanessaE What I suggest is that it should temporarily darken the area behind the remaining chat text in a manner similar to #608, but that darkened area should fade away completely after a couple of seconds. 02:22 VanessaE this will draw the user's eye to the chat text 02:22 VanessaE which of course should contain the output of the command that was executed 02:23 VanessaE (or the line of chat that was just sent) 02:23 VanessaE without something to draw the user's eye to the chat area, you get the impression for an instant that the command didn't execute, because it looks like the console simply closes. 02:24 VanessaE I've seen this same effect on some bug trackers, when you submit a comment (e.g. launchpad), the background of the new comment will turn, say, green for a few seconds and then fade to white. 02:27 VanessaE (right here is where I expect one of, "ehm, wut?" or "fuck no, you moron!" or "oh! I see what you meant" ;) ) 02:41 Exio i still don't get it, grr 02:42 Exio i will try to read it tomorrow; it is getting late and i need to do hw 02:53 VanessaE translation: "ehm, wut?" 02:53 VanessaE ;) 02:54 Exio VanessaE: i need to do two projects for tomorrow 02:54 Exio i have like 5 hours left if i don't sleep; but i need to 02:54 VanessaE relax, I'm just kidding now 02:54 VanessaE you got more important stuff to do. 02:54 Exio but boring* :P 02:55 VanessaE heh 11:15 PilzAdam somone on #terasology just posted this: http://purchasingusa.com/mapsetup.png 11:15 PilzAdam I want this in minetest :-) 11:17 Calinou not that hard, we could add this to world creation menu 11:17 Calinou but we need to make mapgen settings actual per-world settings 11:17 proller +1 11:19 PilzAdam although I would make the preview optional and disabled by default, and put a big red "SPOILER ALERT!" there 11:25 Calinou uh uh 11:43 celeron55 PilzAdam: hmmmm was aiming for such at some time; dunno if he got sidetracked or abandoned the idea 12:22 PilzAdam celeron55, http://forum.minetest.net/viewtopic.php?id=1379 <- can you change the link to point to here: http://dev.minetest.net/Installing_Mods ? 12:53 thexyz PilzAdam: done 13:16 kahrl which of these pull requests shall be merged (next)? #733 #738 #743 #746 #747 #748 #750 13:16 kahrl or any others 13:18 VanessaE 748, hands down. 13:18 VanessaE then 750 and 746 13:19 kahrl any objections to 748? 13:21 Exio the objection is to why it isn't merged yet 13:22 PilzAdam I dont really get what this is fixing, so you need to write it to the changelog 13:23 kahrl I don't really have a concise way of describing it, but I can try 13:23 VanessaE PilzAdam: in most formspecs, when you close the formspec, you have to jiggle the mouse to get the keyboard to work again. 13:24 PilzAdam this really fixes it? then throw it in ASAP 13:24 kahrl actually in all formspecs, but only if you moved the mouse when the formspec was open 13:24 VanessaE e.g. you can't t -> hello! -> enter -> t -> foo....oops, the dialog didn't open -> jiggle the mouse -> t -> foo -> enter 13:25 PilzAdam you have an issue for that open, right? 13:26 VanessaE PilzAdam: ages-old, yes 13:26 VanessaE https://github.com/minetest/minetest/issues/103 13:26 Exio wow, 103 13:27 celeron55 i find it funny how long kahrl's previous attempt at fixing the focus thing was 13:28 celeron55 it was like 1.5 years ago and it modified basically everything everywhere 8) 13:28 PilzAdam https://github.com/kahrl/minetest/commit/04f007404c181a2109db62f9815dcbe470c4a178 <- this one? 13:28 kahrl a-yup :P 13:29 PilzAdam "Fixed bug where you need to move the mouse after closing a formspec" <- would this be ok for changelog? 13:29 VanessaE "jiggle the mouse" :) 13:30 celeron55 you don't need to jiggle it 13:30 celeron55 just move one pixel 13:30 VanessaE celeron55: shush. don't spoil the joke. 13:30 Exio is #132 still valid? 13:30 celeron55 also, as it happens in every menu, so "closing a menu" would be better 13:30 celeron55 -so 13:31 VanessaE Exio: yes, I still get that issue every so often 13:31 VanessaE (whether it's due to the same underlying cause as it was then, I don't know) 13:35 PilzAdam Exio, yes 13:36 Exio other random thing, what about #346 ? 13:36 PilzAdam nobody knows ;-) 13:46 kahrl any objections to 750? 13:49 PilzAdam if it doesnt leak then its good 13:49 VanessaE do we really care about #260, since the described syntax is deprecated anyway? 13:50 VanessaE er #206 13:52 kahrl are quotes in itemstrings really deprecated? 13:53 VanessaE well, no one who pays attention to the api doc uses the old syntax anymore, so.. :-) 13:53 kahrl PilzAdam: the tile code is too complex for me to completely understand it, but afaik textures only get "leaked" when connecting to a server 13:54 kahrl so there's a finite number of them and they get cleaned up on disconnect 13:54 kaeza I guess using quotes in itemstrings is useless anyway 13:54 kaeza nobody in their right mind would use spaces in item names 13:54 kahrl kaeza: strictly speaking the API doesn't prevent it 13:54 kaeza (and I guess the checks don't allow that) 13:54 kaeza hm? 13:55 PilzAdam kahrl, what does valgrind say? 13:56 kaeza if you ask me, I'd say mark it deprecated syntax if it isn't already 13:57 kahrl I haven't been patient enough to try valgrind 14:00 kahrl it spams me with uninitialized value warnings because rotation isn't initialized in the TileSpec constructor 14:01 VanessaE kahrl: https://github.com/minetest/minetest/commit/fbc22f91d511b961b4fbf605b7d8e1b07a7c0e06 14:01 VanessaE the commit where the old syntax was deleted from the api. 14:03 kahrl +eg. '"default:apple" 2' 14:13 kahrl so uhh... valgrind doesn't even show me leak information by default? 14:13 kahrl thank you valgrind for wasting 15 minutes of my time 14:13 PilzAdam valgrind --leak-check=full bin/minetest 14:13 kahrl it could have told me that when I started it not when I quit it 14:13 PilzAdam I also use --log-file=something.log 14:13 kaeza aww 14:14 kaeza but at least you were playing in my server <3 14:14 kahrl not so much playing as getting killed by some jerk while running at 0.1FPS :P 14:14 celeron55 let's make a more randomly broken game so that developers will be playing on servers to debug it 14:15 kaeza ^ 14:15 PilzAdam I get up to 18 FPS when looking at the ground in valgrind 14:15 kaeza interesting: celeron55 was playing on WazuClan the other day 14:16 celeron55 kaeza: what day 14:17 kaeza two days ago IIRC 14:17 celeron55 not me 14:17 kaeza from the looks it was not you actually 14:17 celeron55 also i almost never use the name "celeron55" on servers 14:17 celeron55 more often something else 14:17 celeron55 (if you stumble upon some privilege-less stranger, it has a rare chance to be me 8)) 14:18 kaeza lol IK 14:18 kaeza you don't go talking like "giv me intrct plz!!!" 14:19 celeron55 i probably have played on wazuclan some day to check it out though (can't remember server names) 14:34 kahrl PilzAdam: I don't see any texture related leaks 14:34 kahrl most of them are either in libasound or related to shaders 14:34 PilzAdam good 14:37 kahrl VanessaE: did you try #746? 14:37 VanessaE not yet, I'll do that right now. 14:48 VanessaE 255MB world I use for testing (my old pre-map-reset server), clearobjects maxed out at 601 MB virtual, 128MB resident, ~6.7MB shared 14:48 VanessaE much better. 14:49 VanessaE wait, that was the wrong one. shit. 14:49 VanessaE still, it didn't use much 14:50 VanessaE lemme get the correct figures for the one that was under test. 14:59 VanessaE singleplayer, running that same map. Default textures, my game. Started with ~1.1GB virtual, 481 MB resident, 38MB shared (most of which, doubtlessly, is the client half). Clearobjects maxed out at ~1.2GB virtual, 569 MB resident, 41 MB shared (so a difference of ~88 MB). 15:01 VanessaE I'd call it good, myself. 15:02 kahrl certainly much better than 22.5GB ;) 15:02 VanessaE yep :D 15:02 kahrl does the block unloading seem to cost performance? 15:02 VanessaE I didn't notice, actually 15:02 kahrl awesome 15:03 VanessaE I was only paying attention to the RAM usage :) 15:04 kahrl if it does cause some problems the new config setting can be set to a large value to disable it, anyway 15:04 VanessaE "HA! What'd I tell you!? 88 MB an hour!!" 15:06 VanessaE (go ahead, use that in your official commit message ;) ) 15:06 kahrl heh 17:44 PilzAdam kahrl, just tested max_clearobjects_extra_loaded_blocks = 1 vs. 4096 on a ~70 MiB map; 1 is only ~10 sec. slower 17:45 sapier is bsd license compatible to minetest? 17:50 PilzAdam without the patch it crashes the server (out of RAM) 17:59 kahrl PilzAdam, interesting 17:59 kahrl what default value would you recommend? 18:00 kahrl I thought 1 would be much slower because each unload cycle causes a sqlite transaction 18:00 PilzAdam what is the memory usage in RAM per block? 18:00 kahrl 16k+x 18:00 kahrl where x depends on metadata 18:01 PilzAdam maybe write that info in minetest.conf.example? so people can calculate themselves 18:01 PilzAdam and just leave the current value as default (4096) 18:04 celeron55 just make the default value the smallest that doesn't increase the overhead a lot on regular HDDs 18:04 celeron55 no need for explanations because that will suit practically everybody until forever 18:05 kahrl I don't mind adding a one line documentation to minetest.conf.example 18:06 kahrl 4096 doesn't seem to add much write overhead and consumes between 64MB and 100MB of extra memory usually 18:06 kahrl so it should be a fine default value 18:06 kahrl maybe 1024? 18:07 celeron55 something like "This is a trade-off between sqlite transaction overhead and memory consumption (4096=100MB)" could be a useful comment 18:07 kahrl sounds good 18:10 PilzAdam why do you do this "if(num_blocks_checked % unload_interval == 0)" instead of "if(num_blocks_checked == unload_interval)"? 18:11 kahrl with the latter variant it would only unload once, wouldn't it? 18:11 PilzAdam oh, you dont reset it 18:12 kahrl a similar check is used for report_interval 18:12 kahrl updated the minetest.conf.example comment: https://gist.github.com/kahrl/5700085 18:14 kahrl should the getBlockNoCreateNoEx calls be getBlockNoCreate calls? 18:14 kahrl none of them should fail, but if it does the latter gives a traceable exception 18:21 kahrl since I don't think that really matters, is it ok to merge it (it = the gist)? 18:21 PilzAdam sure 18:29 kahrl https://github.com/minetest/minetest/pull/752 a simple fix for a segfault, looks good to merge 18:30 PilzAdam this function isnt used at all 18:30 PilzAdam ask hmmmm, its part of his mgv7 code 18:30 kahrl sweetbomber was trying something with register_biome 18:30 kahrl but yes it isn't used currently 18:30 celeron55 sweetbomber started developing mgv7 with hmmmm's approval (or that's what i figured from their discussion) 18:32 kahrl hmmmm: does #752 break/conflict with anything you're doing? 18:32 celeron55 not likely at all, as that's in a location recently completely modified by the scriptapi changes and hmmmm has been busy with school stuff 18:33 celeron55 (or something like that) 18:33 kahrl right 18:34 RealBadAngel hi all, a question: new menus based on formspec are gonna be added in o.4.7 or after? 18:34 celeron55 what part of "freeze" you didn't understand? 8) 18:35 hmmmm aa 18:35 hmmmm no 752 is fine 18:35 kahrl hmmmm: ok, I'll merge it 18:35 hmmmm thanks sweetbomber 18:35 RealBadAngel i was away and rather busy for a few days 18:35 hmmmm subtle, that one is. i guess he came across it when he tried to add too many 18:35 RealBadAngel this is my last day on old flat 18:35 celeron55 i think 752 is a good sign of what sweetbomber is doing; it sounds like he's digging into the thing enough to know something 8) 18:36 RealBadAngel theres a box there, chair, some junk and nothing more ;) 18:36 hmmmm yeah, sweetbomber knows what he's doing it seems 18:36 RealBadAngel tommorow im moving to a new place 18:36 Jordach a dev that isnt an IRC asshole \o/ 18:37 PilzAdam RealBadAngel, I dont think your flat is interesting for minetest's development 18:37 Jordach PilzAdam, RealBadAngel: mg_flags = new_flat 18:37 RealBadAngel well, it isnt 18:37 RealBadAngel i am askin bout menus 18:37 hmmmm celeron55, not so much developing mg v7, but improving the biome selection and then he has that bigger idea (which i don't think will work out in the end) for placing discrete structures 18:38 hmmmm if the latter works out, we can use that for caves too somehow 18:38 hmmmm which is big 18:38 RealBadAngel i want to add many menu options, that why im askin for menu system to use 18:39 RealBadAngel i do have many shaders code ready to release 18:39 hmmmm with no need for the extra borders in vmanip, map generation can be completely parallel and it'd remove several long-term problems 18:40 celeron55 RealBadAngel: i'm not fully knowledgeable of what's happening with the menu, but i suggest waiting until after 0.4.7 is released and ask again 18:41 kahrl RealBadAngel: I think the menu will be one of the big new features of 0.4.8 18:42 RealBadAngel celeron55, shaders stuff anyways is too fresh stuff to release it with unecessary hurry 18:43 RealBadAngel but it will need definitely ways to turn parts of it on/off 18:44 RealBadAngel so my guess is waitin for new menu system is better 18:44 kahrl you can do minetest.conf settings for now 18:45 RealBadAngel they will have to be made anyways, of course 18:46 RealBadAngel but the options that basically doubles rendering time deserves meny entry ;) 18:49 RealBadAngel i have almost finished water reflections, but price is high 18:49 kahrl nice 18:49 kahrl does it work at all elevations? 18:50 RealBadAngel by now i am using arbitrary water level 18:50 RealBadAngel sea level in fact 18:51 RealBadAngel and wondering how to make it for varying levels 18:53 PilzAdam terasology has this for sea level only 18:54 RealBadAngel because its the easiest way 18:55 RealBadAngel but usear can easily make many different water surface levels 18:55 RealBadAngel on different heights 18:55 RealBadAngel and thats the problem 18:57 kahrl a very fake but easy way to do it would be screen space reflection 18:58 RealBadAngel current camera view flipped? 18:58 kahrl I'm not sure exactly how it works, but there is code on the internet 18:58 RealBadAngel reflection is a texture 18:59 RealBadAngel you get it by rendering to texture, not screen 18:59 kahrl yeah, that texture would be the current screen 18:59 kahrl what vector calculation have to be done beyond that I don't know 18:59 RealBadAngel then you use this texture as water surface 19:00 RealBadAngel and add effects like distortion 19:00 Exio mix() 19:01 RealBadAngel so basically the texture is a picture in the mirror 19:02 RealBadAngel with border on the sea level 19:03 RealBadAngel this means to get tree view in water, you render the view flipped, clip on the water height level 19:03 RealBadAngel so the mittor image is at the right place 19:05 RealBadAngel the problem is where to clip it 19:05 RealBadAngel if user is not watching the sea level 19:05 RealBadAngel or can see many levels 19:06 kahrl if the user has a supercomputer, clip at every level that has water ;) 19:06 RealBadAngel hehe 19:06 Exio inb4 "For enabling shaders we recomend a SLI with GTX TITANs" 19:07 RealBadAngel cmon, the effects works on ati 4670 19:07 Exio for what kahrl said, i mean :P 19:08 kahrl but, well, you could cycle through the needed levels and only generate one of them each frame 19:08 RealBadAngel which is treated as old one and thus only legacy drivers for ir 19:08 kahrl then the reflections will lag behind the real scene a bit but it might be good enough 19:09 Exio that doesn't sound easy to implement with shaders :P 19:10 RealBadAngel shaders are easier to code than it looks at the begining, but to make something nice with them is different stuff 19:10 RealBadAngel lotsa geometry calculations to do 19:11 RealBadAngel i was wondering for a function to get avereage sea level in view 19:12 RealBadAngel but i guess it will take too much tine 19:12 RealBadAngel *time 19:13 kahrl you could store the average sea level per mapblock (calculate it during mesh generation, and store it in MapBlockMesh) 19:13 kahrl then just compute the average over the stored values of the rendered mapblocks 19:14 kahrl you will also need to store a weight of course 19:18 kahrl getting the mode (rather than mean) seems to be more useful but is more complicated 19:19 celeron55 did RBA add some kind of dual-pass rendering? 19:21 kahrl I assume he did, for the clipped scene for water reflection 19:29 RealBadAngel yes i did 19:29 RealBadAngel second pass renders to texture 19:31 RealBadAngel storing generated water level imho doesnt make any sense 19:31 RealBadAngel user can simply build a pool in a cave 19:32 celeron55 kahrl is not talking about the map generator 19:32 RealBadAngel this should be some render time solution 19:32 celeron55 he is talking about the mesh generator 19:32 celeron55 see MapBlockMesh 19:32 RealBadAngel ah 19:33 RealBadAngel count water levels at generating meshes? that could work 19:34 RealBadAngel seems like a good way to try 19:46 RealBadAngel thx kahrl for the idea 19:46 RealBadAngel i think it may work 19:58 sapier https://github.com/sapier/minetest/tree/ngmm_minizip what do you think about adding full minizip source while only using unzip? 19:59 sapier at least atm only unzip 20:03 celeron55 when we were planning some mmdb stuff, we came up with using .tar.gz 20:03 celeron55 as the automated package format 20:03 celeron55 but dunno 20:04 sapier I guess this would be a minor change ... what about supporting both? 20:05 kaeza gz seems to compress better than zip 20:05 celeron55 the main point is that minetest already has zlib included, which contains functions for handling .gz 20:05 sapier yes but "standard user" is used more to zip files 20:05 celeron55 so you only need to add a tar library, which is small 20:06 sapier minizip s quite less too I actualy use only a single code file 20:06 PilzAdam modders are not standard users 20:06 sapier standard users are the one that need a gui to install mods 20:06 celeron55 the mod database can convert them by itself if needed 20:06 celeron55 or whatever 20:06 celeron55 it's not really an issue as long as there is a single used format 20:06 sapier yes but atm there's only local install support 20:07 sapier and zip seams to be most common used format for mods 20:07 sapier -a +e 20:07 celeron55 if you use only a single file from the zip library, why are you adding like 20 of them? 20:09 sapier that's what I intended to ask to complete license I need to add at least 3 files unzip.h unzip.c and MiniZip64_info.txt 20:09 sapier if someone needs other minizip functions later she/he needs to replace everything 20:10 sapier or find exact version 20:10 sapier so I don't care if you want minimal version only it'll be fine for me 20:11 sapier btw we don't need a tar lib the format is that simple this can be implemented in no time 20:12 celeron55 hmm, well, can you make a more minimal version of that commit? 20:12 sapier of course 20:12 celeron55 like, just rip out all the files that aren't actually used; the version of the library is mentioned in there anyway so files are easy to add if needed 20:14 celeron55 if zip support doesn't require more than a couple of files from that, then i don't see any reason for tar.gz 20:14 kahrl if I had to choose a format I would choose tar over zip 20:14 kahrl zip has some not nice things like the pain it is to find the zip directory 20:14 kahrl making untar a lot less code than unzip 20:15 celeron55 well if there is some actual reason why zip is less handy, then it's obviously not a good idea to use it 20:16 celeron55 i have never used any file archive libraries so i have no idea about anything 20:16 kahrl I assume minizip already handles all that stuff, so it's not like this code needs to be written, but it is still more code that is added 20:19 sapier unzipping is already implemented in that code 20:19 sapier it's not that simple I hoped to be but still not complex 20:21 sapier ok pushed minimal version (for linux) maybe there are two more files on windows as i stripped iowin32.c/h 20:23 sfan5 *reads backlog* ... i'd prefer tar.gz 20:23 sapier maybe we could use version 1.0 too which lacks zip64 support which we most likely don't need 20:24 sapier sfan5 if this was for me I'd prefere tar.gz too but primary users will bin windows users which aren't used to tar.gz 20:24 kahrl the mod repository can accept both formats and convert zip to tar.gz 20:24 sfan5 fow win users .zip would be nicer, indeed 20:25 kahrl the server, when submitting mods, I mean 20:25 PilzAdam doesnt the mod manage handle the archives? 20:25 sapier yes moddb is completely unrelated 20:25 sfan5 zip doesn't store uid and gid, does it? 20:25 sapier it's not required sfan5 20:25 sapier it may even be a problem 20:25 PilzAdam why do users need to know the archive format? 20:26 sapier cause they download it and they get told "don't download anything you don't know" 20:26 sapier I don't intend to add automated moddb downloading in first step 20:26 PilzAdam oh, so the modmanager doesnt download mods? 20:27 sapier not in first version 20:27 sapier in version 2 it will do 20:27 sapier this commit is already huge I don't want to extend it to a level where noone ever will be able to check it 20:30 kahrl I would not show the mod manager by default until it has automatic downloading 20:30 sapier btw minizip addition isn't required in first step too, formspec mainmenu modmanager will use system unzip if available 20:30 kahrl otherwise users will get confused about what it does 20:31 sapier there are two different buttons for local install and download kahrl 20:31 sapier and it's hidden by default atm too 20:31 kahrl what does local install do exactly? 20:32 sapier select a file unzip, try to guess correct filename if folder is incorrect move to global modfolder 20:32 kahrl I see 20:32 kahrl if it does that then it should probably support a number of archive formats 20:33 sapier it's primary intention is to get rid of those "modname-master" errors 20:33 sapier yes it's designed to use different formats but atm it supports only zip on linux ... it's a proof of concept 20:35 sapier i guess zip tar.gz tar.bz2 will be enough ... maybe tar.7z too 20:35 kahrl that's quite a lot, especially if it has to work on windows 20:36 sapier true thats why i started with zip 20:37 sapier libarchive could be used for windows tar support I've found a win32 version and it's available for almost all posix like os too 20:38 sapier but It'd be usable as soon as zip support is available for all os 20:46 sapier is anyone here who could test the minizip code on windows? 20:46 celeron55 i suggest thinking mostly in the long-term on this 20:46 celeron55 short-term benefits don't really count here 20:46 sapier it doesn't do any harm in long term while it adds instant benefit 20:52 sapier and I'll add moddb support once formspec mainmenu is merged upstream for 0.4.8 I guess moddb support will be there no later then 0.4.9 20:54 PilzAdam since we dont have a fixed release schedule that isnt really informative 20:54 sapier since we don't have a fixed release schedule I can't give you more exact information ;-P 20:57 sapier it's sure it won't be added for 0.4.7 pilzadam formspec mainmenu most likely will be added in 0.4.8 but I'm not sure if a online modbrowser with moddb integration voting etc will be ready for 0.4.9 20:57 sapier i mean 0.4.8 21:04 PilzAdam this seems good to me: https://github.com/minetest/minetest/pull/733 21:04 PilzAdam tested with the linked test mod 21:06 celeron55 i really wish people wouldn't assign any major features to any patch versions 21:06 celeron55 it is partly what leads to huge times between versions 21:07 celeron55 but still people will always say "yeah my feature will be in 0.4.current+1" 21:07 celeron55 just please stop it 21:08 sapier you can ask ppl to stop but they wont do "somewhere in future" just isn't an optio 21:08 sapier n 21:08 celeron55 assign a date instead 21:08 celeron55 not a version number 21:08 celeron55 "two months from now" is like a million times better than some arbitrary version 21:08 sapier telling a time is just useless if it isn't a mater of work but a matter of decisions to be made 21:09 celeron55 that's not an argument towards saying a version number 21:09 sapier and after almost two years I still don't have any feeling of how to predict when a decision will be made ;-) 21:10 sapier next release version is same as telling a version number 21:10 sapier I do understand that the version number most likely will be wrong but it's best aproximation I currently have ;-) 21:11 celeron55 maybe instead of coding you should then figure out how to make decisions predictable 21:11 sapier if I would find a way to do that I'd be a very rich man ;-) 21:20 celeron55 anyhow, it could be useful if more people would maintain a roadmap in the wiki 21:21 sapier I guess a roadmap can only be given by a single person or at least a few ... but of course thats my opinion 21:22 celeron55 well it should be done with the people who would be mainly making the things in it 21:22 celeron55 done by* 21:25 sapier that'd be a job for core developers ;-) but they'd have less time for coding if doing this ;-) 21:25 celeron55 you're quite a core developer actually 21:26 celeron55 the listings of developers and github privileges are quite arbitrary 21:27 sapier I'm just contributing what imho is necessary to improve minetest 21:27 celeron55 that's quite what everyone is doing 8) 21:28 sapier true :-) but for some ppl this is only adding fancy new features 21:28 celeron55 it doesn't really get more special than that, altough someone might imagine it in a more glorified way 21:30 sapier glorified? I don't do anything special ... I just want to fix the basics to for more creative ppl than I am 21:30 celeron55 again, that is what i see mostly everyone doing here :P 21:31 sapier sometimes I do have other impressions but maybe I'm wrong 21:31 kahrl PilzAdam, merged 21:31 celeron55 there's wild variation in what people consider "necessary to improve minetest" though 21:31 sapier but back to topic :) release schedule/roadmap 21:32 sapier and it's good to have different oppinions 21:35 PilzAdam kahrl, I could have merged this too, but whatever 21:35 PilzAdam less work for me 21:36 kahrl one less round trip time :) 21:36 PilzAdam and why are there dots at the end of the three latest commit messages? 21:37 kahrl hmm is there a rule about that? 21:38 celeron55 the rule is the rest of the commit messages 8) 21:38 kahrl I know I forgot to change the past tense in 601ab8 21:38 PilzAdam rule 2: http://dev.minetest.net/Git_Guidelines 21:39 PilzAdam "look at the commit log" -> no dots 21:39 Anchakor dat pedantry 21:39 PilzAdam :D 21:39 celeron55 we're north europeans, what do you expect! 21:40 PilzAdam even worse, Im german :-p 21:41 Anchakor dots or no dots always make me stop when making a presentation slides 21:41 sapier ok guys I guess we don't have to proof anytime to be "good" germans .. do we? 21:42 celeron55 Anchakor: the obvious solution is to flip a coin for each line 21:43 Anchakor I decided to alter the convention on every slide and avoid those people who notice and point out 21:43 * PilzAdam hates dots in presentations 21:44 * celeron55 hates presentations 21:44 PilzAdam you most likely dont write full sentences in there 21:44 PilzAdam if you do, your presentation is wrong 21:44 sapier star trek on arte ... I guess star wars has lost cultural war :-) 22:14 * iqualfragile likes chocolate