Minetest logo

IRC log for #minetest-dev, 2015-11-09

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

All times shown according to UTC.

Time Nick Message
00:03 dandelion joined #minetest-dev
00:03 kilbith joined #minetest-dev
00:04 kilbith someone should try out MT on a Wayland session and see what happens with the camera...
00:05 cib0 joined #minetest-dev
00:05 Anchakor left #minetest-dev
00:07 JohnnyComeL8ly kilbith, mind telling me what happens?
00:07 JohnnyComeL8ly I don't want to switch right now. :-P
00:07 kilbith kind of indescriptible
00:08 JohnnyComeL8ly Ah, well, that merits a try!
00:08 kilbith but XWayland ensure a compat with X11 and it works more or less finely on other games
00:08 kilbith in a mediocre way on MT
00:08 kilbith i'm concerned since it'll be the new standard on Linux very soon
00:10 JohnnyComeL8ly kilbith, I want to see Wayland support too, but it is dependent upon Irrlicht... and they said that isn't a priority for them.
00:10 kilbith well that's a real treat for the future of MT
00:11 est31 what is a priority for irrlicht either way?
00:11 est31 irrlicht is more dead than minetest is.
00:14 hmmmm at least upgrade to Antartica (SuperTuxCart fork of Irrlicht)
00:14 dandelion1 joined #minetest-dev
00:15 JohnnyComeL8ly I'm thinking that they are caught up in trying to match a commercial engine....
00:16 est31 I don't really know what Antartica is about... is it more a generic engine? is it one tailored for STK's needs?
00:17 hmmmm read http://supertuxkart.sourceforge.n​et/Antarctica:_Technical_Details
00:17 JohnnyComeL8ly est31, providing a patch for Wayland support is possible, is it not?
00:19 est31 JohnnyComeL8ly, idk whether this mouse pointer capture thing still works
00:19 est31 but ask c55 he has talked in his channel about irrlicht minetest before
00:24 est31 oh this is cool
00:24 est31 starting STK gives me [warn   ] [IrrDriver Temp Logger]: Level 3: Vertex shader compilation failed at position -1: right away
00:24 est31 but it still runs!
00:24 hmmmm they have a Beastie character!
00:29 dandelion1 left #minetest-dev
00:30 dandelion1 joined #minetest-dev
00:30 est31 joined #minetest-dev
00:31 est31 and yes I've timed out because I've tried to start stk
00:31 dandelion1 Hey, I'm new. Would someone recommend a little maintenance issue to get me started?
00:32 est31 seems 4gb aren't enough to have stk + firefox with a moderate number open + hexchat
00:32 est31 dandelion1, a starter issue?
00:32 dandelion1 Yeah
00:33 est31 dandelion1, which part of minetest are you interested in helping in? minetest_game, minetest engine? client? server?
00:34 dandelion1 Engine. Eventually I want to do sqlite or threading
00:37 dandelion1 Those two things I'm new to. I've done a lot with fractals/cellular automata/dungeon generation
00:47 est31 dandelion1, do you know how to use c++ templates?
00:47 dandelion1 Yes
00:47 est31 nice.
00:48 est31 can you write a ring buffer type and place it in src/util/container.h ?
00:48 dandelion1 Okay
00:48 est31 & then use it to speed up on gregorycu's speedup of the profiler code
00:50 est31 the class is called ProfilerGraph and resides in src/game.cpp
00:50 est31 idea is to replace m_log with a ring buffer
01:03 paramat joined #minetest-dev
01:08 paramat now merging https://github.com/minetest/minetest_game/pull/722 Default/mapgen: Fix missing taiga pine (trivial fix)
01:13 paramat done
01:13 hmmmm heh
01:13 hmmmm just played super tux kart for a while
01:13 hmmmm it's a really cute game
01:16 paramat i intend to check out the latest version
01:37 kaeza joined #minetest-dev
01:45 paramat hmmmm with some help i managed to get mgfractal using mgv5 cavegen, it works but please can you review the implementation sometime? https://github.com/minetest/minetest/pull/3351
01:56 hmmmm oh god wtf
01:57 hmmmm paramat:  why do you need to cast mg to a MapgenV5 exactly?
01:57 paramat to fix an error message after applying pilzadam's advice
01:58 hmmmm that doesn't really fix the error...
01:58 hmmmm okay do you see this line?  https://github.com/minetest/minetest/pull/3351/​files#diff-b3b3932e4dba81e4e399d61603c86d8aR33
01:58 paramat https://github.com/minetest/minetes​t/pull/3313#issuecomment-152870648
01:58 hmmmm change that from MapgenV5 to Mapgen
01:58 paramat ^ that was pilzadam's advice
01:58 hmmmm then change this line https://github.com/minetest/minetest/pull/3351/​files#diff-e876a0a129495f2b2dd1182861fa6ea3R36  to remove the cast
01:59 hmmmm then tell me if you get any errors from that
01:59 paramat okay
01:59 paramat i may have tried that before, will report
01:59 hmmmm yes, so you did change the first parameter to a Mapgen, but you didn't follow along with the rest of the steps needed to generalize that cave code..
02:00 hmmmm this is kind of like somebody asking for a blue house and you paint the front door blue
02:00 hmmmm and then when they ask about the rest of it you cover it over with a giant tarp and say, "oh I casted it blue, see this blue tarp?"
02:00 paramat lol
02:05 Miner_48er joined #minetest-dev
02:08 paramat ok i did try this before. 'mg->ystride' in the cavegen code now causes this error ‘class Mapgen’ has no member named ‘ystride’
02:08 cib0 joined #minetest-dev
02:09 paramat no other errors
02:09 crazyR_ joined #minetest-dev
02:11 paramat maybe define ystride in the constructor..
02:11 hmmmm surely you can figure out how to solve that
02:12 paramat ok i'll fiddle for a while
02:12 hmmmm well logically, tell me how you think it could be solved
02:13 hmmmm if MapgenV5 has a member 'ystride' but Mapgen does not, what's the first thing you need to figure out here
02:17 paramat i was going to try defining it in the ctor 'this->ystride = mg->ystride', there's not much logic behind it
02:18 paramat somehow i need to access the subclass
02:23 paramat fiddling..
02:24 hmmmm paramat:  defining it in the ctor is a good idea
02:24 paramat ah it compiles
02:24 hmmmm also consider moving it out to the base class as well
02:25 hmmmm the first question I'd ask myself is, "what is this 'ystride' variable?  what does it do?  where does it come from?"
02:26 paramat i added 'MapgenV5 *mgv5;' to header, then to cpp: 'this->mgv5 = (MapgenV5 *)mg;' and 'this->ystride = mgv5->ystride;'
02:26 paramat but not tested yet
02:26 hmmmm lol
02:26 hmmmm that breaks MapgenFractal now
02:27 paramat heh
02:27 hmmmm in general, don't cast things unless you actually know what it is you're doing
02:27 paramat i rarely know what i'm doing hehe
02:27 hmmmm in that case what it does is reinterprets the memory at the location specified by "mg" as a MapgenV5
02:28 hmmmm if the object at that location is actually a MapgenFractal, are you sure the same variables will line up in the same exact order and offsets as with MapgenV5?
02:28 paramat ah, of course not
02:31 paramat notice i changed the way to get the content ids, to fix an error message, if ystride can be accessed i can go back to getting those ids from mapgenv5
02:31 paramat sorry i have studied pointers a little but still don't understand them, need to work through some tutorials again
02:32 hmmmm maybe it would help if you saw what was going on at a lower level of abstraction
02:32 hmmmm have you ever stepped through code using a debugger?
02:33 paramat nope
02:33 paramat i have used gdb though
02:34 paramat i have spent many hours on this cavegen problem, it's beginning to drive me nuts =)
02:34 hmmmm ahh don't worry I'll get it then
02:35 hmmmm this takes like 5 sec for me to do
02:35 paramat ok remember to include the adding of a lava level constant
02:36 paramat or you could tell me what to do
02:37 paramat probably less work to do it than explain though
02:39 paramat then i'll learn from it and apply this to mgflat also, to use mgv5 cavegen
02:39 paramat thanks
02:40 VanessaE celeron55: impressive "far mesh" image
02:42 paramat VanessaE http://irc.minetest.ru/minet​est-dev/2015-11-08#i_4445983
02:42 VanessaE saw that
02:42 VanessaE I'd tend to agree though :)
02:43 VanessaE celeron55: regarding that ^^^ storing objects in a separate database would at least mean than a /clearobjects could just delete the objects database.  I'd say that would be rather fast :)
02:44 hmmmm the primary problem is that certain databases minetest uses does not efficiently support spatially stored data
02:44 VanessaE how does sqlite handle it?
02:44 hmmmm sqlite can do this because it has a range query and it also has an R* tree implementation
02:44 hmmmm sqlite uses SQL
02:44 VanessaE because really, is there any reason to support multiple backends for the object database?
02:44 hmmmm redis has a similar query language too
02:45 hmmmm it's literally just because leveldb is a key-value store, there's no way to ask for all rows with X >= -400 AND X <= 400 AND Y >= ...
02:46 hmmmm the way minetest hashes its block positions makes it so that it can't do range queries, but increases compatibility with modern "no-SQL" databases
02:47 hmmmm in any case I'm not convinced a really fast /clearobjects is worth the massive design change this entails
02:48 VanessaE well
02:48 VanessaE I don't personally run /clearobjects very often
02:48 dandelion1 left #minetest-dev
02:48 dandelion1 joined #minetest-dev
02:48 VanessaE but I *do* find myself using worldedit's selection-limited version fairly often (//clearobjects with two slashes)
02:49 VanessaE so at the very least, est31's idea would be useful anyway
02:50 dandelion1 What C++ standard are we using?
02:51 VanessaE whatever one step short of C++11 is :)
02:51 dandelion1 98?
02:52 dandelion1 What's the thinking on that?
02:52 est31 03
02:52 paramat ok i'm going to do stony desert another way, making it a highland biome stacked above desert, as in mgv6. it makes sense that sand would collect on lower ground and the redstone mountains will look good
02:53 VanessaE because not all supported OS's have full C++11 support
02:53 est31 and some core devs are even users of those OSes
02:53 est31 (not me)
02:53 paramat also i feel rainforest should stop at some point below the clouds, so i'll vertically limit that biome
02:53 VanessaE paramat: well
02:54 VanessaE belay that for a moment..
02:54 est31 there have been lots of discussions on the c++11 topic
02:54 dandelion1 Sore subject?
02:54 paramat sure input welcome
02:54 est31 earlier this year, there has been a very strong one led by gregorycu, back then c55 decided to change to c++11 when the new debian came out
02:54 VanessaE I live in the mountains of western North Carolina, and while it ain't a rain forest by any stretch, clouds *below* the mountain tops (themselves covered with evergreens and deciduous) are a common occurrence
02:55 est31 then later on, somehow that descision got forgotten
02:55 est31 but it is consensus it seems that we will require c++11 once precise (ubuntu 12.04) is out of support
02:55 est31 which is 2017.
02:55 VanessaE mountain tops here peak in the 1000 to 2000-ish meters range, or perhaps a bit more
02:56 VanessaE (this area would be considered "temperate")
02:59 paramat ok
02:59 est31 Any progress with the ring buffer dandelion1? Any problems?
03:00 VanessaE this is the sort of area where the elevation varies so much that in the winter (or the end thereof), one end of say, a 10 km drive will have wet ground and tolerable temperatures, while the other is snow-covered and too cold to go out without at least a light coat
03:01 paramat certainly appletrees and pines should continue up into floatlands, but rainforest seems to need to be below the primary cloud cover to be wet enough
03:01 VanessaE maybe the clouds themselves ought to be raised up?
03:01 paramat anyway i'm not sure
03:01 VanessaE or a second, high-altitude layer could be added.
03:02 paramat i prefer extra layers of clouds, but i doubt another layer of drifting clouds would be accepted
03:03 VanessaE you never know
03:03 cib0 joined #minetest-dev
03:03 VanessaE it's not like they're made up of tons of polys
03:03 paramat well if we end up with multiple stacked realms..
03:03 paramat but that's the future
03:03 VanessaE yes
03:03 dandelion1 The buffer going to be full all of the time?
03:04 VanessaE but then again, stacked realms wouldn't even have the clouds anyways
03:04 est31 no
03:04 VanessaE and if you give say a 2km separation between layers, with a second clout layer at, let's say 500m above sea level, then no problem.
03:04 VanessaE cloud*
03:04 paramat so currently all biomes go up to 31000, the issue is what to change with altitude, and what biomes/resources will be wanted on high floatlands
03:05 VanessaE to me, elevation correlates with drier, colder air anyway
03:05 paramat yeah clouds disappear at 2kn distance, so we'd never need to render 2 layers at one time
03:05 est31 well dandelion1 once its filled it will be full, all the time, just until then, it won't be.
03:05 VanessaE so unless a stacked realm specifies some other mapgen parameters or something, then it should eventually turn to something like tundra and then eventually glaciers
03:05 dandelion1 It's for the graph thing of a profiler, yeah? It can be filled with nothing values
03:06 paramat sure realistically all floatlands and mountains would be tundra, but for gameplay that would make floatlands unplayable
03:06 VanessaE naw I don't mean like that
03:07 VanessaE a quick google says 9.8 degrees C per 1000 meters of elevation
03:07 VanessaE so if you used real numbers, you wouldn't get into permafrost until you're nearly at the top of the world anyway
03:08 paramat well temp doesn't vary with altitude in MT, it's 2D noise
03:08 VanessaE I know.
03:08 VanessaE just something to consider for the future when these realms become possible.
03:09 paramat cotton comes from junglegrass, so that will be wanted high up, perhaps rainforest should continue up then
03:09 paramat i'll just do the stony desert for now
03:10 paramat that means no cacti or desert sand high up
03:33 paramat hmmmm how about calculating ystride from csize in the cavegen ctor? csize is a member of class Mapgen
03:39 paramat i'll try it, i'd like to do this commit myself if possible =)
03:44 paramat compiles
03:49 paramat it works. thanks for the hints. it came to me while washing-up, as always
03:50 paramat will update the pr
03:51 paramat this->ystride = mg->csize.X;
03:52 sofar $ bin/minetest
03:52 sofar Segmentation fault (core dumped)
03:52 sofar wheps? master not building atm?
03:56 sofar Program received signal SIGSEGV, Segmentation fault.
03:56 sofar 0x00000000006625c4 in ClientActiveObject::registerType(unsigned short, ClientActiveObject* (*)(IGameDef*, ClientEnvironment*)) ()
03:56 sofar at start, no params or anything
03:56 hmmmm sounds like you might be using irrlicht 1.8.2
03:56 est31 sofar, gentoo user?
03:57 sofar est31: archlinux
03:57 est31 if your system is up to date, you should have 1.8.3
03:57 sofar sucks that it segfaults, is it that hard to detect an incompatible api?
03:58 est31 1.8.2 is broken
03:58 est31 that bug has nothing to do with api incompatiblity
03:58 est31 but is rather a bug in irrlicht itself
03:58 est31 fixed in 1.8.3
03:59 * sofar upgrades
04:00 Miner_48er joined #minetest-dev
04:09 Fritigern joined #minetest-dev
04:09 hmmmm https://github.com/kwolekr/minetest/commit​/bdb7b9b2b73929d0315ef5c022ae03cedbe933de
04:09 hmmmm PTAL
04:09 paramat wow
04:09 hmmmm paramat: that's doable too
04:10 paramat acceptable?
04:10 hmmmm sure
04:10 est31 joined #minetest-dev
04:10 paramat ok good pr is updated and ready to go
04:12 hmmmm looks good
04:12 est31 Can you change "Irrlicht 1.8.2 known to be broken with Minetest Client" to "Irrlicht 1.8.2 is known to be broken - please update ..." ?
04:12 est31 so add "is" and remove "with minetest client"?
04:12 paramat great
04:12 est31 I consider a library that can'T even execute its own examples to be broken
04:13 hmmmm ok fine
04:14 hmmmm https://github.com/kwolekr/minetest/commit​/889f893ff37021d97d1805c2e21128b5befa21f5
04:14 est31 the problem is just that irrlicht has so few users these days, that we are the first to have the bug.
04:14 est31 +1
04:14 hmmmm PROBLEM SOLVED.
04:15 hmmmm now we will never get another "can somebody please help minetest crashes on startup" person
04:15 est31 well, there is suse
04:15 est31 they keep ignoring my bug report https://bugzilla.opensuse.​org/show_bug.cgi?id=951710
04:16 hmmmm it's only been like two weeks
04:16 hmmmm give it some more time
04:16 est31 perhaps I should have pointed out that I am dev for minetest and that all our users report it, its no edge case
04:16 hmmmm agh
04:17 hmmmm should have I added a check for gcc version too, you think?
04:18 est31 well if you want amend the commit
04:18 est31 now its still inside 10 minute range
04:18 hmmmm is it broken with ANY other ccs?
04:18 est31 but doing a commit extra for that? I dont think thats needed.
04:19 est31 well it is broken I think if irrlicht was compiled with gcc 5 and the program with non-gcc
04:19 est31 but im not sure
04:19 hmmmm ugh
04:19 est31 it really is the best option to avoid 1.8.2 entirely
04:19 hmmmm true
04:19 ShadowNinja est31: (Re: C++11) IIRC I think we came to a decision to release it when latest debian stable supported it, which is now the case.  PilzAdam then stated the 2017 date and I supported him because someone was starting a "when to switch to C++11" debate and we'd already had that discussion.  2017 is way too late IMO, we should be on C++14 or C++17 by then.  Even if your distro's default compiler is horribly, horribly old you can always add
04:19 ShadowNinja an updated repo to your package manager or cross-compile, so even obsolete distros aren't completely dropped.
04:20 hmmmm i had a lot of trouble compiling minetest with gcc on this machine because it expects a certain libcxxrt version
04:21 est31 who needs type deduction
04:22 est31 rust could have had type deduction in the function return type too but they decided against
04:22 est31 at some point you do want to know the type
04:22 est31 thats the whole point of strongly typed language
04:30 dandelion1 Type deduction is really nice when you're writing templates
04:31 dandelion1 I could also do with range-based for loops and unordered maps
04:31 hmmmm we avoid templates unless it's really really convenient otherwise
04:31 hmmmm just because C++ has it does not mean it's a good feature
04:31 est31 well, its good for generic containers
04:31 hmmmm it's *okay* for generic containers
04:32 hmmmm you have the whole problem that it needs to be wholly defined in a header
04:32 hmmmm forward defining is difficult too
04:32 hmmmm very annoying slow and bloated
04:32 hmmmm also another nitpick but it seems like a lot of IDEs screw up syntax highlighting with templated functions
04:33 hmmmm but the main reason to hate templates is the complexity of them
04:34 dandelion1 They are a tool and a sometimes complex tool but when used properly the make the solving a problem simpler
04:36 hmmmm right it's just that a lot of the language fanboys seem to use them improperly
04:36 hmmmm i don't have a problem with tasteful template usage...
04:37 dandelion1 People get excited about their new toys
04:37 hmmmm dandelion1:  Are you gregorycu?
04:37 dandelion1 I'm learning databases right now and every problem looks like it can be solved with a database
04:38 dandelion1 No
04:39 hmmmm oh.  you kind of talk like him
04:41 hmmmm paramat:  PTAL  https://github.com/kwolekr/minetest/commit​/9f0efd53358f16108dea4c69d9d7834ec024f78e
04:42 paramat ok will check it
04:52 FR^3 joined #minetest-dev
04:57 FR^3 joined #minetest-dev
05:03 paramat hmmmm your pr seems okay AFAIK +1
05:05 est31 PTAL https://github.com/est31/minetest/commit/​63e7137e3bb3ef2531faac995e8817b2c3fedc8c
05:05 est31 before, we had a line with == 90 chars
05:06 est31 & it violated the < 6 lines of indentation rule too
05:07 FR^3 joined #minetest-dev
05:07 est31 (on wish of ShadowNinja)
05:17 FR^3 joined #minetest-dev
05:18 ShadowNinja LGTM
05:19 paramat seems okay to me
05:21 ShadowNinja hmmmm: Why the constants added to the seeds?
05:22 paramat i'd like to merge https://github.com/minetest/minetest/pull/3351 now can i go ahead?
05:23 paramat now merging 3351
05:23 paramat oh i need to rebase nvm
05:23 ShadowNinja If for uniquness, then just add 1, 2, 3, ... if PCG handles that well, or do something more thorough like "seed ^ 0xF0F0F0F0" if not.
05:24 paramat anyway me next
05:28 paramat ok now merging
05:32 paramat done
05:37 hmmmm ShadowNinja: ?
05:40 ShadowNinja hmmmm: PcgRandom ps(blockseed + 53);  --  Why 53?
05:41 hmmmm because i like the number 53
05:45 dandelion1 est31: The ring buffer is finished
05:46 paramat progres on cavegen means mgflat will be ready to go in soon, but hidden while we sort out mgflags
05:48 est31 okay dandelion1. Th next step is to use it in game.cpp
05:48 est31 or have you done that already as well?
05:49 dandelion1 No. I was thinking code review and merge and do the use of it separately.
05:50 est31 having it both in one is okay too
05:50 est31 that way you know right away what it's for.
05:51 est31 but two changes, okay for me
05:51 est31 do you know how to make pull requests on github etc?
05:51 est31 if yes then make one :)
05:52 Player2 joined #minetest-dev
05:53 dandelion1 Okay, I'll finish this up tomorrow. Goodnight
05:54 dandelion1 left #minetest-dev
05:54 est31 night
05:54 est31 I'll go too
06:22 paramat left #minetest-dev
06:33 Calinou joined #minetest-dev
06:42 Hunterz joined #minetest-dev
07:00 est31 joined #minetest-dev
08:01 celeron55 oh, about C++11
08:01 celeron55 the thing is, we can't start using it before deciding how exactly we do it
08:02 celeron55 it's not a good idea to just declare that "now we use it"
08:02 celeron55 that would be just a mess
08:03 celeron55 we need to have a plan that tells us what features are encouraged and what are not, and what things will be allowed to be reworked for the sake of it and what things will be left as-is until they actually need changing
08:03 est31 well, we won't need out threading lib anymore, do we
08:04 est31 agree on the idea to first make a plan before discussing about PRs that change all for loops or make everything auto
08:08 hmmmm i don't think i like those types of PRs anyway
08:10 celeron55 i guess we could propose random things now and try to form a coherent wiki page at some point
08:10 hmmmm well whatever it is, it has to actually help minetest in some way
08:10 hmmmm really want people to get into the habit of asking themselves, "how does this PR help minetest" before they decide to make one
08:11 hmmmm really tired of these massive and random bullshit changes
08:12 celeron55 i guess we're going to be pretty strict about changing old code just for stylistic reasons
08:13 celeron55 so we should mainly talk about what to allow in new code
08:13 hmmmm well right off I can say that unordered_set and unordered_map should go in
08:13 hmmmm i love the new emplace() for containers
08:14 hmmmm foreach is great for iterators
08:14 hmmmm auto is great too, but it's potentitally dangerous
08:14 hmmmm i don't think it should be used for anything aside from iterator types
08:14 hmmmm (that's also where the greatest benefit is imo)
08:15 celeron55 i'll start by saying that allowing auto for iterators and allowing non-static member initializers are things that makes programming an important bit more pleasant
08:15 hmmmm old non-C++11 threading/mutex/what have you code should not be trashed right away if at all
08:15 celeron55 (well you started already, but i started that line like that before you said anything lol)
08:16 celeron55 but changing those just for the sake of changing them makes no sense
08:16 hmmmm initializer lists got better
08:16 hmmmm like for example NoiseParams could benefit from this
08:17 hmmmm I don't want a bunch of PRs going around changing this though unless it's unobtrusive
08:17 hmmmm we can change the statically initialized std::maps (for example the color names to color map)
08:19 celeron55 i would also allow unique_ptr for memory management in new code; maybe not shared_ptr due to the reasons you have said before
08:19 hmmmm yeah, unique_ptr doesn't have all the same problems as shared pointers
08:19 hmmmm I still think it's kind of BS though
08:20 celeron55 well, if it makes someone more comfortable developing things, you don't have a good reason against it
08:20 hmmmm outright ban on lambdas
08:21 hmmmm also there's that custom literal modifier function but I forget what it's called
08:21 hmmmm where you can do H"32" and it'd transform the string to "0x20" or something like that
08:21 hmmmm i agree it's really cute but i think it's a little too implicit
08:22 hmmmm people need to actually be aware that there's code running when they add that little "H" or what have you
08:22 celeron55 i don't like those at all
08:22 celeron55 i like lambdas though, but in current minetest i don't see many places where they would be useful so it's a good idea to make people talk before doing
08:23 hmmmm my take on C++11 lambdas
08:23 hmmmm it's not going to boost our productivity
08:23 hmmmm it's going to confuse the crap out of everybody though
08:23 celeron55 they allow making completely different kind of code than you traditionally do in C++
08:23 hmmmm like you said, there are not many places to actually put them to "good use"
08:23 hmmmm right, AKA confusing the crap out of people
08:23 celeron55 if you'd see the code of one of my projects where i use them, it's like a completely different language
08:24 hmmmm because functional programming is a meme
08:24 hmmmm and C++11 is just C++'s attempt to latch onto it
08:24 hmmmm "am I cool yet guise??? look i have functional programming"
08:25 hmmmm what do you think about nullptr?
08:26 celeron55 well, what is there to think about it? i have found no benefit from using it but i guess it's fancy
08:26 celeron55 i use it in my C++11 projects
08:26 hmmmm my thoughts exactly
08:26 celeron55 no benefit or disbenefit whatsoever
08:26 celeron55 you could just sed -i -e's/NULL/nullptr/g' and whatever lol
08:26 hmmmm i see in tons of other projects people running around adding nullptr everywhere
08:26 hmmmm it's distracting
08:26 hmmmm they think it makes the code better somehow
08:27 celeron55 i guess we could ban if just for the lulz
08:27 hmmmm C++ fucked up early on with some edge cases and now they're trying to fix them all... too late, everybody accepts NULL's quirks
08:27 celeron55 ban it*
08:27 hmmmm I think it should be banned for the sake of consistency with everything else
08:28 hmmmm I mean all of a sudden we're going to have old code with NULL and new code with nullptr, and new code from old people with NULL, and some new people who think nullptr is the greatest thing since sliced bread who go around changing NULLs to nullptr
08:28 hmmmm or modifying old code and using nullptr
08:28 hmmmm basically a mess
08:29 hmmmm i can't think of any other new C++11 features off the top of my head
08:30 celeron55 i'm sure the reworkninjas will just keep their mouths shut until they have made a PR that uses some loopholes in our rules
08:30 hmmmm just one more thing about smart pointers
08:30 hmmmm one thing that I deeply hate about it
08:30 hmmmm the syntax
08:31 hmmmm why is it a template for some certain type?  why isn't it like a new @ or something?
08:31 celeron55 i have STL's bloated identifiers overally and typedef things in my projects started after minetest to ss_, sv_, sm_, sp_, up_ and stuff
08:31 celeron55 being std::string, std::vector, std::map, std::sharerd_ptr, std::unique_ptr
08:31 celeron55 have? *hate
08:32 celeron55 i should wake up before trying to type on IRC
08:32 hmmmm the length of std::unique_ptr< ... > is enough to make me want to never use it
08:32 hmmmm I know it's making my code "better" in some vague manner
08:32 celeron55 up_<Thing> is awesome
08:32 hmmmm Thing @ would be the best
08:33 celeron55 i would really like if i could get these shorthands into minetest but i haven't bothered trying because i guess people that like typing twice the amount of characters all the time would object
08:33 hmmmm that's because you're you, doing it for your own personal project, and you're the only one working on it
08:34 celeron55 how does that explain anything
08:34 hmmmm don't increase the mental load for 100+ contributors...
08:34 hmmmm they don't know your bits of shorthand
08:34 celeron55 i'm sure people would get used to those very fast because they are so common; and rare ones don't need shorthands because they are rare
08:35 celeron55 we have all these u16's and stuff too; they are not that much different
08:35 hmmmm hrmm
08:36 hmmmm is $ a valid identifier character in C++03?
08:36 celeron55 it... is
08:36 hmmmm hm it's not
08:36 hmmmm that's just a GCC-ism
08:36 hmmmm also other old compilers allow it
08:36 celeron55 oh, so it's valid in g++, not c++03 8)
08:36 hmmmm that would've been nice
08:37 hmmmm I can live with $V<Type here>
08:37 hmmmm that looks really nice in fact and I can hold onto the shift bar for three characters
08:37 celeron55 $V is harder to type than something like up_
08:37 celeron55 $ comes from altgr+4 in finnish and german keyboards
08:37 hmmmm I'm trying to type it right now and it seems like the opposite for me
08:38 hmmmm oh okay
08:38 celeron55 and < doesn't need shift
08:38 hmmmm well
08:39 hmmmm I guess it would make sense to optimize for german keyboards since for whatever strange reason a majority of minetest devs seem to be german...
08:39 hmmmm i'd rather try to find something universal though
08:40 celeron55 $v<> would be kind of fine because of not having to switch between pressing altgr and pressing shift
08:40 celeron55 actually really fine
08:40 hmmmm harder for me though
08:40 hmmmm because I have an american keyboard
08:40 hmmmm well what a moot point
08:40 nrzkt joined #minetest-dev
08:40 hmmmm it's not like we can actually use $ for real
08:40 celeron55 why not just V<> though?
08:41 celeron55 it's not like anyone uses a thing like V for anything anyway
08:41 hmmmm macro parameters, template parameters,
08:41 celeron55 oh, template parameters i guess yeah
08:41 hmmmm in any case I'm not really sure if inserting our own shorthand is that great of an idea into an already established project
08:42 hmmmm it's good to be as non-weird as possible
08:42 jin_xi joined #minetest-dev
08:42 hmmmm a lot of the style decisions have been made on the basis of it being a quasi-standard
08:42 celeron55 well, a C++11 transition is the absolute best time for it; no other such chance comes ever again
08:42 hmmmm really?
08:43 celeron55 well, maybe in 10 years but anyway
08:43 hmmmm I don't think a C++11 transition should mean anything except switching a few compiler flags
08:43 hmmmm and if people want to start using their C++11 features... they can
08:43 celeron55 well the transition means that we have to write stuff down and people have to read the stuff
08:43 celeron55 so the stuff could mention something like this if we want to
08:43 hmmmm but god no, no big effort to ctrl+H everything just for the sake of replacing it
08:45 hmmmm alright how about Vec
08:45 hmmmm std::vector, 11 characters, to 3
08:45 celeron55 that would be fine
08:46 hmmmm this needs to be a gradual change though
08:46 hmmmm add the typedef to... util/misc.h (we'll rename basic_macros I guess) and just *start* using it
08:48 hmmmm those big omni-commits really mess with git blame
08:51 celeron55 hmm... i think i'm not ready for doing that now though; i'll try to investigate at some point what the usage of those identifiers is and maybe even approximate how many hard-to-read newlines they would have removed or something
08:51 hmmmm this is the fault of the language
08:52 hmmmm it can be fixed with a single piece of syntax like ^ or % or @ or $
08:52 celeron55 there's a benefit but it really may not be high enough; i'm not going to mess things up just for the sake of it
08:52 hmmmm revision after revision they just keep adding things in the form of a template
08:54 hmmmm anyway
08:54 hmmmm I don't necessarily dislike long type names since they typically give some kind of mental note that the type is bigger than some POD thing
08:57 celeron55 std:: contains many POD types though
08:57 celeron55 it's just a stylistic choice for us to not use them from there
08:57 hmmmm those should be abbreviated for sure
08:57 hmmmm but "std::big_thing_here" makes me think twice before typing that big thing a lot
08:57 hmmmm and unconsciously i know that big thing here is bigger than int or u64 or something
08:58 hmmmm subconsciously i mean
09:26 proller joined #minetest-dev
11:00 proller joined #minetest-dev
11:36 proller joined #minetest-dev
11:46 proller joined #minetest-dev
11:49 celeron55 wtf is wrong with logging
11:49 celeron55 i'm printing truckloads of stuff into infostream and it just isn't showing up anywhere
11:49 celeron55 if i change it to std::cout, it shows up
11:50 celeron55 the weirdest thing is that some of it shows up
11:50 celeron55 but only like two lines out of a thousand lines
12:27 troller joined #minetest-dev
12:36 gregorycu joined #minetest-dev
12:37 gregorycu Someone should tell hmmm that I am not everyone
12:39 Amaz joined #minetest-dev
12:57 leat joined #minetest-dev
13:00 proller joined #minetest-dev
13:00 liq_over_ignore joined #minetest-dev
13:00 nrzkt !tell hmmmm: gregorycu said he isn't everyone
13:00 ShadowBot nrzkt: O.K.
13:00 nrzkt gregorycu, done :D
13:00 gregorycu Thank you shadowbot
13:00 gregorycu And nrzkt
13:13 celeron55 http://i.imgur.com/axtJ31S.jpg
13:13 celeron55 holy fucking shit
13:14 celeron55 at this point i am completely terrified of myself, my computer and this weird piece of software
13:16 celeron55 this has some real character to itself too
13:16 nrzkt celeron55, seems better
13:16 celeron55 it's not just "a LOD system"
13:24 celeron55 that screenshot is quite magical to anyone that has used watershed, actually
13:24 celeron55 nobody has ever had such a view of it while in theory that's what it has been generating all the time
13:28 celeron55 large plain areas are actually pretty fun with this because you can always easily find a forest if you want to; just look around and walk straight into one
13:32 zat joined #minetest-dev
13:38 celeron55 oh, looks like a thing isn't even working correctly in that screenshot
13:38 celeron55 well whatever 8)
13:44 nrzkt2 joined #minetest-dev
13:46 iqualfragile joined #minetest-dev
14:04 cib0 joined #minetest-dev
14:10 proller joined #minetest-dev
14:24 iqualfragile joined #minetest-dev
14:37 iqualfragile joined #minetest-dev
14:41 Darcidride joined #minetest-dev
14:46 proller joined #minetest-dev
14:48 iqualfragile joined #minetest-dev
14:50 RealBadAngel celeron55, thats lookin real good. but it should be blurred somehow
14:51 RealBadAngel i mean just the farmesh
14:52 celeron55 i'm not interested in such things
14:52 RealBadAngel thats eye candy im talkin about, so i do undersrtand you
14:53 celeron55 to me eye candy is clear pictures
14:53 RealBadAngel youre not 90% of the population ;)
14:54 celeron55 i don't need anyone to simulate short sightedness for me in my games
14:54 RealBadAngel have you saw my gimp blur with your farmesh mountain?
14:55 RealBadAngel that made farmesh lose its line-y style
14:55 celeron55 yes; it does not look any better
14:55 celeron55 why on earth are you even here if you don't like lines and cubes
14:55 celeron55 go do something else
14:56 RealBadAngel i do love them, but a bit softer :P
14:57 RealBadAngel have u thought u got rid of me already?
14:57 RealBadAngel sorry, winnetou, mt is my hobby :P
14:59 RealBadAngel celeron55, and i should say im sorry to you and hmmm at this point. i had problems with my health, high pressure
15:00 RealBadAngel im treating it now
15:01 RealBadAngel so, again sorry for my attitude
15:02 RealBadAngel that wasnt me
15:03 celeron55 that we will see
15:03 RealBadAngel ofc
15:03 RealBadAngel but please do accept that im sorry
15:04 RealBadAngel i took long break for everybody to calm down
15:08 iqualfragile joined #minetest-dev
15:12 Hunterz joined #minetest-dev
15:13 RealBadAngel i dont want my code be called "half-assed", im still here and want to work on it
15:33 hmmmm joined #minetest-dev
15:35 iqualfragile joined #minetest-dev
15:39 VanessaE [11-09 08:00] <nrzkt> !tell hmmmm: gregorycu said he isn't everyone
15:50 VanessaE um, wut?  http://pastebin.ubuntu.com/13209282/  (server is at commit 1384108f)
16:02 celeron55 so... i think this may be testable by someone else now
16:02 celeron55 https://github.com/celeron55/​minetest/commits/far_map_wip
16:02 celeron55 there it is
16:03 celeron55 it's fully protocol compatible so it can be thrown anywhere anyone likes
16:04 celeron55 but it might bloat up a world pretty badly; and running a lua map generator with is completely insane
16:04 * VanessaE clones...
16:05 cib0 joined #minetest-dev
16:06 celeron55 really any map generator other than paramat's built-in watershed (that is included in that barnch) is mostly nonsense
16:06 celeron55 branch*
16:11 VanessaE "creating farmap atlas"  hm
16:12 VanessaE very interesting effect.
16:13 VanessaE it's a little fiddly in a couple places but it DOES give the impression of more, distant land.
16:13 jin_xi joined #minetest-dev
16:14 VanessaE suggestion:  ignore plantlike objects when generating the far meshes
16:15 Krock joined #minetest-dev
16:15 VanessaE also, sink the farmesh into the fog a bit more
16:19 celeron55 press ctrl+- to make it not try to render it so far
16:20 VanessaE ah, that helps a bit yet
16:20 VanessaE yes*
16:20 celeron55 it defaults to 500 and it will not adjust the fog to differ from that in any circumstance
16:20 celeron55 and this is intentional
16:21 celeron55 maybe the fog should get thicker earlier but i didn't bother modifying the shader to make that effect
16:22 VanessaE 300 is a good setting for VE-Survival (well, local copy thereof)
16:24 celeron55 take a screenshot or something
16:24 celeron55 i don't really have a library of worlds to try it in
16:25 VanessaE http://digitalaudioconcepts.com/van​essa/hobbies/minetest/screenshots/S​creenshot_2015-11-09_11-25-26.png
16:26 celeron55 i bet that it's not that impressive in existing worlds though; people will start building stuff very differently once they have this in their hands
16:26 celeron55 how about in some open area?
16:26 VanessaE sec
16:28 celeron55 also, everybody should note that my to-do list contains about 20 entries; don't expect it to work or be what i want it to be
16:28 VanessaE no worries.
16:29 VanessaE http://digitalaudioconcepts.com/vanes​sa/hobbies/minetest/screenshots/rando​m/Screenshot_2015-11-09_11-28-37.png
16:29 Krock 404
16:29 celeron55 200
16:30 Krock stil 404
16:30 Krock VanessaE, could it be that I'm being pranked? Your homepage root shows up a Nginx welcome page here
16:30 VanessaE the "stripey" effect can be mitigated somewhat if you always choose the "base" texture, e.g. the dirt layer for dirt with grass.
16:31 VanessaE Krock: the homepage root shows my usual landing page here.
16:31 sfan5 blame your ISP
16:31 VanessaE "Welcome to Abe and Vanessa's website. Please choose a destination:" plus Abe's and my pictures.
16:32 Krock *does a IP lookup*
16:32 celeron55 VanessaE: the stripey effect is very intentional; i could make an option out of it though
16:32 celeron55 VanessaE: what does that blue stuff look like up close?
16:32 VanessaE Krock: should resolve to 198.27.68.162
16:32 celeron55 i need some reference pictures to improve these approximations
16:32 VanessaE checking..
16:32 hmmmm holy shit @ the farmap branch
16:33 hmmmm how is it even possible to generate so many commits
16:33 Krock Got answer from [2607:5300:60:1da2::1] ... must blame my ISP then
16:33 celeron55 VanessaE: also what on earth is that yellow block in the sky?
16:33 hmmmm i wonder if celeron has git commit -a -m ... on a "save" button in his code editor
16:33 sfan5 Krock: what you described happens when your browser doesn't send a Host: header
16:33 hmmmm lol
16:33 VanessaE celeron55: when I fly closer to those blue blocks, they turn into this:  http://digitalaudioconcepts.com/vanes​sa/hobbies/minetest/screenshots/rando​m/Screenshot_2015-11-09_11-33-19.png
16:34 VanessaE yellow block in the sky.. you mean the yellow/black striped thing?
16:34 celeron55 yes
16:34 VanessaE technic HV power cable
16:34 jin_xi nice to see the work on far terrain preview. currently huge generated structures are wasted because they never get fully loaded anyway
16:35 VanessaE goes to the nuke reactor back in the spawn area some 500m away or so
16:35 celeron55 is it just one node wide and high?
16:37 celeron55 is this world downloadable from somewhere, this seems like a suitably impossible exercise for this system
16:37 VanessaE it's a nodebox, 1x0.25x0.25 I think
16:37 VanessaE oh sure, but it's a HUGE download
16:37 VanessaE http://digitalaudioconcepts.com/vanessa/hob​bies/minetest/worlds/Survival_World.tar.bz2
16:37 VanessaE about 10 GB
16:37 celeron55 umm... do we have a tool for cropping worlds
16:37 VanessaE nope
16:38 VanessaE closest would be worldedit, but it can only save small pieces (100 cube maybe) before it becomes unmanageable.
16:38 VanessaE if you want, go to digitalaudioconcepts.com port 30001
16:38 VanessaE turn on local map saving
16:40 Krock sfan5, are you sure? It sends the host header when I visit https://www.whatismybrowser.com
16:41 sfan5 ¯\_(ツ)_/¯
16:41 sfan5 no idea whether thats actually the problem
16:41 proller joined #minetest-dev
16:41 sfan5 it's just something that will cause the same result
16:41 VanessaE celeron55: also, I take no responsibility if your CPU melts down :D
16:41 Krock Okay, thanks for the help anyway
16:42 Krock VanessaE, don't worry. He should use water to cool
16:42 VanessaE Krock: I only say that because that server has a history of stressing peoples' computers :D
16:43 Krock Lol okay :)
16:45 proller joined #minetest-dev
16:46 VanessaE well that's interesting.  I just noticed that, mostly, the shapes of stuff in the farmesh stay put, but the *textures* change as I approach.   this:  http://digitalaudioconcepts.com/vanes​sa/hobbies/minetest/screenshots/rando​m/Screenshot_2015-11-09_11-45-38.png  versus this, when I fly jut a tad closer:  http://digitalaudioconcepts.com/vanes​sa/hobbies/minetest/screenshots/rando​m/Screenshot_2015-11-09_11-45-43.png
16:46 VanessaE just*
16:48 kilbith joined #minetest-dev
16:52 VanessaE <celeron55> god damn it my computer is on fire now.  how did this shit happen?
16:52 VanessaE :)
16:54 Hunterz joined #minetest-dev
16:56 celeron55 okay that's some kind of a mipmapping error
16:58 * Krock wonders why celeron55 is the only person I know who uses the 8) smiley
16:59 celeron55 maybe your GPU doesn't like the 4096x4096 texture atlases this thing uses...
16:59 VanessaE celeron55: possible; this is an R9 280x running the open source gallium driver
16:59 hmmmm doubleyou tee eff
16:59 hmmmm i'm not even sure what that's supposed to look like 8)
17:00 hmmmm anyway I thought node tile textures were supposed to be downscaled
17:04 VanessaE it looks like a lot of older hardware can only do 2048x2048 or smaller, though
17:05 troller joined #minetest-dev
17:08 celeron55 i hope i can get that information from irrlicht somehow
17:18 VanessaE celeron55: GL_MAX_TEXTURE_SIZE
17:18 VanessaE if you can read that var from wherever, there's your limit
17:18 VanessaE http://stackoverflow.com/questions​/18674909/what-is-the-maximum-dime​nsion-a-texture-atlas-should-have
17:19 kilbith left #minetest-dev
17:19 troller joined #minetest-dev
17:20 VanessaE looks like if you split the texture atlas into multiple 2k x 2k chunks, it'll works pretty much everywhere
17:22 celeron55 but that will unnecessarily degrade performance where larger is supported
17:22 VanessaE true
17:22 celeron55 a lot
17:22 celeron55 rendering the far map on a high-end GPU is currently practically a completely free operation
17:22 celeron55 if you start doing stuff like that, soon it won't be
17:24 RealBadAngel with texture atlas you wont be able to get all the texture rotations
17:24 celeron55 ehm... or "high-end"; as if anything i have were high end
17:24 VanessaE RealBadAngel: in this case, it doesn't matter
17:24 celeron55 i just have some recent laptop GPUs
17:25 RealBadAngel either atlas or again copy rotated ones
17:25 RealBadAngel and "modifiers"
17:25 celeron55 thanks for reminding me why i hate to discuss anything with RBA
17:25 RealBadAngel lol
17:26 RealBadAngel i just highlighrted why atlas was off
17:26 celeron55 completely irrelevant complaints
17:26 RealBadAngel and i did that
17:26 RealBadAngel you want it back
17:26 celeron55 this will be three strikes and then kickban
17:26 RealBadAngel really?
17:27 celeron55 yes
17:27 RealBadAngel i dont care, bye
17:27 celeron55 some people are incompatible; we are
17:27 celeron55 i am learning to accept the fact
17:29 * RealBadAngel is waving from far away :P
17:30 VanessaE celeron55:  http://sourceforge.net/p/irrlicht/patches/152/
17:30 VanessaE looks like it IS possible to query irrlicht for max texture size.
17:30 celeron55 thought so; that's good
17:31 crazyR_ dont most games use an atlas of some sort anyway?
17:32 celeron55 yes; minetest used to use but then someone broke it and everyone were too lazy to fix it
17:32 celeron55 now i am using one for this FarMap thing because... well, everything should use one
17:32 crazyR_ lol, Well i gues someone need to fix it :/
17:32 celeron55 it was fully removed
17:32 crazyR_ *hides*
17:33 VanessaE crazyR_: you just volunteered. :D
17:33 crazyR_ haha i wish im not that technically advanced or i really would have volunteered
17:33 crazyR_ *and then i would probably have regreted it afterward
17:34 crazyR_ *correction: not my area of expertise :D sounds better
17:38 younishd joined #minetest-dev
17:41 RealBadAngel celeron55, with atlases u cant do reverse mapping of the texture, flipping it
17:41 RealBadAngel thats why we dropped it
17:41 hmmmm this new version of the atlas you have; I take it that it's been generalized this time around for farmap usage and regular node tiling usage?
17:41 hmmmm RealBadAngel:  really?  because I thought the reason why it got dropped was because it wasn't useful and it caused bugs
17:42 RealBadAngel no
17:42 RealBadAngel that was the main reason
17:42 hmmmm that's sort of a dumb reason
17:42 RealBadAngel not really
17:42 hmmmm at least, in theory, a texture atlas is a huge speedup
17:42 hmmmm not having to switch textures while drawing
17:42 RealBadAngel and memory eater
17:43 hmmmm that's why it was optional
17:43 hmmmm enable_texture_atlas
17:43 celeron55 these logging bugs really bother me right now
17:43 celeron55 i can't use infostream at all
17:43 celeron55 i have to use std::cout because otherwise i don't see the stuff i output
17:43 RealBadAngel that was not a big problem for 16px
17:43 celeron55 how can *this* be broken?
17:44 RealBadAngel but higher resolution tps?
17:44 hmmmm maybe you didn't set your logging level to "info"?
17:44 celeron55 i use --info
17:44 hmmmm yeah dunno.
17:44 celeron55 and expect to see everything on the terminal where i run the thing
17:44 hmmmm does it work if you set it to info in the config file?
17:44 celeron55 it isn't happening; i only see some info lines
17:44 hmmmm interesting
17:45 hmmmm ShadowNinja:  it sounds like there's a big bug in logging
17:45 hmmmm hey, at least it's been cleaned up.
17:45 hmmmm that's what's truly important :-)
17:46 RealBadAngel celeron55, we both know we are incompatible, but im still here and will do my job. chill out.
17:46 RealBadAngel as i will
17:46 hmmmm RealBadAngel you spend 70% of your time on this channel talking about how people don't like you instead of talking about code
17:46 hmmmm seriously wtf
17:47 celeron55 that was the second strike, by the way
17:47 RealBadAngel hmmm, youre right, i should shut up and post links to code
17:47 celeron55 i am *not* joking
17:47 VanessaE RealBadAngel: the reason texture atlases were originally dropped was because there was no way for them to expand to handle large textures or large numbers of tiles.  e.g. HDX32 on default game was enough to break it.
17:48 celeron55 i will be the RBA kick machine if that's where this is going
17:48 VanessaE and no one wanted to fix it.
17:48 VanessaE when it DID work, it worked perfectly fine as far as I could tell
17:49 RealBadAngel VanessaE, reason for atlases to drop was they were not working with 6d facedir
17:49 VanessaE no.
17:49 celeron55 i don't have an opinion on whether RBA can contribute, but i suggest not talking here
17:49 VanessaE they were dropped long before that.
17:49 RealBadAngel a bit after 6d facedir
17:50 celeron55 IIRC they were dropped after RBA failed to implement 6d facedir with them
17:50 RealBadAngel atlases was a blocker for 6d facedir code
17:50 celeron55 and frankly, it should have been possible
17:50 VanessaE the code might have been removed, but the use of them was dropped ages before.
17:50 celeron55 there's nothing that would have made it impossible
17:50 RealBadAngel celeron55, rolling over the texture, that was a blocker
17:51 VanessaE anyway c55 is right.  less talk, more code :)
17:51 celeron55 i don't know what that even means
17:51 RealBadAngel if we have single one, we can reverse it
17:51 celeron55 now stop it
17:51 proller joined #minetest-dev
17:51 RealBadAngel in atlas we were at another texture simply
17:51 celeron55 ok; i have to kick you out
17:52 RealBadAngel was kicked by celeron55: RealBadAngel
17:52 celeron55 that's a knockout
17:52 VanessaE celeron55: the texture-switching glitch actually looks more like some kind of off-by-1 error; I can't imagine 4096^2 being too big for my GPU (though it's surely possible)
17:52 VanessaE fwiw, it also switches if I just pan sideways a little, if I'm in *just* the right spot
17:53 VanessaE s/pan/yaw/
17:53 celeron55 that's very characteristic of mipmaps
17:53 VanessaE hm, ok
18:00 nrzkt joined #minetest-dev
18:02 DFeniks joined #minetest-dev
18:05 cib0 joined #minetest-dev
18:08 nrzkt joined #minetest-dev
18:14 RealBadAngel joined #minetest-dev
18:43 celeron55 VanessaE: grep for 4096 in util/atlas.cpp and see if changing it fixes the problem
18:43 celeron55 wait
18:43 celeron55 there's no such number in there
18:44 celeron55 oh it's 2048
18:44 VanessaE well that ain't it, then :)
18:44 celeron55 well... you could still try
18:44 VanessaE sure.
18:44 celeron55 it's not like i know what the problem is anyway
18:47 VanessaE trying 1024..
18:51 VanessaE nope.
18:51 VanessaE still the same textures-swapping-weirdly glitch
19:08 * VanessaE throws HDX256 at it, just for shits and grins :D
19:11 VanessaE well it works, and the glitch isn't any worse :)
19:11 VanessaE so that's a good sign
19:13 zupoman joined #minetest-dev
19:13 zupoman joined #minetest-dev
19:15 celeron55 yes the farmesh atlas is independent of original resolution; it always generates the atlas from low-resolution versions of textures it's always enough at that range
19:15 celeron55 +because
19:15 VanessaE right
19:17 VanessaE anyways, at far_map.cpp around like 547, I would recommend you filter-out plantlike, torchlike, and probably also signlike and anything else that's "thin"
19:18 VanessaE line*
19:18 celeron55 i'll get to it once i manage to get this otherwise working
19:18 VanessaE ok.
19:18 celeron55 this exploded at me when i tried to make an option so that i can disable the mapgen triggering from farmap requests
19:18 VanessaE damn it I wish I could read this better.
19:19 celeron55 and now i'm trying to get logging even remotely working so that i could see what is wrong
19:19 celeron55 because LOGGING IS BORKED
19:19 celeron55 this is stupid
19:21 rubenwardy joined #minetest-dev
19:24 kahrl_ I think dstream still works
19:24 kahrl_ I always use dstream so I never noticed that infostream and friends are now FUBAR
19:26 RealBadAngel im using dstream too for debug and output all the time
19:28 RealBadAngel left #minetest-dev
19:33 celeron55 ShadowNinja: holy shit wake up already
19:34 celeron55 ShadowNinja: my respect towards you is declining so fast after this particular utter fiasco that i don't even know what to say
19:35 Amaz joined #minetest-dev
19:38 kilbith joined #minetest-dev
19:38 hmmmm celeron is management material
19:39 hmmmm errr oops.... the correct quote is "celeron has upper management written all over him"
19:41 celeron55 what the fuck is the point of this rework even
19:41 celeron55 log.h and log.cpp is much larger for no good reason
19:41 celeron55 and there are more random classes
19:41 celeron55 and everything depends on miscellaneous things from the new versions for no reason
19:41 hmmmm he removed the "ugly C non-OOP interface"
19:42 celeron55 this is it; ShadowNinja is not allowed to commit anything anymore unless i review it
19:42 celeron55 this is utter junk
19:43 rubenwardy Who approved?
19:43 hmmmm here's the PR discussion:  https://github.com/minetest/minetest/pull/2628
19:43 rubenwardy hmmmm did
19:44 rubenwardy as far as I can see
19:44 hmmmm no definitely not
19:44 hmmmm PRs require more reviewers the larger they are
19:44 hmmmm there are others here
19:45 rubenwardy Must have been in IRC :P
19:45 hmmmm there was a lot of talk about this
19:45 hmmmm the main win was the addition of warningstream
19:46 celeron55 warningstream is a loss in my opinion; i intentionally left it out
19:46 celeron55 but it's a small problem compared to everything else that PR contains
19:46 hmmmm the rest... I was really neutral about but if a core developer wants to do that stuff who am I to say no exactly?
19:46 hmmmm I'm not the minetest boss or something
19:50 hmmmm in any case I feel like you would be way more productive if you channel the energy to fixing infostream instead of bitching
19:51 celeron55 i'm not going to fix that crap; i'm porting my old logging back
19:51 celeron55 at least i know how it works
19:54 hmmmm in any case I hope you now realize why I've become so anal about commit quality and usefulness
19:55 hmmmm here, it's not like i left my guard down or anything, i believe i did my due dilligence as a reviewer/approver
20:01 Robert_Zenz joined #minetest-dev
20:02 JohnnyComeL8ly joined #minetest-dev
20:09 cib0 joined #minetest-dev
20:13 proller joined #minetest-dev
20:16 VanessaE joined #minetest-dev
20:17 RealBadAngel joined #minetest-dev
20:20 VanessaE celeron55: btw, once you sort out that glitch, might I recommend binding the farmesh to 'r' so that one can cycle between normal, normal w/farmesh, and all-range
20:23 Darcidride joined #minetest-dev
20:25 kilbith left #minetest-dev
20:31 crazyR_ ^ +1
20:31 RealBadAngel celeron55, https://github.com/celeron55/minetest/commi​t/059a8bb888b4096adeb21d504bcd303b96105e35#​diff-f9ac0e5fb8fc639e1307b9e92244357bR649
20:31 RealBadAngel typo here?
20:39 celeron55 VanessaE: i like it that i can toggle farmap and all-range separately
20:40 celeron55 but dunno
20:40 celeron55 it's bindable to a separate key in case you didn't notice
20:55 proller joined #minetest-dev
20:58 celeron55 this logging thing is just pure what the fuck
20:58 celeron55 i replaced the class thing with the old implementation, and the problem still exists
20:58 nrzkt celeron55, did you look at the ncurses implementation ? i think your problem comes from here
21:01 celeron55 my branch does not contain it
21:11 celeron55 if i step with a debugger to the place where something should be printed to infostream and step through it, it generally seems to work
21:11 celeron55 but when i let it run freely, it never works after some time from startup
21:13 celeron55 this damned thing makes no sense at all; i even did a complete rebuild
21:14 celeron55 except that in one place where i tried, gdb wasn't even able to break at the infostream statement and acted like it didn't exist
21:18 hmmmm i hope you're compiling with -O0
21:19 celeron55 i guess -O1, so that can explain it not finding it; however everything is still weird
21:20 celeron55 ...i just checked my GCC version; it's 5.1.1
21:20 celeron55 is GCC 5 even stable yet
21:21 celeron55 i don't trust this
21:24 sofar it is
21:24 sofar (and I wouldn't either)
21:26 celeron55 with -O0 gdb acts exactly like with -O1, it seems
21:30 liq_over_ignore joined #minetest-dev
21:33 rubenwardy celeron55, segfault. Will rebuild in debug to see if I can get better info.
21:33 rubenwardy https://gist.github.com/rube​nwardy/66875c356f0f2be9f252
21:34 hmmmm heh, not very helpful
21:34 rubenwardy Yeah
21:35 hmmmm reminds me of those new Windows 8 "Something happened" bluescreens
21:35 rubenwardy bt full in release build is useless
21:36 celeron55 now i recompiled with clang 3.7.0, and get pretty much the exact same behavior
21:38 celeron55 there is no conceivable way this could happen
21:38 celeron55 i need to throw my computer in the trash and become a farmerr
21:39 celeron55 -r
21:39 celeron55 logic has ended
21:41 celeron55 i guess clang++ is using the same libstdc++ though
21:41 celeron55 it could be shit
21:48 rubenwardy I can't see any difference in terms of view range with far map on and off
21:48 rubenwardy or FPS
21:49 rubenwardy wait
21:50 rubenwardy nevermind, it was because the areas weren't generated
21:53 celeron55 eh, i guess i'm taking the original minetest laptop to the rescue - it has debian; it will work unlke fedora 22
21:53 celeron55 maybe i'll be disappointed though
21:53 Robert_Zenz joined #minetest-dev
22:03 rubenwardy Weird color for a sea: https://cdn.pbrd.co/images/217avOR2.png
22:05 rubenwardy and weird blocks
22:07 liq_over_ignore joined #minetest-dev
22:16 Taoki joined #minetest-dev
22:16 cib0 joined #minetest-dev
22:18 Taoki joined #minetest-dev
22:23 celeron55 i think this behaves exactly the same on this gcc 4.9 -based build
22:24 celeron55 i'm going to give up on this and hope that this is reproducible and someone figures this out
22:42 proller joined #minetest-dev
22:43 VanessaE you can't give up :)
22:45 proller joined #minetest-dev
23:02 proller joined #minetest-dev
23:26 proller joined #minetest-dev

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