Time Nick Message 00:00 Warr1024 exio4: I think polymorphism is more useful in cases where you either (1) have multiple different behaviors, and want to override only SOME of them, or (2) are writing an API for consumers who expect you to use that kind of style :-) 00:00 exio4 Warr1024: please don't use polymorphism alone to mean subtyping polymorphism :( 00:00 Warr1024 sorry, I don't really know any other kind of polymorphism 00:01 Warr1024 Ironically, my day job is writing C# code :-/ 00:02 exio4 Warr1024: outside that, the "API problem" seems to be a good job for interfaces 00:02 exio4 classes with multiple inheritance are a mess, and without it they're so limited they are also a mess 00:05 Warr1024 hmmmm: so, any chance of you being the one to merge PR 2462? Zeno's signed off on it already. 00:05 hmmmm hold on i've gotta test it 00:05 Warr1024 oh, cool, thanks 00:07 Warr1024 I'm also frankly not sure exactly where to balance between "being an active advocate for my PR" vs. "not annoying the core devs" 00:07 Warr1024 it has been indicated to me in the past that IRC is a perfectly fine place to ask for a review/merge 00:07 hmmmm yes definitely 00:07 Warr1024 but then I guess I don't know how patient I should expect to have to be to get something merged 00:08 hmmmm i at least appreciate you annyoing me because i don't think it'd ever get done otherwise 00:08 Warr1024 ah, good, thanks then :-) 00:08 Warr1024 I figured that if it got buried onto the second page of PR's on github, it'd probably never see the light of day, realistically :-) 00:10 hmmmm hmm I set anisotropic filtering on and i don't see a difference 00:11 hmmmm do I need to use a different texture pack? 00:11 Warr1024 I was conservative with the default settings 00:11 Warr1024 try bumping texture_min_size to something like 64 or higher 00:11 Warr1024 and use a LOW-res texture pack to see the difference 00:12 hmmmm texture_min_size is 64 by default though 00:12 Warr1024 also, it might help to have a shitty video card, so if yours it too high-end and implements aniso without implicit linear blurring.... 00:12 Warr1024 oh? I thought I set it to 16 00:12 hmmmm yeah, i have a gtx660 00:12 hmmmm you have it 16 in the conf.example but 64 in the code 00:12 Warr1024 ah, shit 00:12 Warr1024 well, that's a bug then 00:12 Warr1024 I can fix that, unless you wanna do it as part of the mrge process. 00:13 hmmmm i'll fix it real quick 00:13 Warr1024 if you enable linear filtering, do things get blurry? 00:13 hmmmm you want the actual setting to be 16? 00:13 Warr1024 16 default is conservative 00:13 Warr1024 it'll prevent it from impacting anybody 00:13 hmmmm might be a good thing or a bad thing 00:13 Warr1024 64 will make it visible, but may have a performance impact for very low-end users. 00:14 hmmmm trilinear filtering + anisotropic looks nice 00:14 Warr1024 and you won't notice the impact if you don't have linear filtering. 00:14 Warr1024 on my cheap intel GMA, enabling aniso IMPLIES linear filterng 00:14 Warr1024 so it looks like arse without this setting. 00:14 Warr1024 I can handle 64 on my intel GMA 950-something-or-other. 00:15 Warr1024 256 drops my FPS measurably. 00:15 hmmmm wait a minute... 00:15 hmmmm what's going on 00:15 Warr1024 eh? 00:15 hmmmm something keeps resetting my config changes 00:15 hmmmm hrmm 00:15 Warr1024 fair sure it ain't me 00:16 Warr1024 my touch in the config area was light as hell :-D 00:17 hmmmm i had like 4 instances of trilinear_filter = 00:18 hmmmm ahh i can clearly see the effect 00:18 hmmmm okay, so you want me to change the actual default setting to 16? 00:18 hmmmm very nice btw 00:20 Warr1024 sure, 16 will have the least risk for existing players 00:20 Warr1024 if they bitch about things being too jagged, we can just tell them to enable that setting :-) 00:20 Warr1024 maybe someday I'll learn how to put it in the UI or something, if there's enough real estate to warrant it. 00:21 hmmmm heh 00:21 hmmmm my advice is to run as far away as possible from the UI code 00:25 Warr1024 awesome, thanks for merging that! 00:27 hmmmm np 00:29 Jordach if we're running out of space why not separate everything into sub formspecs 00:29 Jordach audio, visuals, gameplay etc 00:41 hmmmm because it's a lot of work 00:45 Warr1024 I guess you could add more tabs, assuming that those tabs scroll or wrap or something 00:45 hmmmm haha, they don't 00:45 Warr1024 it might actually be nice to have some "advanced" levels of config or something... 00:46 Warr1024 I'm not surprised they don't scroll :-) 00:46 hmmmm I think we can make a custom control that does 00:46 hmmmm that wraps, maybe 00:46 Warr1024 or I guess you could just nest tabs on that Settings screen, to keep the main tabs simpler? 00:47 hmmmm make a secondary tab list visible/invisible when the first one gets too full :) 00:47 hmmmm sure 00:47 Warr1024 I installed MT on a Windows machine and tried to edit the config file in notepad... no dice, apparently it was a \n text file, not a \r\n fle. 00:47 hmmmm i don't like how they need to look like tabs, though 00:47 Warr1024 I don't think they do need to look like tabs 00:47 hmmmm look at other games' config screens 00:48 hmmmm i think we're ultimately wasting a lot of screen real estate with the current setup 00:48 Warr1024 yeah 00:49 Warr1024 frankly, what I'd really like to see would be an interface that parses (or uses a compiled-in version of) minetest.conf.example and just dynamically adds controls 00:49 Warr1024 I mean, obviously we should consider having "standard" settings and hiding the full monty behind an "Advanced" button or something 00:49 hmmmm you know, another thing... 00:50 hmmmm the move from std::list to std::vector is probably a performance *regression* in many cases 00:50 hmmmm look at Server::m_mods for instance. a ModSpec is a fairly heavyweight structure 00:51 Warr1024 IIRC lists grow faster, vectors do lookup by index faster, right? 00:51 hmmmm each time a new mod is pushed back, it's going to delete and then create a new one 00:51 hmmmm it's way more nuanced than that 00:51 hmmmm but here, especially, a vector is just a bad choice 00:51 Warr1024 it's sort of like hashtables vs. red-black trees. The correct answer to "which one is better?" is "for what purpose?" 00:51 hmmmm maybe if it were ModSpec * 00:52 Warr1024 oh, those are large structs/classes stored inline? 00:52 hmmmm see, with a std::list, when a new element is pushed back it'd just add a new link to the linked list 00:52 Warr1024 ha, it actually didn't occur to me that they wouldn't be pointers. 00:52 hmmmm with an std::vector, if it happens to trigger an allocation (which it does early on), it clears, destroys, etc. all of the heavyweight std::sets and std::strings (which internally have a lot of heap-allocated members) 00:52 Warr1024 yep, ouch 00:52 hmmmm and then it recopies all of the old elements 00:53 hmmmm calls the dtor and calls the ctor 00:53 hmmmm doesn't do a dumb copy 00:53 Warr1024 vectors might be more efficent if you could preallocate intelligently... 00:53 hmmmm vectors might be more efficient if you store a lightweight structure 00:53 Warr1024 oh, it reconstructs them? damn, I'd have expected a memmove() :-) 00:53 hmmmm it does 00:53 hmmmm try it yourself 00:53 hmmmm this is the source of the bug i reverted earlier 00:54 hmmmm because of the reallocation, a push_back() operation even will invalidate any iterators 00:54 Warr1024 this is one of the reasons I prefer straight C when I want to write performance-demanding code... 00:54 hmmmm nevermind an erase() call 00:54 hmmmm with a std::list, iterators are preserved 00:54 hmmmm well tbh this is a problem with using somebody elses' containers 00:54 hmmmm not the C++ language 00:55 Warr1024 true, but you'll have different problems trying to reinvent this stuff... 00:55 hmmmm and they're usually much more subtle :( 00:55 hmmmm but nonetheless dangerous 00:55 Warr1024 not for me, I tried to naively reimplement a binary tree and got a SIGSEGV right up. 00:56 Warr1024 turns out that learning about std::map was more efficient than trying to fix my own impl. 00:56 hmmmm erm, I mean using the STL containers the bugs are more subtle 00:56 hmmmm yea heh 00:56 Warr1024 ah, ok 00:56 Warr1024 well, I guess you end up with subtle bugs because everyone fixes the obvious ones 00:56 hmmmm i implemented my own B+ tree and needless to say there's a lot of code involved 00:56 hmmmm lots of bugs too 00:56 hmmmm but once you do get it correct... you're gold 00:56 Warr1024 subtle ones also tend to become those "load bearing" ones you can't fix because major consumers are dependent on them. 00:57 est31 lol 00:57 Warr1024 https://xkcd.com/1172/ 00:57 hmmmm https://github.com/kwolekr/imgcmp/blob/master/src/bptree.c 00:57 hmmmm i put a lot of effort into this 00:58 Warr1024 yeah, holy shit. 01:03 Warr1024 hmmmm: you can also probably definitively close issue #2154, too: just instruct them to run master, and use the texture_min_size config setting now. 01:03 ShadowBot https://github.com/minetest/minetest/issues/2154 -- Texture filtering smoothes textures 01:05 hmmmm i love closing issues and prs 01:06 Warr1024 me too :-) 01:29 est31 #2460 still open for review... 01:29 ShadowBot https://github.com/minetest/minetest/issues/2460 -- Optimize minetest.get_(all_)craft_recipe(s) by ShadowNinja 01:36 hmmmm I wish GameDefs weren't being passed around but rather an ItemDef, the only part actually needed 01:38 hmmmm the lack of encapsulation here is astounding, but it's a low-priority problem for another day 01:47 est31 thanks hmmmm 01:47 hmmmm sure 03:27 WarrTab Hey, isn't for(y...) for(x...) better than for(x...) for(y...) for image processing? 03:29 hmmmm sure 03:30 crazyR can anyone suggest a string that i can send to my minetest server via udp and an expected responce? 03:31 crazyR its to allow me to use pingdom to track the uptime 03:52 est31 crazyR, https://github.com/minetest/master-server/blob/master/server.py#L135 03:52 est31 I also suggest reading this page https://github.com/sfan5/minetestbot-modules/blob/master/COMMANDS.md 03:52 est31 the bot is at #minetest 03:54 crazyR thanks :D just what i was looking for :D 03:58 Zeno` Robots! 04:00 est31 http://www.ricdes.com/wp-content/uploads/2008/06/bild-1.png 04:01 est31 (or if you have firefox type about:robots) 06:49 hmmmm you know what i just realized? 06:49 hmmmm the most time-intensive thing about a new feature is coming up with the interface for it 06:50 hmmmm at least for me... i change my mind on things 50 times 06:50 VanessaE hmmmm: worse yet, you change your mind half-way through coding the feature ;) 06:51 VanessaE (I'm guilty of the same though) 06:51 hmmmm well 06:51 hmmmm 80% of the done really 06:51 hmmmm if we didn't have that damn lua api 06:51 hmmmm we would've gotten away with it, too! 06:51 VanessaE AHHH! blasphemy! :P 08:05 hmmmm ahhh 08:05 hmmmm I wonder how we didn't catch these 08:06 hmmmm clientlauncher.cpp:379, clientlauncher.cpp:398 08:08 hmmmm https://github.com/minetest/minetest/blob/master/src/util/numeric.cpp#L134 what, everybody else needs to abide by the rule of small rand() ranges except clientlauncher? 11:04 nrzkt 2459 is now closed. The discussion is finished, this PR will not be merged and problems have been discussed. 11:11 Krock nrzkt, I would like to see 2493 merged but I need two devs to agree. Could you please look at it and tell me what you think? 11:12 Zeno` #2493 11:12 nrzkt i'm not the best for this algorithm, it seems simpler, it's stable, but haven't tested it 11:12 ShadowBot https://github.com/minetest/minetest/issues/2493 -- Connect rails with connect_to_raillike again by SmallJoker 11:13 Zeno` It could do with a few more comments 11:13 Krock Okay, I'll try to add simple comments 11:14 Zeno` E.g. the meaning of is_rail_x_all[0] = is_rail_x[3] || is_rail_x[0] || is_rail_x[5]; is not immediately apparent (although spending a bit looking at it, it's correct. The idea though is not having to look at it for 10 minutes and read every line of code :)) 11:15 Krock yes, those codes were also confusing me 11:15 Zeno` The more comments the better IMO (well, apart from int x = 0; // set x to 0 kind of crap heheh) 11:16 Krock :3 11:19 Zeno` I actually looked at the PR about a week ago and it works; and I spent some time deciphering it and trying to think of a different way but I couldn't think of an obviously different way 11:31 Krock is_rail_x[3] || is_rail_x[0] || is_rail_x[5]; was confusing, [0] [3] [5] is now much easier to read 11:31 nrzkt 0 3 and 5 are cube faces, right ? 11:33 Krock yes, kinda. 11:33 Krock well, I've found more unused indexes 11:35 nrzkt why not have constants to define these ? it could make the code simpler to read 11:35 nrzkt for example CUBE_FACE_BOTTOM 11:50 Krock travis/jenkins is useless in this case - going to test that by myself :3 11:55 Krock nrzkt, I hope the current comments explain enough :) 12:01 Zeno` note to self: be careful when testing shader code 12:02 Krock since when are shaders agressive? 12:03 Zeno` I used the wrong word... proper word is opencl 12:15 Warr1024 shader code can be shady sometimes 12:15 Warr1024 damn github offline for maintenance 12:17 Warr1024 I got that GUI filtering code further along. 12:17 Warr1024 VanessaE: fixed the imagebutton filtering. 12:17 VanessaE oh good 12:17 VanessaE + 12:17 Warr1024 Check out pr_guiscaling now for the squashed edition; guiscaling is now my own development history. 12:18 Warr1024 I'm going to start using that branch convention for my PR's from now on. 12:19 Warr1024 I did figure out how to automate a git squash, btw. 12:20 Warr1024 env EDITOR=myscript git rebase -i upstream branch # :-/ 12:22 Warr1024 guiscaling doesn't work on android yet. I haven't positively identified the cause yet, but I strongly suspect it's the non-power-of-2 thing. 12:23 VanessaE it probably is. sapier as I recall had to pad all images to a power-of-2 to make them work 12:26 Krock Zeno`, may I get a +1? :) 12:31 Zeno` Krock, ok 12:34 Zeno` Can another dev also check #2493 please? 12:34 ShadowBot https://github.com/minetest/minetest/issues/2493 -- Connect rails with connect_to_raillike again by SmallJoker 12:34 jin_xi have you seen blockmens client side fences vid? 12:35 VanessaE Warr1024: I can't find your pull request link..what was it? 12:38 Zeno` this is silly. The PR results in code that's better than before and is (slightly) less confusing. It also has the advantage of working. I'll merge in 30 minutes if nobody else can peer review 12:39 VanessaE nevermind, found it. 12:43 VanessaE Warr1024: seems to work fine now 12:44 VanessaE it adds a noticable delay to the paging in U_I but that can't be helped 12:58 Zeno` ok merging now 13:02 Krock \o/ thanks! 13:08 VanessaE Warr1024: can you please test homedecor modpack's signs_lib with your patch in place? 13:08 VanessaE (just place a sign and write something on it) 13:08 VanessaE the rendered text is getting corrupted 13:21 VanessaE ok it's not your patch. 13:22 Warr1024 VanessaE: one other thing you could test that would help out would be making sure that you can turn my patch OFF properly 13:22 Warr1024 i.e. that it doesn't affect anything when disabled 13:23 VanessaE I'll check that shortly 13:23 Warr1024 btw, what's the word on use of inline functions? If I have a strong intuition that a function should be inlined, should I mark it inline, or should I always leave that to the compiler to decide? 13:27 VanessaE idk on that 13:28 Warr1024 eh, I left it off. Micro-optimization is the root of all evil. :-) 13:29 VanessaE fuck. something broke signs_lib :( 13:29 Warr1024 something in guiscaling, or something in HEAD? 13:29 Warr1024 (or maybe something in signs_lib?) 13:29 VanessaE not sure where, except it's not your patch 13:30 Warr1024 ah, good...? :-) 13:30 VanessaE I wonder if it's 1e4fb80d4 13:30 Warr1024 well, at least I didn't break anything; I hate being the one to break something. 13:30 Warr1024 try disabling that. 13:31 Warr1024 if you set texture_min_size to 1, it'll never upscale anythig 13:31 VanessaE rolling back to test. 13:31 Warr1024 and I think texture_clean_transparent. 13:31 Warr1024 You should be able to disable it via config without having to roll back the actual change. 13:31 est31 confirmed sth is broken in master 13:31 Warr1024 IIRC I think we might have made one of those filters, considered low-cost, to be on by default. 13:32 Warr1024 what's the git address for this signs mod? 13:32 Warr1024 oh, er, wait, I can try this out on VanessaE's server? 13:32 VanessaE Warr1024: yeah, digitalaudioconcepts.com 30005 13:32 VanessaE that's the quickest one 13:32 VanessaE the spawn's signs will make it obvious 13:33 VanessaE ok, works fine with commit 26153bab 13:33 VanessaE so 1e4fb80d is what broke it 13:34 VanessaE because the texture filter is being applied to entities 13:34 Warr1024 it's applied to anything with textures 13:34 Warr1024 ok, it works if you set texture_min_size to 1 13:35 Warr1024 texture_clean_transparent doesn't seem to affect it. 13:35 Warr1024 hm, that means I might need to either bypass the filtering for bitmap fonts 13:35 Warr1024 or make the bitmap font code handle upscaled textures. 13:35 VanessaE those aren't bitmap fonts, per se 13:36 VanessaE they're a bunch of images bodged together with ^[combine 13:36 Warr1024 ah, I see how that works. 13:36 VanessaE so if you move your filter to the end of the compositor chain, that might suffice 13:36 Warr1024 yeah, but my filter isn't part of the compositor... :-/ 13:37 VanessaE question then is why are some pixels being duplicated and others being erased 13:37 Warr1024 I guess the tricky part is that it'll work fine with textures that are composited without pixel offsets 13:37 Warr1024 but the problem is that those compositor pixel offsets are for the ORIGINAL textures, and don't account for the resized ones 13:38 Warr1024 ok, so what I probably need to do is find some way to keep the original textures around 13:38 Warr1024 and when constructing the composited image, use the originals 13:38 Warr1024 then upscale after compositing. 13:38 Warr1024 I think I can make that work 13:39 Warr1024 the workaround for that is to set texture_min_size to 1, for now, which will effectively disable the upscaling filter. 13:44 VanessaE ok. 13:45 VanessaE ok yeah, setting it to 1 works around that glitch 13:57 Warr1024 oh, actually, it looks like the fix might be even simpler 13:57 Warr1024 I just need to defer the upscaling to right before the image is converted to a texture 13:58 Warr1024 so that the original image, for image use purposes (i.e. compositing) is unscaled. 13:59 VanessaE ah 13:59 * VanessaE waits for a patch ;) 13:59 * Warr1024 tests a candidate patch 14:03 Warr1024 success 14:03 Warr1024 check out the compositefix branch in my git 14:03 VanessaE k 14:04 VanessaE I'll apply that on top of a clean HEAD first and see 14:04 Warr1024 also, since you have a better video card than me, you can test with texture_min_size = 256 or 512 or more, to see if the sign text gets sharpened :-) 14:05 VanessaE ok. (you shoulda seen the mess it made of the text before, when I set that to 128 ;) ) 14:05 Warr1024 I also can't log onto your creative server, btw, I apparently don't have enough RAM for it. 14:05 Warr1024 my signs were mostly blank, with just a few pixels here and there, before the fx. 14:06 VanessaE yeah, 30000 and 30001 need 2-3 GB of RAM on the client, at minimum. 14:06 Warr1024 I've got 4GB total, but I'm generally using at least half that for something else at any given time 14:07 est31 Its very bad that minetest prerenders all textures 14:07 est31 It wasnt done in earlier versions 14:07 Warr1024 btw, make one of your rules "when requesting interact, also say your name backwards, to prove you read these rules" 14:07 VanessaE FAIL 14:07 Warr1024 didn't work for you? 14:07 VanessaE nope. 14:07 est31 with all that slabs stuff the list is just sooo long 14:07 est31 although its only a theory for the ram requirement 14:08 est31 btw _every_ single combination of cable is its own node 14:08 Warr1024 argh, I gotta recompile from scratch then 14:08 Warr1024 I was doing incrementals 14:08 Warr1024 but they act up for me sometimes, probably an issue with modtime stamps or something 14:08 Warr1024 just takes SOOO long to compile. 14:08 Warr1024 compared to an incr 14:13 VanessaE fwiw, 256 is as high as one ever needs to go, with a 1600x1200, 20" screen 14:14 Warr1024 VanessaE: clean build worked for me 14:14 Warr1024 you abs sure you got the right branch? 14:14 Warr1024 yeah, 256 is really nice, but just a little bit much for my GFX card. 14:14 VanessaE lemme rebuild just to be sure, then 14:15 Warr1024 though it's ironic hearing that 256 is all you need, from someone who I believe maintains a 512px texture pack...? 14:15 VanessaE are you also using your pr_guiscaling patch on top of that? 14:15 VanessaE yeah, I do maintain the 512px pack but it is indeed overkill ;) 14:15 Warr1024 no, my fix branch shouldn't contain the gui scaling... 14:16 VanessaE ok. 14:16 Warr1024 since this is a bugfix, I want a simple patch that's quick to review and merge.. 14:16 Warr1024 I better do a git log --all --graph to be sure 14:16 Warr1024 yep, this should apply cleanly to HEAD. 14:19 VanessaE oh it applies fine, but it's definitely not working - in fact it has no effect at all. 14:19 VanessaE maybe I need to clear my cache 14:20 VanessaE nope. 14:21 Warr1024 I can't see any reason why it would work for me and not you 14:22 Warr1024 this isn't in any really machine-specific or driver-specific area either 14:22 VanessaE you don't still have texture_min_size=1 set do you? 14:23 Warr1024 nope 14:23 Warr1024 64 now 14:23 Warr1024 I also tested with 256 14:23 Warr1024 the text was notably sharper, though my framerate less so 14:24 Warr1024 also tested with texture_clean_transparent both true and false 14:25 Warr1024 do you do incremental builds or clean? 14:25 VanessaE clean. 14:25 Warr1024 I actually keep my source in one place, then rsync it to a fresh temp dir for the build. 14:25 Warr1024 I don't 100% trust clean scripts 14:26 Warr1024 though I guess a git reset --hard && git clean -dfx might work, but be careful with git-clean if you RUN_IN_PLACE, as it may nuke your configs. 14:26 Warr1024 I RUN_IN_PLACE, but the IN_PLACE from which I run is a dir with a bunch of symlinks to the originals. 14:26 Warr1024 I did fix a memory leak in the code, so it might not be a bad idea to refetch. 14:27 VanessaE http://digitalaudioconcepts.com/vanessa/hobbies/minetest/screenshots/Screenshot%20-%2003212015%20-%2010%3a28%3a57%20AM.png 14:27 VanessaE this is with your patch in place, texture_min_size left at its default, aniso+mip+trilin turned on 14:28 VanessaE ok, re-cloned, re-fetched, building. 14:28 Warr1024 you're on ed36b00? 14:30 VanessaE I am now. 14:30 * VanessaE tests... 14:30 Warr1024 That's what I'm on, with aniso+mip+trilin, min_size 64, and clean_transparent on, and it looks like https://imgur.com/gCxkbDz to me 14:30 VanessaE THERE WE GO 14:30 VanessaE now it works 14:30 Warr1024 nice 14:30 VanessaE I must have got it between pushes before 14:31 Warr1024 could be 14:31 Warr1024 my auto-rebase script also probably needs work. 14:31 Warr1024 I automatically squash a bunch of commits on XXX branch into one on pr_XXX branch 14:32 * VanessaE adds pr_guiscaling 14:33 Warr1024 heh, I'm planning on creating a psuedo-fork at some point that will include all my unmerged PR's atop head, automatically rebased :-) 14:33 * est31 wonders what became out of that extruded stuff thing 14:34 est31 and what paramat has against it 14:34 VanessaE est31: you mean e.g. extruded torches? 14:34 est31 yep 14:34 est31 I mean when you drop them, they render fine 14:35 est31 just placed they can disappear when you look at them from the wrong angle 14:35 VanessaE oh that 14:35 VanessaE BlockMen has an issue for making torches into models 14:36 Warr1024 #2521 14:36 ShadowBot https://github.com/minetest/minetest/issues/2521 -- Fix composite textures with texture_min_size. by Warr1024 14:36 est31 its also farming stuff 14:36 VanessaE but we have mesh nodes now, and torchlike drawtype is used for some non-torch stuff so it can't be changed 14:36 est31 when you look from above, you always get problems 14:36 VanessaE so better to change the default torches into mesh nodes 14:37 est31 perhaps add a torchlike_legacy drawtype? 14:38 VanessaE no, can't change the existing one 14:38 VanessaE otherwise you break everything that relies on it 14:38 VanessaE and you make it impossible for those things to be used on older versions of minetest 14:38 VanessaE (e.g. older clients) 14:38 VanessaE better to make a new drawtype that won't ever have a chance of breaking existing mods/games. 14:38 VanessaE Zeno`: please review and merge #2521 14:38 ShadowBot https://github.com/minetest/minetest/issues/2521 -- Fix composite textures with texture_min_size. by Warr1024 14:39 Zeno` problem 14:39 VanessaE ? 14:39 est31 I dont know any use case where the old "it disappears" mode looks better 14:39 Zeno` I am in Windows :-o 14:39 Zeno` I have no idea how to use git in Windows heh 14:39 VanessaE est31: well having something disappear isn't desirable, but old games rely on the old look 14:39 VanessaE e.g. Minetest Nostalgia game 14:40 Warr1024 Zeno`: Take a look at msysgit and tortoisegit, they're really nice when you have to take a walk on the Windows side. 14:40 est31 they are already extruded as wielditems 14:40 Zeno` Warr1024, I am so rarely in Windows I'm not sure if it's worth the effort hehe 14:41 Zeno` so... these textures don't need to be clamped to the ceil power-of-two? 14:41 Warr1024 Zeno`: you mean in the texture fix? 14:41 Warr1024 Zeno`: I think the clamping happens in other places. 14:41 Zeno` oh it fixes signs 14:41 Zeno` I see 14:42 Zeno` Yes, it looks good 14:42 Warr1024 Zeno`: yeah, I ran into the clamping issue on android in my other upcoming PR code... android is a good platform to confirm that you're being PO2 strict when you need to be :-) 14:42 Warr1024 awesome, do we need one more coredev to okay it? 14:42 Zeno` I have some very fast PO2 ceil and floor functions 14:42 Warr1024 Jenkins and Travis are having a race to see who can verify the builds first :-) 14:42 est31 VanessaE, same argument applies for leaving the torchlike behaviour 14:43 Warr1024 Zeno`: where? I did it the naive way, myself. 14:43 Zeno` well, we need another dev to review and also to merge it :) 14:43 Zeno` Warr1024, I haven't added them to MT yet 14:43 Zeno` I suppose I could reboot :( 14:43 Warr1024 ah, are you going to add some generic reusable PO2 code then? 14:44 Warr1024 that'd be nice 14:44 Zeno` Warr1024, I already have the generic code for use in my personal projects heh 14:44 Warr1024 if you're feeling generous, MT could use something like that. 14:45 Warr1024 even something as simple as s32 nextpowerof2(s32) might be nice to have standardized. 14:45 Zeno` yeah, I've been meaning to make a PR with them for ages anyway 14:45 Warr1024 I did mine using the old while(a) { a >>=1; b <<=1; } method with a trivial-case check above. 14:47 Zeno` you can generalise this: https://github.com/Zeno-/snippets/blob/master/Esieve.c#L91 14:47 Zeno` But I already have a better function which is already generalised 14:47 Warr1024 oh, wait, there's https://graphics.stanford.edu/~seander/bithacks.html#RoundUpPowerOf2 14:48 Zeno` which seems very similar to mine, lol 14:48 Warr1024 yeah, though it'll work with something with known size like u32, but not something potentially platform-specific like int. 14:48 Warr1024 ironically 14:48 Zeno` int won't work anyway 14:48 Warr1024 since the code uses unsigned int and apparently assumes sizeof(int) == 4 14:49 Zeno` doing bit twiddling on ints is not portable 14:49 Warr1024 i.e. their example is not strictly correct to begin with. 14:49 Zeno` yeah it's not 14:49 Zeno` I'll make a PR for numeric.h tomorrow 14:49 Warr1024 you wanna add your +1 to the texture fix PR? 14:50 Zeno` yep 14:50 Warr1024 good to have a written record in case we can't find another dev while you're around to confirm it. 14:50 Zeno` done 14:50 Warr1024 thanks 14:50 Warr1024 VanessaE and I have both built and tested it. 14:51 Warr1024 ah, and we got the thumbs up from CI too 14:51 Zeno` reading it, without testing it looks correct 14:52 Zeno` oh, btw 14:52 Zeno` thank you for actually commenting your stuff! 14:54 VanessaE wow, I forgot how much performance I lose with shaders and aniso et al when I go on VE-S :-/ 14:54 VanessaE (at least at spawn - 17 fps @35m, with HDX256 and all the good stuff turned on) 14:55 Warr1024 heh, you're lucky you still even GET fps. I'd get spf. 14:55 VanessaE haha 14:56 kilbith just updated my build now, and text on sign_lib looks odd now : https://lut.im/Aj96jFaB/9SRm01aC 14:57 VanessaE kilbith: get https://github.com/minetest/minetest/issues/2521 14:57 kilbith i've missed something ? 14:57 Warr1024 kilbith: you're just a little late to the party :-) 14:57 Warr1024 no, I did. 14:57 VanessaE it'll fix that issue. 15:02 Zeno` brb 15:06 Zeno` Since #2521 fixes a critical error I'll merge in 15 minutes 15:06 ShadowBot https://github.com/minetest/minetest/issues/2521 -- Fix composite textures with texture_min_size. by Warr1024 15:06 Zeno` Unless someone finds something wrong with it, of course 15:14 VanessaE huh. one thing the engine still has a problem with is fps sags/spikes when I'm going full-tilt 15:14 VanessaE oh well 15:16 Zeno` VanessaE, I'll do some profiles again in a few days. I just hope this isn't the same thing I noticed about 2 weeks ago hehe 15:16 Zeno` bug fixes first, optimisation second 15:16 VanessaE it's really only noticeable with e.g. HDX + all the filters + shaders etc 15:16 VanessaE probably due to mesh regen hangs 15:16 VanessaE e.g. mesecons blinkyplants 15:17 Zeno` The thing I noticed was the framerate (everything, actually)... "freezing" 15:17 VanessaE yeah 15:17 VanessaE very short pauses 15:17 Zeno` and it was pretty much not noticeable using standard textures (HDX128 showed it) 15:18 VanessaE equivalent to a few frames at a tiome 15:18 Zeno` and it's very annoying 15:18 VanessaE time* 15:18 Zeno` yep, probably the same thing I saw then 15:19 VanessaE I assumed it was mesh regen from something periodically changing, hence the blinkyplant comment. meshes are already laggy as it is, and an order of magnitude worse with a HD texture pack 15:19 Zeno` nah, it happens with no mods 15:20 VanessaE hm 15:20 VanessaE wonder what it is then 15:20 Zeno` not really sure :/ It's pretty hard to track down 15:21 Warr1024 mesh regen is slower with a HD texture pack? Really? That doesn't make sense to me... 15:21 Zeno` I tried for a day when I first noticed it but haven't really looked at it since 15:21 Warr1024 unless it's e.g. reuploading the textures each time... 15:21 Warr1024 I did notice that using HD textures can increase the complexity of weilditem extrusions significantly, though... 15:22 Zeno` VanessaE, the event I am talking about is when I gave you the config for a world where I was moving into and out of jungle tree leaves (with no clip) (if you can remember) 15:22 VanessaE the extrude meshes are precalculated now - textures are sorta "plugged in" I guess 15:22 VanessaE Zeno`: OH yeah. 15:23 VanessaE I didn't see the effect there, but I sure see it when wandering around the spawn area of VE-S. 15:27 Zeno` if you still have the world config I'm sure you'll notice it now that you know what to look for ;) 15:28 Zeno` merging 2521 now 15:28 Zeno` hmm link not working 15:29 * Zeno` refreshes browser 15:29 Zeno` lol, is it just me or is github being very stupid the last day or two? 15:29 Zeno` people's avatars not showing etc etc 15:30 Zeno` anyway I can't refresh the PR 15:30 Zeno` will merge as soon as I can 15:31 Zeno` done 15:31 Warr1024 thanks 15:33 * Zeno` watches tv 15:40 est31 http://pastie.org/10043989 15:40 est31 ^ happened on 0.4.12 server, dunno what I or sb else did. 15:40 est31 wasnt even online 15:41 Tablet_One Does increasing texture_min_size increase RAM useage? 15:42 VanessaE it does a bit 15:42 VanessaE depends on how many textures are being upscaled 15:43 Tablet_One Oh... because using texture_min_size 256 on the default textures, my RAM use jumped to 78% 15:43 VanessaE what game are you using? 15:44 Tablet_One I'm on JD-PUB 15:44 VanessaE ah 15:44 VanessaE that would do it 15:44 Tablet_One Oh lol 15:45 VanessaE Jordach's games are pretty heavy but not as bad as Dreambuilder ;) 15:45 Tablet_One Heh 15:45 VanessaE Jordach: how many textures does JD-Pub have? 15:46 Jordach enough 15:46 Jordach http://jordach.net/jdpub/ 15:46 VanessaE or more to the point, what's the nominal amount of RAM a client needs for that server? 15:46 Jordach my client is just consuming over 1gb 15:47 Jordach (1,044,644K) 15:47 sfan5 Jordach: thats the outdated way of doing remote_media 15:47 Jordach VanessaE does it too 15:47 VanessaE sfan5: none of us understand the index.mth method. 15:47 Jordach and your script isn't idiot friendly 15:47 sfan5 it isn't? 15:48 Jordach https://gist.github.com/Calinou/d58bd9c1a2ed26cb0b37 15:48 Jordach not even any comments 15:48 sfan5 the index.mth method is just that each file is named by it's sha1 and index.mth contains MTHS<00><01> and all raw binary sha1 hashes 15:49 sfan5 Jordach: is it not self-explanatory? 15:49 Jordach no, it isnt 15:49 sfan5 you set the options and run the script 15:49 sfan5 the variable names of the options are self-explanatory too 15:50 sfan5 LEGACY_SYMLINKS could use an explanation but the rest doesn't need one 15:50 Jordach no 15:50 Jordach you've written "self-documenting" code which doesn't mean anything to server hosters who are getting started 15:50 sfan5 you aren't supposed to look at the code 15:51 sfan5 you are supposed to look at the part between "#Options" and "function die {" 15:52 est31 bash is standard knowledge for linux users who want to do more than click some fancy buttons 15:53 est31 When I got linux, I sat down and learned bash 15:53 sfan5 also " uses the old way to is probably the worst argument you can make for not doing something 15:53 sfan5 to -> too" 15:53 est31 I still don't know each command's every single option, but I do know how all stuff is tied together. 15:54 sfan5 Jordach: what lines should i add comments to? 15:54 Jordach all of em 15:54 sfan5 you are not being useful 15:55 sfan5 i need specific lines 15:55 Jordach you also need to add .b3d to the list of supported formats 15:56 sfan5 this is exactly why i did not want calinou to have his own version of the script 15:56 sfan5 https://gist.github.com/sfan5/6351560 contains b3d support 15:56 VanessaE http://digitalaudioconcepts.com/vanessa/hobbies/minetest/Server-scripts/update-server-mediafiles.sh does, too 15:57 VanessaE that's the full, copypaste-enfused version of what Jordach uses. 15:57 VanessaE (yeah, I got lazy :P ) 16:05 VanessaE bbl 16:10 Zeno` There is two very concerning issues with minetest. The first is "this code is not used by anything else and therefore it should not be a function" (bullshit). The second is the lack of comments... "You can see what it does by reading the code" (bullshit... a comment should be added to say what the code intends to do) 16:11 Zeno` the index.mth method is just that each file is named by it's sha1 and index.mth contains MTHS<00><01> and all raw binary sha1 hashes <--- that should be the comment 16:12 Warr1024 argh, I wish I could figure out how to view a list of PR's and issues in github that I'm watching 16:12 Warr1024 i.e. any that I've touched, commented on, mention me, I submitted, etc. 16:12 Zeno` stop with the "wahhhhhhhh it's not used by any other code so it should not be a function" and "wahhhhhh the code explains it" crap :( 16:13 Zeno` If this doesn't happen then minetest cannot progress beyond being a toy project full of code that is hard to read and hard to maintain 16:17 Zeno` seriously. 16:17 Warr1024 I usually like to use a "screenful" rule, i.e. code that can be understood without scrolling is small enough to inline... 16:17 Warr1024 though variations on screen size cause me to be inconsistent :-/ 16:18 Warr1024 Zeno`: That code I added for the transparency filter is actually going to move to a separate function, if my next pr gets merged. 16:37 kilbith PilzAdam, sfan5 : your [yay/nay] needed on game#443 please. 16:37 ShadowBot https://github.com/minetest/minetest_game/issues/443 -- Add vessels shelf by kilbith 17:12 Wayward_One is there a new setting i need to change? sign text is messed up for me with the latest git 17:15 paramat #2501 is still a bug in latest dev 17:15 ShadowBot https://github.com/minetest/minetest/issues/2501 -- creative & damage disabled: tnt deals damage and kils player 17:16 Calinou Wayward_One, update to latest Git & recompile 17:16 Calinou there has been a fix 17:19 paramat the clear registered ores bug is confirmed fixed, thanks 17:24 Wayward_One ok, thanks Calinou 17:30 Jordach paramat, you're biomedev works great with current git 17:31 Jordach i'd also like those trees availble via saplings 17:31 paramat cool man 17:31 paramat oh yeah.. 17:32 paramat i forgot about saplings =/ 17:42 paramat as the tree designs are different we may need a second set of saplings 17:43 paramat also i need to write code to create .mts files from lua table defined schematics 17:45 est31 yaaay now I wanted to confirm whether somebody uses the official app, and no I cant as the server isn't compiled in a debug mode 17:45 est31 yes really great 17:46 est31 Its already hard enough to describe the official app to them 17:49 paramat biomesdev still has a mysterious bug, patches where the trees dissappear for no apparent reason, very noticeable in the jungles which are supposed to be consistently dense 17:51 Wayward_One hmm, just git pulled and sign text is still messed up :/ 17:52 hmmmm hrmm 17:52 hmmmm warr, where are you! 17:52 hmmmm darnit 17:52 hmmmm in the meantime you could just set texture_clean_transparent = false 17:53 hmmmm paramat: is the clear ore/biome/decoration api still broken for you?? 17:53 Wayward_One oh, ok 17:54 Wayward_One thanks hmmmm 17:55 hmmmm aside from the sign text, do you like the texture cleanup? 17:55 paramat hmmmm, the 'clear registered ores' bug is confirmed fixed 17:56 hmmmm k good =] 17:56 hmmmm that's what happens when people make changes to other peoples' code and they don't test it at all 17:57 Wayward_One i haven't noticed anything different with the textures 17:58 hmmmm you run with antialiasing and either bi or trilinear filtering, no? 17:58 Wayward_One ... 17:58 hmmmm ok 17:58 Wayward_One well there's my problem xD 17:58 Wayward_One trying it now 17:58 est31 paramat, why do you think extruded torches are bad? 17:59 paramat i bumped issue #2501 up to high priority, anyone know how to fix? 17:59 ShadowBot https://github.com/minetest/minetest/issues/2501 -- creative & damage disabled: tnt deals damage and kils player 17:59 paramat hi, i explained in the PR 18:00 nrzkt paramat, only with breath ? 18:00 hmmmm Warr1024: maybe it would be a good idea to tune texture_min_size = 64, and texture_clean_transparent = false, and then in the texture loading code it'll auto-adjust the defaults if anisotropic is on.. sound good? 18:00 paramat nrzkt, didn't test with other damage types 18:01 nrzkt to be sure can you try to take damage from TNT , because this is solved. For breath i'll look at this soon 18:02 paramat ah TNT would be lua damage so i guess different 18:03 est31 paramat, so is your main point that 3d torches are bad because they are just fancy stuff nobody needs, or is it more technical issues (whether to extrude the torchlike drawtype, or make meshnodes). 18:04 paramat est, the plantlike drawtype is so extremely simple and abstract it's just *wrong* when extruded to be fake-3D. so yeah 3D objects should be designed meshnodes 18:05 paramat i have nothing against 3D torches or plants, i love OCD plants 18:05 paramat and the Vox game 18:06 BlockMen paramat, 3d torches with meshes are possible, but you need 2 meshes and 2 nodes and then the placing need to be done serverside, which causes lags 18:06 BlockMen and furthermore there is needed an abm to convert all current torches to not break existing maps 18:06 BlockMen doing 3d torches with meshes is shit! 18:07 paramat the plantlike drawtype should stay flat, and also i feel should remain at 45 degree angles to voxels 18:07 BlockMen i agree that the plantlike drawtype should stay flat 18:07 Wayward_One i think i'm missing something... tried the new settings with trilinear filtering and antialiasing enabled, and can't see a difference. also, sign text is still broken for me 18:08 paramat extrusion for torches only seems okay to me now, for those reasons, but nothing alse 18:08 paramat (else) 18:08 BlockMen a new drawtype would be the best for 3d torches 18:08 paramat although i have nothing against enabling plantlike rotation for modders 18:09 paramat yeah 18:09 est31 my main issue with missing torch (and plant) extrusion is that when you look at something from above, it almost dissappears. 18:10 est31 and the "fake" 3d is some "blockification" of the plant/torch 18:10 paramat well that's just Minetest. only happens when precisely above, you always know something's there as you move 18:11 paramat yeah microvoxels are fine but should be done with meshnodes and designed from scratch 18:12 est31 ok thanks paramat for clarifying your position on this. 18:14 paramat i just find extruded textures tacky and tasteless, it's sort of an artistic/design issue for me =) 18:17 paramat so yeah i feel plants should remain at 45 degrees in the world, otherwise it's like rotating voxels to random angles, simply should not be done in a voxel world. the 45 degree angle of plantlike drawtype is ery much part of the character of a voxel world 18:17 paramat (very) 18:17 paramat but enable it for modders 18:18 est31 btw what do you mean with 45 degrees? 18:21 paramat the way plantlike drawtype is at 45 degrees to the world grid when seen from above 18:21 paramat = diagonal 19:54 rubenwardy https://github.com/skywinder/github-changelog-generator 19:55 est31 why did I knew it was ruby even before I clicked the link? 19:56 rubenwardy Lol 19:59 paramat hmmmm, #2519 is ready to go if you want to review it, but no rush i'll push it later or tomorrow 19:59 ShadowBot https://github.com/minetest/minetest/issues/2519 -- Mgv7: 1 up , 1 down overgeneration for chunk border continuity by paramat 20:13 Jordach why does the master server not pickup on the --port minetestserver flag 20:31 nrzkt #2523 is good. Will merge it in ~10 min if no problem (a little SendInventory improvement which is good). 20:31 ShadowBot https://github.com/minetest/minetest/issues/2523 -- SendInventory on Use (not just Place) by poet-nohit 20:32 Calinou when was this broken? 20:32 Calinou (the eating) 20:32 nrzkt it was related to this problem ? 20:34 nrzkt if don't know when, but if it's related to this function, maybe it can be related to 1b2f64473ed4f222d3b7f02df853730d4382105e and will be fixed by 2523. 20:35 nrzkt i will look at all setWieldedItem to be sure there is no new missing call 20:36 nrzkt no other than the PR fix. This fix is perfectly correct 20:37 nrzkt it's a missing SendInventory , it's right. I merge it now, no need to check another time, each setWieldItem call SendInventory except the call mentionned by the PR 20:38 hmmmm paramat: I'll take a look 20:38 hmmmm paramat: I saw you added a heightmap to mapgen v6 but you didn't replace findGroundLevelFull() calls in placeTreesAndJungleGrass() 20:39 paramat ah good catch 20:39 hmmmm I think we should fix any instances related to the mapgen where myrand_range() is being used 20:40 hmmmm they really should not be based off the global PCG; I don't know why they're like that but I am pretty sure I didn't do that 20:40 hmmmm aside from the usages in schematics, that's my laziness for not adding a 'seed' parameter :/ 20:41 paramat i'll use the heightmap in 'place trees and junglegrass', i'm on it 20:41 nrzkt pushing 2523 now 20:42 hmmmm :) 20:42 hmmmm thanks 20:44 paramat so i guess myrand is not pseudorandom determined by seed, so yes i agree 20:55 celeron55 hmmmm: what is this?https://github.com/minetest/minetest/pull/2459 20:55 celeron55 hmmmm: what is this? https://github.com/minetest/minetest/pull/2459 20:55 hmmmm what does it look like 20:56 celeron55 i don't know, it looks childish and unappropriate 20:56 hmmmm it's a political statement 20:56 celeron55 what do you expect minetest to gain from it 20:57 hmmmm stability, an increase in network performance, a few bugfixes 20:58 hmmmm i'm sorry to say but his network changes were shit 20:58 nrzkt as i said, there are no problem of stability in the network stack, if you remember. 20:58 hmmmm i think they had nice ideas behind them but their implementation was poor. 20:59 celeron55 do you have intention to merge the PR yourself 20:59 hmmmm no 20:59 hmmmm obviously that PR can't be merged. at this point in time it's very difficult 21:00 nrzkt you PR is only a revert which doesn't help anybody to do anything. 21:00 hmmmm it could have been done for real, if merged two weeks ago 21:00 hmmmm nrzkt, I found the best way to deal with your shit commits is to revert them 21:00 hmmmm you don't discuss anything with anybody prior to merging things and you keep justifying yourself under the guise of them being trivial (they're not) or they're under your domain (they're not) 21:01 hmmmm you broke HP 21:01 nrzkt please be polite, and we could talk. If you don't want to be polite i don't need to talk with you. You think you are minetest saver and you think you want to manage the team whereas you only complain on network when i'm away and you don't proove anything. 21:01 hmmmm you broke ores/decorations/schematics 21:01 nrzkt HP aren't due to network stack. Ores are not broken by myself 21:01 hmmmm you broke unreliable packet sending 21:01 nrzkt no 21:01 nrzkt no 21:01 nrzkt no 21:01 nrzkt and NO. 21:01 Calinou https://github.com/minetest/minetest/pull/2459#event-261321271 21:01 Calinou what a kid 21:01 Calinou we know GitHub issues is incredibly exploitable, this isn't a reason to do this 21:02 celeron55 i'm going to allow that PR to be open as long as some of the non-vague and non-ridiculous reasons listed in the comments stay valid 21:02 hmmmm nzkrt, I would be really happy with you if you weren't so utterly arrogant, posessive, and thoughtless 21:03 nrzkt and hmmmm: why do you edit my answers on github ! 21:03 hmmmm arrogant because you have never failed once to admit a mistake with your code, even when there's very obvious evidence 21:03 hmmmm instead you'll say, "well then why don't you fix it?" 21:03 nrzkt hmmmm: stop edit my answers on my PR. 21:03 hmmmm didn't edit anything 21:03 hmmmm that's how stupid you do sound in reality 21:03 celeron55 both of you are arrogant and posessive, frankly 21:04 hmmmm when there's a problem with my code, I always give the reporter the benefit of the doubt 21:04 hmmmm and I do not tell them to fix it on their own 21:04 hmmmm and I don't tell people they're not allowed to edit mapgen code 21:04 hmmmm the one time you did edit mapgen code, though, you broke it 21:05 nrzkt if it's not you, who edit my last comment on PR 2459. I refix it but i said "No, i was the network maintainer", and when the last link was done on IRC i pointed our the "i'm the network maintener" disappear, just when you re-open the issue. Strange in fact, no ? 21:05 hmmmm you didn't even test to see if your changes worked 21:05 hmmmm you just changed it, it compiled, and then you committed a consistent, instantaneous crash whenever a feature is used 21:05 nrzkt no. 21:06 hmmmm I reverted your commit that caused it, and voila, it works all of a sudden 21:06 nrzkt stop talking and saying stupid things. I'm doing more tests than you think. 21:06 hmmmm nrzkt, yes. this is reality 21:06 hmmmm I can't believe you're denying something that has already happened and is done with 21:06 nrzkt i agree the commit you revert in fact, because you are right. But stop telling we must revert all my work. 21:06 nrzkt you also introduce bugs in mapgen sometimes. 21:07 hmmmm and I fix them right away 21:07 nrzkt me too. then stop 21:07 hmmmm no you don't 21:07 hmmmm you put them in your 0.5-dev branch and say 21:07 hmmmm "nope, not fixing it here, it'll get fixed in 0.5-dev" 21:07 nrzkt and you also removed a comment on this PR. that's very strange in fact. 21:07 hmmmm christ, you're worse than proller even 21:07 hmmmm what did the comment say 21:08 celeron55 what's up with this PR comment thing 21:08 nrzkt you have a good strategy, remove the communication process to let the PR say what you want, good guy :) 21:08 celeron55 this is the only thing that still seems a valid concern in this discussion 21:08 hmmmm in fact 21:08 hmmmm I'm going to do that right now 21:08 hmmmm I'm going to delete all your inflammatory postings on that PR, and I'm going to delete my responses to them 21:08 nrzkt celeron55, agree. 21:08 celeron55 i'm going to remove anyone from being a core developer who edits github comments and/or actions of other people 21:09 nrzkt let it close and stop this discussion. Find real bugs, point them, not telling concepts, we are doing code, please 21:09 hmmmm nrzkt: I don't understand what problem you have with discussion 21:09 celeron55 (but i can never prove it) 21:09 nrzkt celeron55: maybe the github API have some logs ? 21:10 hmmmm it's too late, I already deleted everything 5 seconds after I said I would 21:10 hmmmm because it's unproductive and looks stupid 21:10 hmmmm anything that doesn't have specific details about issues or whatever should go 21:11 nrzkt if you want to help to improve the network stack you are welcome, but reverting and telling "it's shit" is not a good argument. 21:11 hmmmm rules: 21:12 hmmmm - don't close that PR anymore 21:12 hmmmm - don't add any tags to that PR anymore 21:12 hmmmm - don't comment anything that isn't constructive 21:12 nrzkt this PR mustn't be opened. 21:12 nrzkt it's not mergeable. 21:12 nrzkt and will not be merged. 21:12 hmmmm it just needs to be rebased 21:12 hmmmm nrzkt, what I would like to know is how you got such an attitude 21:12 nrzkt no hmmmm, you won't revert the network stack. 21:13 hmmmm why do you feel as if you have the final word here 21:13 nrzkt because you work on mapgen and me on network. 21:13 nrzkt i'm the network maintainer. you're the mapgen maintainer. 21:13 hmmmm the? 21:13 hmmmm so I can't commit anything else? 21:13 hmmmm I've done more work elsewhere than I have the mapgen 21:13 nrzkt /applause 21:14 hmmmm those labels are arbitrary groupings referring to what your area of expertise is 21:14 nrzkt and i'm working on the network more than other devs who are connected here, but... and then ? 21:14 hmmmm the reason why they were created was to make it easier for people who are frequent committers to that area and are more knowledgable of it than anybody else 21:14 hmmmm no you're not 21:14 nrzkt please reread the rules. 21:14 hmmmm you're a newcomer who doesn't look at the code you're modifying before you do it 21:15 hmmmm you just assume that your changes will work and that's it 21:15 hmmmm show me the rules that contradict what I say 21:15 nrzkt i'm using a good IDE which permit to point the modifications area, but if there is a missing thing in the code which is hidden by a bloated test in the AsyncRunStep, sorry i can't see it when i do all my tests, and i do many tests :) 21:16 hmmmm then how did you miss the clear_registered_ores bug? 21:16 hmmmm can you please enlighten me 21:17 hmmmm it's a consistent crash that happens every time any of those apis are called, DIRECTLY resulting from your modification of that code 21:17 nrzkt http://dev.minetest.net/Organisation 21:17 nrzkt and this is fixed in master ? 21:17 hmmmm yes, by reverting your shit 21:17 nrzkt stop telling "my shit", and we can talk stupid guy 21:17 hmmmm I'm convinced the best way to deal with your broken nonsense is to just revert it all 21:18 nrzkt no, because you are just angry and you think every core dev must think like you. 21:18 hmmmm maybe if you had a better attitude we could work on it 21:19 nrzkt you mean here: "if you agree with all my rules". It's like i always saw you on IRC with every dev. You want to manage everything like if you are essential, and when you left the team you connects on IRC when i'm away to say, "i must save minetest". SOrry i didn't agree with that, you are only a core dev like others. You are the mapgen maintainer, but a core dev like others. 21:20 hmmmm all my rules? you mean getting approval for your commits before committing them is my rule? 21:20 nrzkt and everybody can make mistakes, me , you, anybody. We are humans, not bots 21:20 hmmmm I must save minetest because it was going in a bad direction when I left 21:20 celeron55 i'm going to say something extremely boring: the actual truth and the actual right thing to do resides pretty much exactly at the halfway point between what hmmmm and nrzkt think 21:21 hmmmm you were turning upstream into some sort of playground for you to break everything 21:21 hmmmm no. if you want an unstable playground, Freeminer is your place 21:21 nrzkt there is no real direction. Where is the roadmap for a future version for our users ? 21:21 hmmmm there doesn't need to be a "roadmap" 21:22 hmmmm this isn't business software, this is just an open source project 21:22 hmmmm where some people can commit some cool features and everybody's happy 21:22 nrzkt i have my own fork for my playground, sorry, you must clean your glasses :p and sorry it's based on minetest and it's perfectly stable :) 21:22 hmmmm you're acting like this is some kind of business 21:22 nrzkt because we have users. 21:22 hmmmm in any case, if I were to actually fix your commits that'd be a full time job 21:22 nrzkt you want retrocompat for users no ? It's business ? No. 21:22 hmmmm clean up after the baby's mess 21:23 hmmmm I want retrocompat for the users because I *respect* them 21:23 nrzkt if you respect them devel you mapgen and let me work on my network. 21:23 hmmmm ... 21:23 hmmmm that's a non-sequitor 21:24 nrzkt sorry i don't understand the word "sequitor" 21:24 hmmmm oh well then 21:25 hmmmm the reason why I said I need to save minetest is because I'm the only one with large enough balls to call you out on your BS. 21:25 PilzAdam praise hmmmm's balls 21:25 hmmmm I'm not the only one here who feels strongly about the recent turn of events, by any means. I'm just the most vocal 21:26 jin_xi its a pattern of people arriving eager to contribute and not familiar with mt code and history leading to bugs and alienation all around 21:26 nrzkt hmmmm stop think you are minetest god. It's a free software we are a community. 21:26 nrzkt and nobody is perfect, you too. 21:27 hmmmm jin_xi: good summary... 21:27 hmmmm at first I left because I just didn't want to deal with it. I don't like arguing 21:27 nrzkt it's not an argument in your favor, it's a fact :) 21:27 VanessaE see also, NFG Syndrome 21:28 nrzkt i'm not perfect too, and when some code was cleanup maybe there is a loop in the main thread which hide a un-seeable race, that can happen. The network is a little bit different with mapgen, it's also why i want to rewrite all the stack, it's because we are too complicated. 21:29 hmmmm open source projects don't usually take too kindly to newcomers making huge, sweeping "cleanups" and then expecting them to get merged 21:30 jin_xi see mgv7 for an earlier example 21:30 hmmmm it was made reverse compatible though 21:30 hmmmm and it's still not the default 21:30 nrzkt in fact hmmmm if you see the network stack it's not simple to do retrocompat on it, but i think you don't read the low level code, you only argue on the fact there are issues somewhere without a real code line in your hand. 21:30 hmmmm back then, I should have been put in my place though, I agree 21:30 hmmmm that's not the correct way to do things 21:31 celeron55 hmmmm: you need to provide a reasonable path for newcomers to actually get into the project though; you can't just skip that and watch as the project dies 21:31 hmmmm nrzkt, if you knew the exact line of code breaking things, that'd be already 70% of the work of fixing the bug 21:31 celeron55 or, you can, but it's dumb 21:32 nrzkt if you want me to look at the ore problem, please tell me how to reproduce it and i will fix it soon if you think it's network. 21:32 hmmmm I don't think that all newcomers want to jump in and make some huge sweeping rewrite 21:32 nrzkt there is no huge rewrite atm 21:32 nrzkt the huge rewrite will come later, because i need to write all the stack and do many tests on it 21:32 hmmmm just the arrogant ones who experience Not Invented Here syndrome 21:33 jin_xi i think this project lacks appropriate channels for discussion prior to writing code 21:33 nrzkt jin_xi: agreed :p 21:33 nrzkt jin_xi: we must create #minetest-fight 21:33 exio4 there were suggestions for having a mailist list 21:34 PilzAdam hmmmm, can you help nrzkt fix the bugs he introduced? 21:34 celeron55 we have three places for having those discussions (irc, forum, github), but nobody is around to listen to anyone when they want to discuss about new code 21:34 hmmmm PilzAdam: I can, sure, but I'd rather work on my own things 21:35 hmmmm look at prestidigitator for example 21:35 hmmmm all he did was come around and rewrite two files in his own style, with no actual enhancements, and expects us to just merge that 21:35 hmmmm then start crying when we don't 21:35 jin_xi irc has temporal problem, forum is too noisy and github is mostly good for smallish stuff about already written code 21:36 hmmmm this is what I'm talking about the NIH syndrome 21:36 hmmmm with the& 21:36 PilzAdam hmmmm, cleaning up the networking code was on our todo list, though 21:37 hmmmm and cleaning it up is fundamentally fine 21:37 celeron55 jin_xi: i'm fine with someone setting up a mailing list 21:37 hmmmm like I was saying 21:37 hmmmm the changes to the network code are fundamentally good 21:37 hmmmm just poor in execution 21:37 nrzkt then let me do my work and please help if you have ideas. 21:37 celeron55 people have been suggesting a mailing list for like 4 years but nobody has actually ever made one 21:37 nrzkt and if you haven't looked at our network stack, i invite you to look at it 21:38 PilzAdam hmmmm, don't just blame people for doing stuff wrong, help them 21:38 hmmmm but he's the 'network maintainer' 21:38 hmmmm he's supposed to be responsible for making it work 21:38 nrzkt i'm perfectly responsible of what i do. 21:38 hmmmm okay good then 21:39 VanessaE ok I'd like to interject: 21:39 VanessaE who's responsible for fixing the flat player glitch then? 21:39 VanessaE (players signing in and either coming up as "the green guy" or as a flat player sprite but with the skin texture that belongs on the 3d model) 21:39 hmmmm the one who broke it :) 21:39 nrzkt if there is a bug i work on it, and if somebody say it's not the case, please shut up, especially VanessaE for who i worked ~ 24 hours this month to fix problems. 21:39 VanessaE let's just start with that one bug. 21:40 PilzAdam hmmmm, you spend the last hour telling that nrzkt's code contains bugs 21:40 hmmmm it was more general talk 21:40 PilzAdam we are all very surprised by the fact that code can contain bugs, but this doesn't help anyone 21:40 VanessaE hmmmm: clearly :P 21:41 PilzAdam you could have just spend the last hour helping nrzkt to find / fix bugs 21:41 hmmmm PilzAdam: the discussion was more about what the appropriate response should be when a bug is found by somebody 21:41 VanessaE but can we stop this argument and focus on finding and fixing these issues? or at least identifying who broke what so that at least the blame is legit? 21:41 hmmmm I too easily get sucked into unproductive conversations 21:41 PilzAdam then stop now 21:41 nrzkt in fact the problem is not WHO cause the cause, the real problem is WHICH CODE cause the bug. 21:42 nrzkt cause the bug* 21:42 VanessaE nrzkt: which, in turn, is the responsibility of whoever wrote that code. hence the person who wrote the code is the person who broke it, and therefore that person should fix it, in theory. 21:42 hmmmm aahhh * 21:42 hmmmm see that 21:43 hmmmm that's continuing on the same conversation we just agreed was unproductive 21:43 hmmmm idk about you guys, but I'm going to start hiding my IRC window 21:43 nrzkt VanessaE: maybe but that doesn't mean nobody else can fix it if the commiter is not here or stucked ? 21:43 VanessaE (and yeah, I just realized how repetitive that sounded :P ) 21:43 sfan5 this was probably talked about, but who removed Zenos and my comment on #2459? 21:43 ShadowBot https://github.com/minetest/minetest/issues/2459 -- Reset all network-related changes by kwolekr 21:44 hmmmm sfan5, I deleted everything that wasn't directly related to specific issues 21:44 sfan5 i see 21:44 hmmmm that's probably the only time I'd ever delete github comments 21:44 hmmmm because it's a shitshow and doesn't belong there 21:45 VanessaE nrzkt: of course not. anyone interested in a given bug has the authority to fix it (or at least propose a fix), but whoever originally caused the bug should be the one to fix it, if that person is active. "You broke it, you bought it." 21:45 PilzAdam nrzkt, do you have a list of bugs that are / might be related to your changes? 21:45 nrzkt VanessaE, i think i must retrieve the original commiters for some bugs, but maybe they leave the project two years ago , then :p 21:45 VanessaE nrzkt: I said "if that person is active"... 21:45 nrzkt PilzAdam: no. No issue on github 21:45 VanessaE "leave the project" != "active". :P 21:46 nrzkt the only bug i'm working on it is #2501 21:46 ShadowBot https://github.com/minetest/minetest/issues/2501 -- creative & damage disabled: tnt deals damage and kils player 21:46 PilzAdam hmmmm, can you help nrzkt create this list of bugs? 21:46 nrzkt because i know where does it comes, but i don't have the solution yet 21:46 PilzAdam hmmmm, since you seem to have an oversight over this 21:46 VanessaE PilzAdam: that flat player glitch I mentioned may or may not be network related. it's unclear now. 21:47 hmmmm - HP bug 21:47 hmmmm - flat player glitch 21:47 nrzkt i don't think it's network related because packets are forged with the same way. If there is a bug on a packet you will have a corruption on all packets 21:47 hmmmm - packets sent as unreliable now being sent as reliable 21:47 hmmmm - packets being copied before sending 21:47 celeron55 if people promise to be around often enough to look at PRs, i could remove nrzkt's network maintainer status if that helps people maintain their trust to each other 21:47 hmmmm - (unconfirmed) lower network performance 21:47 nrzkt hmmmm: please create issues on github with facts 21:47 celeron55 i'm not convinced of that being the case though 21:47 nrzkt lower network performance is wrong. 21:47 VanessaE players being spawned at 0,0,0 even when there's a static spawn point may be network related also 21:48 hmmmm that one I'm just going off of peoples' reports to me about it 21:48 hmmmm yep, there 21:48 nrzkt VanessaE: i looked at this , but it's not the case as i saw. The packet is strictly identical 21:48 PilzAdam nrzkt, can you work with that list? 21:48 hmmmm basically if it's a bug that popped up in the past month or so, it's probably caused by nerzhul 21:48 VanessaE nrzkt: the packet's content may be correct, but what about when the packet arrives relative to other data? 21:49 hmmmm - bad spelling, grammar in comments, and poor code style on network related code :-P 21:49 nrzkt no. I need issues and also i need more informations. For the 3rd point it's already fixes. For the 4th point it's on the road but not merged/proposed 21:49 hmmmm we also need to go through all of the instances where lists were changed to vectors to verify that they work correctly 21:49 hmmmm also we need to verify they aren't performance regressions as is the case with several of the moves 21:49 nrzkt stop trolling, i'm not english my english is not perfect, stop believe you can laugh on this. 21:50 PilzAdam hmmmm, can you create a new issue on github, with the list of bugs, and close #2459 ? 21:50 ShadowBot https://github.com/minetest/minetest/issues/2459 -- Reset all network-related changes by kwolekr 21:50 hmmmm why 21:50 hmmmm why is that being open such a bane for everybody 21:50 celeron55 because it's not helpful 21:50 PilzAdam PRs that don't get merged get closed 21:50 nrzkt hmmmm: for the convertion, the only commit which could lead problems but i avoid many problems was the commit you reverted , others doesn't use those methods, they are stupid iterators in loops 21:51 VanessaE hmmmm: just a guess, because it starts with the premise of a huge revert. if it started as a bug report with suggested problem commits, it might be more useful. 21:51 PilzAdam and an issue with a list of bugs is better structured 21:51 celeron55 hmmmm: if you actually want to get those issues solved, they need to be reported separately 21:51 celeron55 hmmmm: now you're just wanting to make a war 21:51 hmmmm :\ 21:51 celeron55 and i also hate nrzkt's sloppy coding and testing 21:51 hmmmm I'll close it if we get a decent place for discussion of these matters 21:51 celeron55 but i am able to weigh on other benefits 21:52 PilzAdam I'll create an issue then 21:53 nrzkt thanks PilzAdam, you are helpful :p 21:53 nrzkt don't report HP bug, there is already an issue for it. Don't report the packets reliable/unreliable, it's already fixed. And the lower performance is not mesured 21:54 celeron55 you can report all of those if you think they need to be re-evaluated 21:54 celeron55 github will happily waste their HDD space for us 21:55 nrzkt VanessaE: please note i will not fix your problems since you will be more polite and less offending. 21:55 VanessaE ? 21:55 hmmmm vanessae's problems? 21:55 VanessaE wat 21:55 hmmmm minetest users' problems 21:55 celeron55 lol 21:56 VanessaE wait, so I have to insult you to get my bugs fixed? ;) 21:56 hmmmm nrzkt: I don't think that's the only instance of reliable/unreliable packet regressions 21:56 hmmmm it's just one that I happened to spot as I was looking at an unrelated piece of code 21:56 nrzkt scream on me in MP everytime and tell i'm a retard on IRC is not very elegant. 21:56 hmmmm you need to be way more careful about the code you change. 21:56 PilzAdam #2524 21:56 ShadowBot https://github.com/minetest/minetest/issues/2524 -- Issues with new network code 21:56 nrzkt thanks PilzAdam 21:57 kilbith "Players being spawned at (0,0,0)" is an old bug from years now 21:57 VanessaE kilbith: hence the "(unconfirmed)" 21:58 PilzAdam kilbith, comment that in the issue 21:58 PilzAdam it might be helpful in the future 21:58 VanessaE it's unclear if that bug went away and just came back, or what's the case there 21:58 kilbith i had a server hosted by OldCoder and i had exactly this bug happened 21:58 nrzkt yes, but don't say it comes from me. 21:58 kilbith one year ago ^ 21:58 nrzkt somebody can think i introduce this bug. 21:59 celeron55 nrzkt: note that VanessaE happens to be one of our most dependable testers of heavy usage of minetest; it would be quite dumb to not take advantage of the information she can sometimes give 21:59 hmmmm I just wasted a crapload of time to discuss bugs with somebody elses' code instead of working on my own code 21:59 PilzAdam nrzkt, you'd be better of if you just ignore everyone who just blames you 21:59 VanessaE [03-21 17:50] players being spawned at 0,0,0 even when there's a static spawn point may be network related also <--- notice I said "may be network related" and not "is nrzkt's fault" or anything even close. 21:59 PilzAdam nrzkt, as long as you take bug reports seriously and are eager to fix them 21:59 est31 revert all commits until one year ago. obviously there is an error with them 21:59 PilzAdam hmmmm, shame on you for working together with other people 21:59 est31 and its not my, code, so I'll send a PR to revert, and not to fix. 22:00 nrzkt PilzAdam, i do it, but at a moment my patience is submerged 22:00 celeron55 hmmmm: you can use these discussions beneficially by giving motivation to everyone 22:00 celeron55 hmmmm: which has a way greater end result than one person working for one hour 22:00 nrzkt PilzAdam, please look at github logs if you think i don't fix anything, i also look at many PR and talk with maintainers to make the PR mergeable. 22:01 PilzAdam nrzkt, I didn't say otherwise 22:01 est31 but I'm happy you partially resolved it. 22:01 est31 (the dispute) 22:02 PilzAdam nrzkt, now have fun fixing the bugs; I'm out 22:02 nrzkt i don't get time to develop at this hour (it's 11PM here), and i'm another project :p 22:02 nrzkt i'm on* 22:03 celeron55 yeah the thing is, it's not enough to fix "many bugs"; you need to fix as many of them as possible, as thoroughly as possible, before continuing on other things; it's the only way a large project can survive 22:04 nrzkt celeron55: agree it's what i'm doing everytime, fix the bug, improve some performances, talk with community about needs, talk about PR and issues to find bugs :) 22:10 hmmmm the issue with creating a roadmap is that nobody here is guaranteed to have time to work on minetest 22:11 hmmmm try adding *that* to the already high level of uncertainty in software development 22:11 nrzkt we don't have a deadline, there are only ideas and some things requested by community. Every point can be reported if necessary 22:11 VanessaE minetest also has a history of taking sudden detours that would render any roadmap somewhat less useful than intended, as well. :) 22:12 nrzkt VanessaE, then we update the roadmap if it's no the same way :) 22:23 nrzkt VanessaE: sorry i misunderstand what you said on NFG, translators are poor sometimes :( 22:23 VanessaE nrzkt: indeed so 22:24 VanessaE also I got the letters out of order, which confused your translator and google. 22:24 nrzkt then: working together but don't forget to remove your CAPS LOCK :p 22:37 * VanessaE makes a point to insult nrzkt more often... it results in jokeworthy responses sometimes ;) 22:42 nrzkt this doesn't work sorry :p i prefer calm persons with reproducable conditions :) 22:42 est31 NFG? 22:44 VanessaE s/NFG/FNG/ 22:44 VanessaE http://en.wikipedia.org/wiki/FNG_syndrome 22:44 kilbith anyways, thanks both of you, nothing at TV tonight 22:44 * kilbith runs away 22:44 VanessaE haha 22:48 hmmmm I thought it meant No Fuckin Good 22:49 VanessaE now now 22:49 VanessaE let's not start up again 22:49 nrzkt #troll spotted :p