Time Nick Message 05:04 hmmmm jesus christ schematics are so fubar 05:04 hmmmm two memory leaks within 10 lines of eachother 05:04 hmmmm =[ 06:58 hmmmm I've changed my mind on adding update/delete operations for nodedef 07:49 Zeno` hmmmm, I've been profiling the client and I really think that the exceptions in getNode() need to go. They are being used in lieu of conditions for the majority of cases and the function spends >90% of its time handling exceptions 07:50 Zeno` I've reimplemented stuff so that they are not used and will profile again (each profile takes about an hour or so :/) to see if there really is a difference 07:50 hmmmm are you looking at calcLighting 07:51 Zeno` no collision handling really 07:51 hmmmm in any case there is getNodeNoEx 07:51 hmmmm you just need to use it 07:51 Zeno` that's mostly what I did 07:51 hmmmm celeron was exception-happy with the original code 07:51 hmmmm i agree, it wasn't the greatest of choices. 07:51 Zeno` Once I get the results of the new profile session I'll post the results somewhere 07:58 RealBadAngel https://github.com/RealBadAngel/minetest/commit/3ed109d307bdab1791c5e9bfabc1166ff2a73977 07:58 RealBadAngel fixes for wielded, WIP 07:59 RealBadAngel everything works ok for default textures, there are some glitches with high resolution ones and shaders enabled still 07:59 RealBadAngel but all of the drawtypes are working ok 08:04 Zeno` I'll test in a few hours if you like RBA 08:04 Zeno` But... I didn't experience the first issue so I dunno how much testing I can actually do :) 09:38 kilbith bug : https://cdn.mediacru.sh/zs_4Z2o03D5F.png 09:39 kilbith probably related with the latest commit 'Serverlist: announce mg_name' 09:40 kilbith ^ kahrl 10:25 Zeno` I like that screenshot 11:06 Zeno` are you a time traveller? 11:18 kilbith who ? me ? 11:19 kilbith eh no, john 12:27 kahrl kilbith: can't reproduce that 12:29 Zeno` kahrl, refactor_main is "testing" again :p 12:32 celeron55 i actually had never used exceptions in C++ when i started minetest, so i chose to try them out in various extents... yes, not the best of ideas 8) 12:32 Zeno` =D 12:33 Zeno` anyway, profiling (again :/) 12:34 Zeno` celeron55, at least they're easy to get rid of in this circumstance 12:45 Zeno` with exceptions removed I can almost play when running under callgrind 12:48 jin_xi Zeno`: are you just looking at map access in collision handling or all of collisionMoveSimple? i ask because it produces wrong collisions which you can see if you bump particles radius and expiration time enough 12:49 Zeno` looking at it all 12:49 jin_xi i have a shitfix for the issue in the irrlicht particles wip PR which might help figureing out whats wrong with it 12:50 Zeno` I'll clean up these changes and make a branch for testing 12:50 Zeno` But I'll show you some pictures if you want 12:50 jin_xi no need to do extra work for me ;) 12:52 Zeno` I will anyway 12:53 kahrl Zeno`: I'm still trying to figure out what to do with current_playername and so on 12:53 kahrl it's really ugly to have these kinds of duplicate variables\ 12:53 kahrl -\ 12:53 Zeno` kahrl, do you see why I added a TODO comment? :D 12:54 Zeno` or maybe it was FIXME 12:54 kahrl yeah 12:54 kahrl btw you can remove the one above setup_log_params ;) 12:56 Zeno` oh, probably yeah 12:56 VanessaE kaeza: about kilbith's report, I wonder if it's related to this: take a look at the server list (I mean the site at servers.minetest.net).. many of them don't show the mapgen at all (just "?"), even where they're clearly using mgv6, and those servers have no mod list popover. Any clue why? 12:56 VanessaE fuck. 12:56 VanessaE kahrl: ^^^^^ 12:57 Zeno` with exceptions (1 image): http://i.imgur.com/1Ps1Rvt.jpg 12:58 Zeno` without exceptions (2 images): http://i.imgur.com/RnT1WJ3.jpg and http://i.imgur.com/SazNHG1.jpg 12:58 Zeno` it's "slightly" faster (I am understating slightly) 13:00 kahrl VanessaE, wtf 13:00 kahrl that doesn't make any sense at all 13:00 Zeno` those are for the client btw, not the server 13:01 VanessaE kahrl: should I hide? :P 13:01 kahrl VanessaE: some of them have be running longer than when I pushed the patch 13:08 kahrl almost all of them have a high update time though 13:10 kahrl s/update time/uptime lol 13:10 kahrl so I don't think many of them are affected by my patch 13:10 VanessaE kahrl: well, not saying that anyone is being affected by your patch or not. just pointing out that the list is not exactly consistent right now :) 13:11 VanessaE I know my servers have that patch. 13:11 VanessaE but as you can see in the list, they're not all reporting the same way 13:12 kahrl hrm 13:12 VanessaE compare digitalaudioconcepts.com:30005 with 30004 13:14 kahrl VanessaE: anything special about :30005 and :30002 compared to the others? 13:14 VanessaE nope. 13:14 VanessaE they're bog standard configs 13:15 kahrl very strange 13:15 kahrl I should stop pushing stuff, it just leads to bugs I can't explain :( 13:16 VanessaE naw 13:16 VanessaE maybe this is a problem on the master list, not the minetest server? 13:18 kahrl https://github.com/minetest/master-server/blob/master/server.py <-- it has some special handling for "mapgen", "mods" etc. depending on if the server was already in the list 13:19 Zeno` for testing: https://github.com/Zeno-/minetest/compare/getnode_exception_free (note/warning: I have only tested the changes with the client so far!) 13:19 kahrl but it shouldn't have an effect if server["action"] is "start" 13:20 VanessaE Zeno`: inb4 a certain troll gripes about spacing changes ;) 13:20 Zeno` I don't care about those 13:20 VanessaE oh btw, line 888 13:20 Zeno` where are the spacing errors? 13:20 VanessaE of game.cpp 13:20 VanessaE trailing spaces there 13:21 Zeno` those were already there :p 13:21 Zeno` this is not a PR yet, but thanks... I'll check them all for sure 13:24 Zeno` I suppose I should set my editor to remove all trailing spaces and not just lines modified by me (it looks like it was modified by me but it wasn't... just changed position in the diff) 14:00 Zeno` https://github.com/Zeno-/minetest/compare/getnode_exception_free 14:01 VanessaE ^^^ improves fps on my Survival server significantly 14:01 VanessaE (which is saying a lot) 16:40 ShadowNinja kilbith: I've seen those error always, intermittently. It seems like the request doesn't finish and you get partial data. It's rare though. 16:42 ShadowNinja VanessaE: Old servers that don't support announcing mapgen/mods/clients will display like that. 16:42 VanessaE ShadowNinja: that doesn't explain git HEAD servers like mine displaying wrong. 16:43 proller but now it on python! 16:43 ShadowNinja Hmmm, odd. 16:44 ShadowNinja sfan5: *Please* quiet him. 16:44 proller how many months you cant ever repeat what works normally long time 16:44 sfan5 *sigh* 16:45 sfan5 umm 16:45 sfan5 ShadowNinja: the quiet already expired 16:45 proller banning and changing spaces - goot problem solution 16:46 sfan5 ShadowNinja: and tell your bot that it's ##fix_your_connection with 2 ## 16:49 ShadowNinja sfan5: Not here. And single-# is more network-independent. On freenode it's just a forward to the ##. 18:45 paramat hmmmm, please could you review, possibly merge? https://github.com/minetest/minetest/pull/1834 18:47 * Krock writes "Add a Lua option for non-eased 3D-noise" onto paramat's TODO list 18:48 Krock Also, how about using constants instead of 31000? 18:49 paramat the lua eased noise might be best left to the expert :) 18:50 paramat next i need to fix the biome discontinuity at y = 47 18:50 paramat in mgv5 18:50 Krock Add it to that pull and rename it to "mgv5" fixes? :3 18:52 paramat nah best a separate pull i think 18:53 paramat hmmmmm i did actually try accessing eased noise at point by creating single-node perlinmaps heh but somehow it didn't work 18:58 Krock I actually wonder how I can check if a lua_state contains further data 18:59 Krock specially, to add optional arguments 19:03 celeron55 http://stackoverflow.com/questions/5448654/lua-c-api-handling-and-storing-additional-arguments 19:03 celeron55 "The C function lua_gettop will return how many parameters were passed to your C function." 19:03 Krock oh nice. thanks 19:04 celeron55 i.e. the function parameters are the only things on the stack at positive indices 19:05 celeron55 also note that nil is counted as a parameter too even if it's the last one 19:06 celeron55 (that distinction is lost only when reading parameters in lua when doing it in the handy regular way) 19:06 paramat sfan5 please could you look at this? https://github.com/minetest/minetest_game/pull/339 this is a blocker for biomes in mgv5 19:06 Krock okay. thanks for this explanation 19:07 sfan5 paramat: looks good 19:08 paramat aw thanks 19:27 Krock added a noiseparam, 'eased'.. trying to compile now 20:12 Krock #1835 20:12 ShadowBot https://github.com/minetest/minetest/issues/1835 -- Add option eased to NoiseParams by SmallJoker 20:15 Krock comments please 20:16 paramat wow it worked? 20:17 paramat it was inevitable anyway, hmmmmm's just busy 20:17 paramat first screenshot is beautiful 20:19 stakewinner00 I have a "question" about the code style. 20:20 Krock paramat, yes. 20:21 stakewinner00 is better something like "if (rows == 0) return 0; else return formatted_count - rows;" or "if (rows != 0) return formatted_count - rows; return 0;" ? 20:22 Krock return formatted_count - rows; 20:22 Krock because division by zero is not a problem here 20:22 Krock and you can subtract 0, that's possible 20:22 Krock ^ stakewinner00 20:23 Krock paramat, caves in mgv6 use eased (or similar) noise already IIRC 20:23 paramat dont think so 20:24 paramat caves are random, not noise based 20:24 Krock oh! 21:52 ShadowNinja hmmmm: I found the bottleneck in one of my VM mods -- update_map. The function takes about 0.02s without it and about 0.5s with it. The docs are very vague as to what it does and what makes it different from write_to_map, but not calling it leaves some phantom blocks on the client (non-updated blocks) 21:53 ShadowNinja Can it be optimized and what does it do? 21:53 ShadowNinja These figures are for a ~20x20x20 area BTW. 21:54 ShadowNinja It's important that this function complete in 0.1s or less. 21:59 ShadowNinja Just calling it the first time and not on later updates should help the speed without leaving the phantom blocks. 22:04 paramat i think update map recalcuates the lighting, and that is slow 22:59 RealBadAngel ok, i got all the possible cases working correctly for wielded: http://i.imgur.com/TPuGWyJ.png 23:02 VanessaE nice 23:03 RealBadAngel shaders on items are cool :) 23:16 hmmmm agh paramat left 23:17 hmmmm ShadowNinja, yes, I use Map::updateLighting() for non-mapgen vmanip since the mapgen lighting algorithm relies on having the mapgen... this was just something I never got around to doing 23:18 hmmmm hrmm 23:18 RealBadAngel https://github.com/minetest/minetest/pull/1836 23:19 RealBadAngel pr is rdy 23:19 hmmmm this biome/ore/decoration/schematic reorganization is really large and messy. i rely on covariant return types but i'm beginning to wonder if this was a mistake 23:19 hmmmm will you guys kill me if I use templates instead? 23:19 RealBadAngel idk what are you doing so im ok with it ;) 23:21 VanessaE hmmmm: so long as it doesn't break anything or result in a performance regression, why not? 23:21 hmmmm because we just don't use templates 23:21 hmmmm it's an implicit style rule 23:23 RealBadAngel kahrl, here? 23:24 y templates are good as long as you don't write code like http://stackoverflow.com/a/275295 23:24 VanessaE hmmmm: don't let style "rules" get in the way of making your code do its job. 23:25 VanessaE I don't know what templates are in the C++ sense of the word, but if using them is the right way, then do it and make a note in the style guide of when they could be considered allowable. 23:25 VanessaE (I have a definition of a template that likely differs from yours) 23:26 RealBadAngel VanessaE, can you check 1836? 23:26 VanessaE RealBadAngel: I just did. I'm not sure you'll like the result. 23:27 RealBadAngel go on 23:27 VanessaE one moment. 23:27 VanessaE with aniso + mipmap + tri, shaders + bumpmap + auto normals + parallax all turned on: http://digitalaudioconcepts.com/vanessa/hobbies/minetest/screenshots/Screenshot%20-%2011122014%20-%2006%3a28%3a02%20PM.png 23:28 RealBadAngel since when you turn on filters for 16px ? 23:28 RealBadAngel :P 23:28 VanessaE normally I don't, but you know me, I test everything 23:28 VanessaE turn of mip/aniso/tri and it looks correct, i.e. the same as your screenshot 23:28 RealBadAngel ofc 23:29 RealBadAngel filters are for high resolution tps 23:29 VanessaE yeah but we need a threshold function here. 23:29 VanessaE you and I both know high rez texture packs will never be able to support 100% of everything 23:29 RealBadAngel selecting the RIGHT option is enough 23:30 VanessaE so there needs to be a cutoff below which those filters will simply not be used. 23:30 RealBadAngel some folks like it blurry 23:30 VanessaE so make the cutoff configurable 23:30 RealBadAngel theres too much setting and 90% of them never used 23:30 RealBadAngel lol 23:30 VanessaE here's with aniso/mip/tri turned off, shaders/bump/auto-normals/parallax turned on: http://digitalaudioconcepts.com/vanessa/hobbies/minetest/screenshots/Screenshot%20-%2011122014%20-%2006%3a31%3a07%20PM.png 23:31 VanessaE I'd say this looks fine 23:31 RealBadAngel check it out with hdx 23:31 * VanessaE checks 23:33 VanessaE http://digitalaudioconcepts.com/vanessa/hobbies/minetest/screenshots/Screenshot%20-%2011122014%20-%2006%3a33%3a30%20PM.png 23:33 VanessaE that's with mip/aniso/tri, shaders/bump but without auto-normals ^^^ 23:33 * VanessaE turns that on just for giggles. 23:34 RealBadAngel items can use only auto normals 23:34 RealBadAngel not like the nodes 23:35 kaeza so entities can now take advantage of shaders? cool 23:36 kaeza is there a way to add at least simple lighting to entities (like it was done for wielditems)? 23:36 VanessaE auto-gened normals look funky with HDX :P 23:36 kaeza I mean shading, rather 23:36 RealBadAngel ofc 23:36 kaeza or is that part of the patch? 23:36 VanessaE anyway I see no obvious problems aside from the need for thresholding. 23:37 RealBadAngel kaeza this covers only wielded and item entities 23:37 RealBadAngel but i can do the same with all the entities, including players 23:38 VanessaE hm. I can't seem to pick up those seeds. 23:38 VanessaE but I can pick up a dropped fence. weird. 23:38 VanessaE oh, full inv. 23:38 RealBadAngel lol 23:38 VanessaE that would help :) 23:39 VanessaE ok, yeah I think it's all good here 23:44 RealBadAngel hmmmm, are you ok with 1836? 23:44 hmmmm #1836 23:44 ShadowBot https://github.com/minetest/minetest/issues/1836 -- Wielded fixes. Add shaders support. by RealBadAngel 23:45 hmmmm why did you need to include "main.h" to wieldmesh.c? 23:45 hmmmm just curious 23:45 RealBadAngel g_settings 23:46 hmmmm ahhh 23:46 hmmmm grr. 23:46 hmmmm I think it should be exposed through settings.h instead. 23:46 hmmmm sorta wonder how much compilation time that would save on 23:46 proller "already solved in.." 23:47 RealBadAngel ofc it should 23:47 hmmmm RealBadAngel: just to make sure I'm not crazy, bilinear and trilinear texture filters are mutually exclusive, right? 23:47 hmmmm you can't have both applied with irrlicht, right? 23:48 RealBadAngel trilinear overrides bilinear 23:48 hmmmm wonder why irrlicht would have it as a set of flags then 23:48 RealBadAngel so if you have both set, trilinear is used 23:48 hmmmm i know why we have bilinear/trilinear as boolean options 23:49 hmmmm because there's no good "radio button" control for irrlicht so checkboxes had to be used instead 23:49 RealBadAngel should be toggleable: none/bi/tri 23:49 hmmmm of course 23:49 VanessaE hmmmm: because I didn't think to make them mutually exclusive. 23:49 VanessaE no other reason :) 23:49 hmmmm VanessaE, I remmeber when this was implemented 23:49 hmmmm it's because there was no good control to represent a tri-state variable 23:50 hmmmm RealBadAngel, just a nit: https://github.com/RealBadAngel/minetest/commit/9f08f9e21d72cd6f7ed391044472986d6f6582f0#diff-60aa608aac2fefb1a86d4c2861a08eb7R321 23:50 hmmmm you always do stuff like this 23:50 hmmmm how hard is it to make sure your code is properly formatted?? 23:50 RealBadAngel oh shit 23:50 hmmmm perhaps you code too fast and you should take your time and pay more attention to what you're actually typing 23:50 VanessaE hmmmm: at the time, I'd have just used a radio button had I known better....but then again, did we even have those? 23:50 RealBadAngel because i was changing that code hunders of times last few days 23:50 hmmmm VanessaE, no, that's the point 23:51 hmmmm we'd have to make them.. 23:51 VanessaE that explains it 23:51 hmmmm just like everything else with the completely deficient irrlicht UI toolkit 23:51 RealBadAngel you wouldnt belive what gimmicks i have to make to get all the cases working 23:51 VanessaE well nothing says you can't just auto-uncheck one if the other gets checked. 23:51 VanessaE it's hacky and stupid, but it would work 23:52 hmmmm looks good 23:52 VanessaE (then you have users asking "why didn't you just use a radio button there, you twit?") 23:52 VanessaE : 23:52 VanessaE :) 23:52 kaeza don't we already have dropdown lists? 23:52 VanessaE kaeza: yeah 23:52 VanessaE but a dropdown is ugly for that idea 23:52 VanessaE a spinbox is best 23:52 hmmmm I guess you couuuuld use a dropdown for that 23:52 kaeza (those are kinda radiobuttons, and are used for e.g. mapgens) 23:53 kaeza +like 23:53 hmmmm i always thought a radio option set is ideal for things that only have a few options that are mutually exclusive and are always static 23:53 hmmmm this would be the perfect example 23:53 hmmmm the perfect use i mean 23:53 RealBadAngel hmmmm, added that space 23:53 hmmmm cool 23:54 hmmmm LGTM 23:55 RealBadAngel ok, so lets get this merged, those fixes have to be widely tested (previous kahrl's code was irrlicht version sensitive)