Time Nick Message 00:00 TBC_x that makes sence 00:06 Routh This is weird. I've defined an alternate water block for a biome, and for some odd reason minetest is added a 1 block layer of water on top of the other liquid. 00:11 * VanessaE peeks in 00:12 VanessaE Routh: check your "dust" setting for the biome 00:12 Routh VanessaE: Not set 00:18 Routh Werid. I explicitly defined node_water_top with the same liquid and it worked. 00:19 Routh Dunno if that is a bug with biome API or my liquid. 00:19 paramat will look into it, might run your subgame to see 00:21 Routh I think my liquids definition in moreliquids may be lacking, I need to review other liquid defs to see whats possible and what I may be missing. 00:21 Routh I know it doesn't drown you, seems I missed drowning = 1 00:26 Routh Oh, figured out what I did wrong. Doh. I left depth_water_top = 1 on the biome def when troublshooting, and it added water since node_water_top was left undefined. 00:27 paramat aha =) 00:31 VanessaE paramat: oh btw, moretrees now uses the default acacia nodes (and the pine ones too), in case you're curious. 00:32 paramat good, the previous acacia leaves were ridiculous, didn't even tile 01:25 paramat now pushing trivial #2947 01:25 ShadowBot https://github.com/minetest/minetest/issues/2947 -- Conf.example: Update mgv7 noise parameters by paramat 01:30 paramat complete 02:12 Sokomine has something been changed regarding minetest.serialize? some behaviour seems odd. it wasn't throwing complaints at me for passing a function in a table to it, and the output looks diffrent. or is it just me? 03:46 est31 hmmmm, have you seen #2941 ? 03:46 ShadowBot https://github.com/minetest/minetest/issues/2941 -- Serialization error 03:50 hmmmm well, i have now 03:50 hmmmm est31: that's good 03:51 hmmmm est31: under normal circumstances that would've flooded the chat window with a bunch of bogus characters and possibly get further into the AOM processing chain 04:31 VanessaE regression: now all the wielded items in the wield3d mod are black 04:32 VanessaE i.e. what you see other players carrying 04:34 VanessaE RealBadAngel: ^^^^^^ 05:42 RealBadAngel VanessaE, link to that mod? 05:59 VanessaE https://forum.minetest.net/viewtopic.php?id=6407 06:00 VanessaE RealBadAngel: better to just sign onto VE-Basic and you can see it in use there. 06:00 VanessaE however now it's not doing it 06:01 VanessaE RealBadAngel: I see the trouble - it doesn't react to the ambient light, only to the sunlight level 06:02 VanessaE I'm standing next to a torch at night and it's still solid black 06:10 RealBadAngel VanessaE, when you drop an item on the ground, next to that torch, is it properly lit? 06:11 VanessaE yes 06:11 RealBadAngel can you check shaders on/off ? 06:11 VanessaE this is without shaders. 06:12 RealBadAngel having them on changes anythin? 06:13 VanessaE no effect. 06:13 VanessaE dropped items look correct but not the wielded item, which is probably an attached entity 06:13 RealBadAngel i see, compiling now fresh sources and will get that mod, will look for the reason 06:15 VanessaE reference screenshot: http://digitalaudioconcepts.com/vanessa/hobbies/minetest/screenshots/random/Screenshot_2015-07-22_02-14-33.png 06:15 VanessaE that's a block of dirt-with-grass that I'm holding, same as is dropped on the ground in front of me. 06:18 VanessaE (that screenshot was taken with shaders enabled; normally I have them turned off, same effect either way) 06:19 RealBadAngel lol that mod is glitchy as hell 06:19 RealBadAngel items keep disappearing. when you walk it swaps the hands too 06:20 RealBadAngel and the item shows on the other hand 06:20 RealBadAngel with item rotated by 180degs 06:20 VanessaE the hand swap is an engine bug I'm sure of it 06:21 VanessaE but doesn't matter, this is about the lighting on the wielded object, not how the wield is implemented :) 06:24 RealBadAngel https://imgrush.com/upZF7zNLKcaj 06:24 RealBadAngel https://imgrush.com/yG-fEsFNxMPW 06:24 RealBadAngel i cant see anythin wrong with it 06:25 VanessaE now go into a dark cave during the daytime. 06:26 VanessaE the wield will be full bright 06:28 RealBadAngel nope.avi 06:28 VanessaE well it's broken for me. 06:31 RealBadAngel can i see a screenie how it looks for you? 06:33 VanessaE reference screenshot: http://digitalaudioconcepts.com/vanessa/hobbies/minetest/screenshots/random/Screenshot_2015-07-22_02-14-33.png 06:34 VanessaE compare the dirt-with-grass block I dropped on the ground with the one in my hand 06:34 VanessaE (the dropped one is just above the hotbar in the screenshot) 11:09 VanessaE https://github.com/VanessaE/hdx-16/issues/3 11:10 VanessaE one of the oddest messages I've gotten in recent times, but he's right. Can this be remedied at some point? 11:13 rubenwardy some texture packs store alternatives in subdirectories 11:13 VanessaE right, but I think he's talking more like "default" in one dir, "coloredwood" in another, and so on 11:14 rubenwardy yeah 11:14 rubenwardy For compatibility, you'd need to add a meta data file which says look in sub directories 11:14 VanessaE yeah 11:18 TBC_x I hate reference counted memory leaks 11:18 TBC_x but I may have come up with a way to hunt them 11:28 kilbith first at all, find out a memory free-er for the distant mapblocks 11:29 kilbith this is insane that they're always kept in RAM 11:29 TBC_x you know... caching 11:30 kilbith caching ad nauseum, yeah 11:31 TBC_x how are mapblocks sent? on client request or are just sent by the server? 11:31 VanessaE I'd say unload based on distance from the nearest client, multiplied by time since loading. if the result exceeds some arbitrary threshold, unload the block 11:32 VanessaE TBC_x: by the server based on which way the client is looking 11:32 TBC_x i mean client-side right now 11:32 TBC_x if that makes sense 11:33 VanessaE it does. 11:34 TBC_x I know, that mapblocks have their lifetime defined by timeout value 11:48 TBC_x this is annoying, when I WANT it to leak memory it does not leak 11:54 nrzkt TBX_x timeout is not sufficient 14:04 TBC_x http://sprunge.us/HAgC leak sanitizer report from inside of running minetest 14:07 VanessaE only 7.6MB of leaks? throw dreambuilder at it and see what happens ;) 14:08 TBC_x well... we're keeping 4k mapblockmeshes around 14:09 TBC_x I got the output so I could see how many objects there are hanging out 14:09 VanessaE hm, you should sort the output by leak size 14:09 TBC_x grep " leak " leaks | uniq |less 14:09 VanessaE that's cheating :P 14:11 Krock lol 14:11 TBC_x I needed a way to count objects 14:12 VanessaE 2015-07-22 15:09:04: ERROR[main]: generateImage(): Could not load image "door_steel.png" while building texture 14:12 VanessaE so I'm not the only one who gets that error huh? 14:12 TBC_x I thought you're aware of that 14:13 nrzkt i didn't know we sell tshirts.... http://www.cafepress.co.uk/mf/83297301/mt-hdx-logo_tshirt?productId=998795960 14:13 VanessaE nope. Id I've seen any such report, I've forgotten about it :) 14:14 * kilbith would like a t-shirt autographed by RBA :3 14:14 nrzkt is this somebody who want to earn money on minetest community ? 14:14 VanessaE nrzkt: that's rubenwardy's cafepress store. 14:14 VanessaE been there for a couple years now I think 14:14 rubenwardy yeah 14:15 rubenwardy there's $6 there, I can't withdraw until there's $25 14:15 Calinou is the quality of the mug good? 14:15 nrzkt okay 14:31 RealBadAngel VanessaE, i cant reproduce that black wield3d issue even when using old server and new client 14:32 est31 I've seen it on VanessaE's server too 14:33 VanessaE so clearly it ain't unique to my client :P 14:33 VanessaE and the mod is just a normal, unmodified install 14:33 RealBadAngel is your server still an older ver or you have it updated? 14:33 VanessaE nope 14:33 VanessaE it's up-to-date as of 6am this morn. 14:34 VanessaE (4 1/2 hours ago) 14:34 RealBadAngel ok, going to waste some of my limits and join it ;) 14:34 VanessaE heh 14:35 RealBadAngel which one? 14:35 VanessaE Basic 14:35 VanessaE (30005) 14:38 VanessaE ssssshhh I'm hiding from the legion of n00bs over there. :P 14:46 TBC_x I think it is driver-dependent issue 14:46 VanessaE I'd be inclined to blame irrlicht first 14:47 VanessaE seeing as how all other lighting seems to work like it should 14:48 RealBadAngel VanessaE, ok, im in 14:48 RealBadAngel where are you? 14:48 VanessaE I'm not online at the moment. hiding from the legions of n00bs 14:49 VanessaE you caj ust hit f7 and check in that mode, it's the same 14:49 VanessaE can* 14:56 RealBadAngel i think we shall look for the bug in code that set player->light_color 14:56 VanessaE so you can reproduce it? 14:56 RealBadAngel by joining your server yes 14:57 VanessaE ok 14:57 RealBadAngel locally i cant 14:57 VanessaE that's what made me think it's network-related. 14:57 VanessaE that's surely the only difference between local client+server instances and going online. 14:57 RealBadAngel you also wasnt able to reproduce it on local right? 14:57 VanessaE yeah, locally it worked for me, too 15:00 RealBadAngel anyway, i think best way to fix it is making wield3d mod obsolete anyway 15:00 VanessaE agreed 15:00 VanessaE but until then.. 15:00 RealBadAngel if we have it done properly, client side, we simply wont have such problems 15:00 gravgun Hum I can't create an account on the mt-dev wiki due to some "Incorrect or missing confirmation code.", is anyone else experiencing this issue? 15:00 VanessaE gravgun: everyone has that problem. 15:00 VanessaE Calinou: ^^^^ 15:00 gravgun Alright... 15:01 RealBadAngel VanessaE, so, im going now to code wielded engine and client side 15:01 RealBadAngel not so hard i think 15:01 VanessaE no 15:01 RealBadAngel i do have all what i need 15:01 VanessaE whatever it is that's causing the problem needs fixed first. 15:02 VanessaE because you can't fix old clients 15:02 VanessaE and I'll be using this mod until at least the point where old clients can safely be ignored. 15:03 RealBadAngel many of our bugs are caused by thinkin this way (offtopic) 15:03 est31 actually I agree with rba on this point 15:03 VanessaE yeah 15:03 RealBadAngel "stfu and get client up to date" 15:03 est31 yea 15:03 VanessaE old clients can't be updated easily. 15:03 VanessaE think tablet/phone users 15:03 RealBadAngel just dl new one 15:04 est31 we have to maintain actual backwards compat, but wield items being shadowed is a "luxury bug" for me 15:04 RealBadAngel whats so hard? 15:04 VanessaE there isn't a new one. 15:04 VanessaE that's what's so hard. 15:04 VanessaE tablet clients are slow to update. 15:04 VanessaE (the maintainers I mean) 15:04 RealBadAngel are you kidding me? 15:04 RealBadAngel check out "HayDay" 15:04 VanessaE no, I'm talking from experience here. 15:04 RealBadAngel without client up to date youre not able to connect 15:05 RealBadAngel and are FORCED to get new one 15:05 VanessaE someone the other day produced a disturbing chart of current-in-use clients. at least half of the Minetest userbase is out of date. 15:05 est31 no 15:05 gravgun How to make something good: 1. Solid and flexible protocol 2. Code 3. Done 15:05 RealBadAngel which is circa 50mb 15:05 RealBadAngel and they do updates at least once a month 15:05 VanessaE this isn't HayDay., 15:05 RealBadAngel but philosophy is correct 15:05 gravgun This isn't 2000's either 15:06 VanessaE this is minetest with its ongoing problem of clone clients. 15:06 RealBadAngel sure, lets think on the code to provide compability with all the versions around including fm :) 15:06 VanessaE I'm still getting users who haven't updated enough to see mesh nodes, for crying out loud. 15:07 est31 well thats their fault then 15:07 RealBadAngel because you let them in 15:07 est31 but at least you can tell them "update your client" 15:07 VanessaE just yesterday for example. fortunately that guy was able to update 15:07 RealBadAngel if they were told, go get new one... everybody would be happy 15:07 RealBadAngel after a single download 15:07 VanessaE no 15:07 est31 I have done a PR to check the clientside version clientside 15:07 VanessaE I'd lose like 80% of my userbase 15:07 est31 but it has been muted in a shitstorm 15:07 RealBadAngel thats what you think 15:08 VanessaE that's what I know! 15:08 RealBadAngel its totally wrong 15:08 VanessaE I lost half of it just by changing my G*d damned server name 15:08 est31 #2508 15:08 ShadowBot https://github.com/minetest/minetest/issues/2508 -- Pass version information to joinplayer callback by est31 15:09 est31 even displaying a hint "please update your client" is too much for some people 15:09 VanessaE I'm afraid so 15:09 VanessaE don't get me wrong, I am not saying DON'T do in-client 3d wields. 15:10 nrzkt VanessaE: phone users should use official version and shut up 15:10 VanessaE I'm saying you can't expect most server owners to rely on it so quickly when there's an existing solution. 15:10 VanessaE nrzkt: yes they should, but they won't/ 15:10 VanessaE good luck getting them to switch. 15:10 VanessaE I have tried 15:10 VanessaE and tried 15:10 VanessaE and tried,. 15:10 nrzkt VanessaE: we can add a fuction in minetest server when they connect to insult them 15:10 est31 consult is a more polite term 15:11 VanessaE nrzkt: and make them turn away from minetest entirely? that's hardly a good idea. 15:11 nrzkt xD 15:11 VanessaE these are 8-10 year old kids 15:11 nrzkt it was a joke 15:11 nrzkt :p 15:11 VanessaE well 8-15, average seems to be about 12. 15:11 nrzkt fucking kid, please update your minetest or i hit your ass 15:11 nrzkt this is a good message no ? :p 15:11 VanessaE no :p 15:11 nrzkt :( 15:12 nrzkt but i really think we could tell users to update their clients 15:12 VanessaE I don't disagree with telling users to upgrade 15:12 nrzkt when they connect to server, server can tell them to update to newer version 15:12 VanessaE but there's no way to do so that they'll actually listen to. 15:12 nrzkt you can do it with a mod no ? 15:12 VanessaE nope. 15:12 nrzkt i do it in my server 15:12 nrzkt how i do it... 15:13 est31 you can do it with a mod, but only if you compiled in debug mode 15:13 RealBadAngel age doesnt matter 15:13 nrzkt https://github.com/nerzhul/minetest/blob/nrzforked_experiment/src/contrib/server.cpp#L38 15:13 nrzkt oh i do it natively 15:13 est31 minetest.get_player_information(player_name) 15:13 VanessaE RealBadAngel: sure it does. the smaller the kid, the less likely they'll be to understand "get the official client" 15:14 RealBadAngel just show message "updating your client. please stand by...." 15:14 RealBadAngel and fuckin do it 15:14 nrzkt we can add: if you are a kid, tell you father to di it 15:14 VanessaE nrzkt: these are *american* kids. ergo, equally stupid parents :-/ 15:15 nrzkt tell your brother ? :p 15:15 nrzkt tell your grand ma ? 15:15 est31 perhaps we need better marketing 15:15 nrzkt not marketing but communication 15:15 VanessaE est31: we do. 15:15 nrzkt and the first thing is to tell players to upgrade their clients 15:15 nrzkt we should add this function in core or buildint before next release 15:15 nrzkt builtin* 15:15 VanessaE nrzkt: the problem is still getting out from under 4 million old,outdated clones 15:15 VanessaE and kids just won't do it 15:15 VanessaE I have trieed 15:15 RealBadAngel such ways used in HayDay effects in millions of players worldwide miliking stupid cows 15:15 nrzkt clones are not our problem 15:16 RealBadAngel can be done? 15:16 RealBadAngel ofc it can 15:16 est31 RealBadAngel, we arent hayday 15:16 nrzkt the main thing is to have an automated message for update 15:16 nrzkt each time they connect they will see it, they will learn to update their cleints 15:16 VanessaE nrzkt: which doesn't exist in old clients. 15:17 est31 yea 15:17 RealBadAngel est31, no we are not. but idea of having client and server in sync is valid 15:17 VanessaE you can't detect the client version from the server 15:17 est31 VanessaE, you can 15:17 est31 minetest.get_player_information 15:17 VanessaE est31: I mean in Release builds. 15:17 est31 yes 15:17 est31 that has to be changed IMO 15:17 nrzkt what doesn't exist in clients ? 15:17 nrzkt it said the server send the message 15:17 nrzkt i* 15:17 VanessaE nrzkt: an automated way of checking versioning and offering to update. 15:17 nrzkt then it's not a problem with clients :) 15:18 nrzkt oh, i don't talk about offering an update 15:18 est31 at least one should be abled to check the protocol version 15:18 nrzkt but telling them how to do it 15:18 nrzkt ti's the minimum to do, and we don't do it atm 15:18 RealBadAngel also theres another problem with the sync, minetest_game is in continuos async mode 15:18 Tesseract hmmmm: Does my threading PR look good now. I've fixed everything you mentioned (except for the pthread leaking thing, which was really a question). 15:18 nrzkt i don't agree with automated updating because we are packaged in distros 15:18 VanessaE nrzkt: android. 15:18 kilbith VanessaE: it's not our problem if android players don't get the official release on android 15:18 VanessaE kilbith: it IS our problem. 15:19 nrzkt no 15:19 VanessaE it *shouldn't* be 15:19 VanessaE but it is. 15:19 nrzkt because we are not maintainer 15:19 VanessaE it's a problem we have to deal with in the most graceful way possible to avoid alienating our userbase. 15:19 est31 VanessaE, automated updates are forbidden by play store AFAIK 15:19 nrzkt it's the problem of maintainers of clone if their client doesn't work with servers 15:19 nrzkt they will be forced to update they stupid outdated forks 15:19 VanessaE est31: even *offering* an update? 15:19 kilbith that's wrong, i do have auto. updates everyday from the play store 15:20 kilbith but it's configurable 15:20 RealBadAngel what i do think we should forbid connecting different versions. put the game in sync with engine. make the official releases come out more often 15:20 nrzkt kilbith, i think est31 talks about app updated itself 15:20 nrzkt not playstore checking update 15:20 kilbith i talk about app update too 15:20 nrzkt RBA : not 1 server with 1 version, but 3 versions is sufficient i think 15:20 nrzkt not 20 15:21 nrzkt i looked at stats and we can drop Minetest 0.4.7 and lower 15:21 RealBadAngel if we had chosen syncing we would have anything in the statistics other that last official 15:22 nrzkt yes, but it's too late 15:22 VanessaE nrzkt: drop everything before 0.4.12 if anything is dropped. 15:22 nrzkt now we should do it smother 15:22 RealBadAngel cut the rope 15:22 est31 man 15:22 nrzkt wtf VanessaE, i don't understand 15:22 nrzkt i looked at master server stats with graphs 15:22 est31 why do you not want to give server owners freedom? 15:23 est31 server owners can enable strict protocol checking 15:23 kilbith VanessaE: if some player complain about the outdated homedecor on launchpad, that's not your problem either 15:23 nrzkt est31: agree, but with limit, a maximum version supported 15:23 est31 if they think its so great 15:23 est31 how many have done it 15:23 RealBadAngel if i can hear folks are complainin about meshes? 15:23 VanessaE kilbith: that's not my problem because I don't maintain those packages, but at the same time it IS my problem because I have to field support requests. thankfully not too many in that instance. 15:24 est31 nrzkt, RealBadAngel my offer is still there, if you get me 8 of the top 10 server owners on the public list to agree to dropping protocol, we can do it. 15:24 RealBadAngel damn how old is meshnodes drawtype? 15:24 est31 but not until then 15:24 kilbith VanessaE: we don't maintain the wacky android forks either 15:24 hmmmm ugh 15:24 est31 if server owners thought it would be so great, they would have done it 15:24 est31 but they dont do 15:24 VanessaE RealBadAngel: that's exactly my point - I had to tell someone just yesterday to update their client for meshnode support. they were still using 0.4.10. 15:24 est31 we have to follow our users here 15:24 VanessaE kilbith: and yet, we have to deal with the fallout from them. 15:24 hmmmm why are you being so pushy about dropping reverse compatibility 15:24 est31 NONE of the servers has enabled strict protocol checking AFAIK 15:24 hmmmm is there an actual technical reason why this needs to be done 15:25 nrzkt est31: supporting the two last years versions is sufficient, no need to support minetest before that 15:25 VanessaE est31: for a while, my two dreambuilder servers had that. then I realized it cut off too many people./ 15:25 est31 nrzkt, why is there need to drop support 15:25 hmmmm i asked this question before, there is no need, he just wants to 'clean up' the code 15:25 est31 thats not worth it 15:25 nrzkt 1. users update their clients. 2. have a clear code well supported and well known 15:25 RealBadAngel one thing i know its extremaly hard to solve issues not only in current code but also in all previous ones 15:25 hmmmm of course 15:26 hmmmm i dunno about you but i'm not in the business of forcing users to do things they don't want to do 15:26 nrzkt 3. can evolve minetest 15:26 hmmmm optionally executed code is not holding back minetest 15:26 nrzkt if we support minetest 0.3 we cannot add some features because they don't support it. 15:26 est31 nrzkt, if you have spare time to evolve minetest, document the protocol on the wiki. Then people can do clean re-implementations 15:26 VanessaE there's surely a limit of how much backward compat to retain. 15:27 VanessaE the argument here is where that limit should be and what we do uip to that point. 15:27 VanessaE up* 15:27 hmmmm vanessae: unless there's an actual reason for doing so I still don't see why it needs to be dropped 15:27 nrzkt the net protocol works like it does now. But it's very very very very hard to evolve the protocole because we should have many conditions to handle, and some has bugs. 15:27 hmmmm the discussion about what the limit should be is how far back reverse compatibility is necessary 15:28 hmmmm i.e., if there is a technical reason for cutting out reverse compat beyond that point, we can do so safely 15:28 est31 but only if there is a technical reason 15:28 hmmmm right 15:28 nrzkt this can take 1 year, but 1 not wait 1 year to do it smoothy 15:28 est31 nrzkt, tell us more about those bugs. 15:28 nrzkt look at minetest 0.4.10 which send u16 for u32 to clients 15:29 est31 ? 15:29 nrzkt when i looked at the code, 7 month ago i see some packets sending u16 to clients but the variable was u32 serverside and client side. 15:30 nrzkt i also see that the server has a race which permit to crash it. It's fixed now, and packets are properly read 15:30 est31 what variable was it? 15:31 nrzkt search my first commits on the repo 15:31 nrzkt when i started to design protocol v25 15:31 VanessaE well this sure went off-topic fast at any rate. 15:32 est31 I know protocol v25, I've done most of it. 15:32 nrzkt you do authentication 15:32 nrzkt and i fix many protocol problems with datas. 15:32 nrzkt you don't do most of it. We worked together. 15:33 est31 I see you did TOCLIENT_DELETE_PARTICLESPAWNER 15:33 nrzkt yes, it was the packet which has u16 problems 15:33 est31 TOCLIENT_ACCESS_DENIED was done by you too 15:33 est31 and parts of the new init packed 15:34 est31 packet* 15:34 est31 also TOCLIENT_HELLO 15:34 nrzkt we are not talking about a recompense for the devel which has many lines to code here. 15:34 nrzkt please come back to the initial discussion. 15:34 est31 but rest was auth, which was done by me 15:34 nrzkt RBA: what are your arguments ? 15:35 nrzkt i also patched the Packet handling, changed the packet reading mode 15:35 nrzkt and improve performance in some low layers, but it's not the problem here. 15:35 VanessaE what does all this got to do with fixing attached entity lighting? or client-side wielding? 15:35 VanessaE what has* 15:35 est31 nrzkt, thats nice, but has nothing to do with protocol v25 15:35 nrzkt nothing, and we are talking about minetest communication problem about version update 15:35 nrzkt stop that est31 this is stupid, you want a medal, take it. 15:36 VanessaE nrzkt: yeah, y'all went off on a tangent and that off onto another tangent :P 15:36 est31 nrzkt, no I just wonder what technical arguments you have 15:36 nrzkt too many maths for me, i forget my graduate courses :p 15:36 est31 to drop old versions 15:36 est31 you just list what you have improved 15:36 nrzkt you are out of course, you talk about new protocol, not old 15:36 est31 so yes take your medal 15:36 nrzkt i don't want a medal 15:37 nrzkt stop that now. 15:37 est31 but what does low level speedup to do with dropping old protocol versions 15:37 nrzkt you mixed everything... 15:38 est31 I see now nrzkt what you've meant with the u16/u32 thing 15:38 est31 agreed, its a bug 15:38 est31 so servers which want to have more than 2^16-1 particle spawners have to enable strict protocol checking 15:39 est31 but particle spawners are not worth it IMO 15:39 est31 lets give the choice to server owners 15:39 nrzkt particle spawner is is increasing no ? 15:40 est31 well thats a problem with id management serverside 15:40 est31 classical id assignment problem 15:45 RealBadAngel whatever the feature, whatever the fix. point is when we fix it older clients wont get it 15:45 RealBadAngel or make the end effect even worse 15:46 est31 RealBadAngel, can you please ask server owners to enable strict protocol checking? This is not a discussion you have to have with me, you have to have this discussion with the users of our software. 15:47 nrzkt est 15:47 RealBadAngel i know, but thats the effect of our politics 15:47 RealBadAngel not theirs 15:47 nrzkt est31: i really think we should add a message sent to clients with outdated client to tell us to upgrade 15:47 nrzkt just a message to inform them, if they don't know, why they should update ? 15:48 RealBadAngel for starters just a warning message could do 15:49 RealBadAngel like: it looks like youre using older client. if youre experiencing some drama please do update it in the first place ;" 15:49 est31 nrzkt, I agree with that, but you should be abled to customize the message, and to turn it off. Best is that you make it moddable (minetest.player_uses_outdated_client(playername)), then server owners can integrate it into their system. 15:49 nrzkt use a minetest.conf string value then 15:49 nrzkt and we can add it to buildint 15:50 nrzkt builtin 15:50 nrzkt called at every joinplayer 15:50 est31 and optional, yes 15:50 nrzkt then server owners only have to disable it 15:50 nrzkt enable it by default, because it's the most important 15:50 est31 ok 15:50 nrzkt this is the most important to do 15:50 nrzkt communicate. 15:50 est31 ? 15:51 nrzkt the most important is to communicate information to our users, server admins & players 15:52 nrzkt we should add this important feature in the changelog and warn admins about the importance of this message 15:52 VanessaE most will ignore it 15:52 VanessaE or not see it 15:52 VanessaE (that's not an argument to not add it) 15:52 VanessaE (merely an argument not to *rely* on it) 15:53 est31 we can't do much else 15:53 est31 except more marketing for the official minetest 15:53 VanessaE *nod) 15:53 VanessaE *nod* 15:54 est31 it is basically marketing 15:54 nrzkt maybe, but they are advertised, we are not minetest support team which come to each user by helicopter 15:54 VanessaE nrzkt: no, that's Tesla. :) 15:54 nrzkt xD 16:02 est31 I just think we shouldn't feel that the problems we solve (u32 vs u16 spawner issue) are more important than our users. I think users are more important. 16:08 TBC_x I would just append build date after the version string 16:08 nrzkt i agree 16:08 nrzkt but our users didn't use minetest < 0.4.7 16:08 nrzkt we can drop support. 16:09 rubenwardy It's not like we have real support 16:09 rubenwardy new clients can't connect to Redcrab 16:10 est31 well, redcrab has to update then 16:10 rubenwardy which is 0.4.4 16:10 VanessaE from the previous discussion last week: http://meow.minetest.net/tmp/serverlist_stats_2015-03-14.txt 16:10 rubenwardy so dropping remaining support for versions where the support doesn't actually work is good 16:10 est31 thats their problem. We dont have a majority of users on redcrab do we 16:10 VanessaE rubenwardy: actually they can as long as the user deletes one file from the cache that's supplied by recrab. 16:10 rubenwardy oh 16:11 est31 The core problem is that the people who develop the unofficial clones want to make money with inserting ads/demanding money. As long as the money is still coming in, they have no motivation to update. So we have to find some way to convince the app maintainers to update the version. 16:12 est31 without losing users on the way. 16:13 rubenwardy that's a different issue, one I wanted to solve with #2442 16:13 ShadowBot https://github.com/minetest/minetest/issues/2442 -- Notifications of new engine versions. 16:13 est31 rubenwardy, that doesnt solve it 16:13 est31 its just for new versions 16:13 est31 we need a solution that works for old ones 16:13 rubenwardy it could be done server side 16:13 est31 yes 16:14 est31 thats what nrz wants to add. 16:14 est31 nrzkt, can you make a pr? 16:14 rubenwardy what's the problem, then? 16:15 est31 nrz and RBA want to drop older versions, for no apparent reason 16:15 rubenwardy what's the point in keeping half support? 16:16 est31 its sufficient for most people 16:16 est31 if 90% of the features are right, its better than none 16:16 VanessaE I say use the server list itself. 16:17 VanessaE feed to clients a modified server list that puts a two- or three-line warning in above all the public servers. 16:17 VanessaE drop support for THAT in say a year 16:17 est31 ? 16:18 rubenwardy That's a bit hacky :P 16:18 VanessaE hacky yes 16:18 est31 it does the job 16:18 VanessaE but guaranteed to work on nearly every client in existendce. 16:18 VanessaE existence* 16:18 TBC_x is the server list notified of client version? 16:19 est31 so please nrzkt RealBadAngel dont run around demanding to cut out half or more of minetest's population. please try to find ways to get the population to newer versions first, if you are so motivated to drop support for old versions. You have my approval for a message on join. 16:19 nrzkt VanessaE i will show you graphs later, i have a kibana which have parsed all logs 16:19 VanessaE after a year (or less?), stop issuing the warning via that method because by then, in-client support for a *proper* warning dialong would surely be in and everyone would hopefully have it 16:19 nrzkt all logs from the begging to last week 16:19 VanessaE nrzkt: I think I saw that log. 16:20 TBC_x I think that warning should be there for good 16:20 nrzkt i don't send the address of the graph public 16:20 VanessaE nrzkt: ok, I'm thinking of a different log then. 16:21 VanessaE it was like this one, but focused just on versions, without platform info: http://meow.minetest.net/tmp/serverlist_stats_2015-03-14.txt 16:21 nrzkt est31: i don't want to drop older version, just unused version, usage < 5% 16:21 nrzkt est31: cannot provide a PR these day, i'm at work and i need to add a note to many of my students for tommorow :) 16:21 VanessaE of course that one's old, march of this year 16:22 est31 ok 16:22 VanessaE sfan5: *poke* 16:22 sfan5 VanessaE: *poke back* 16:22 VanessaE sfan5: do you have a newer version of ^^^^ lying around? 16:22 sfan5 no 16:23 VanessaE ok. 16:23 sfan5 but if you want just a single statistic i can give you that 16:23 sfan5 making a whole report is pretty time consuming 16:23 rubenwardy Just the version % 16:24 est31 sfan5, nrz has said that from the 25% for 0.4.7 and older, 20% are "unknown", he says that they can be disregarded. Do you know whether older versions didnt send the version in the user agent?? 16:24 nrzkt i will give you vanessaE 16:24 nrzkt please wait 16:24 VanessaE wow, lots of gratuitous use of cat there ;) 16:24 nrzkt they send a git version hash 16:24 nrzkt http://minetest-stats.unix-experience.fr/#/dashboard/Monthly-stats?_g=%28refreshInterval:%28display:Off,pause:!f,section:0,value:0%29,time:%28from:now-2y,mode:quick,to:now%29%29&_a=%28filters:!%28%29,panels:!%28%28col:1,id:Version-usage-per-month,row:1,size_x:12,size_y:5,type:visualization%29,%28col:1,id:%27Version-usage-per-month-%28stacked%29%27,row:6,size_x:12,size_y:6,type:visualization%29%29,query:%28query_string:%28analyze_wild 16:24 nrzkt card:!t,query:%27*%27%29%29,title:%27Monthly%20stats%27%29 16:25 sfan5 est31: 0.4.7 or older do not send version info 16:25 VanessaE eeek 16:25 nrzkt please set this link in your browser 16:25 sfan5 god damnbit nrzkt just use a shortener 16:25 nrzkt and look at the graph :p 16:25 sfan5 -b 16:25 VanessaE nrzkt: please bitly it. 16:26 nrzkt http://goo.gl/H3xXis 16:26 nrzkt time can be consuming, please only consult it 16:27 nrzkt i let it public for 15 min, after i stop the service 16:27 nrzkt in the logs empty users agents are only 8 in all logs 16:28 sfan5 VanessaE: gUnknown is android and 0.4.11 or newever as mentioned in my stat files 16:28 VanessaE that's what I figured. 16:29 nrzkt for the graphs you have the top 30 versions here 16:29 nrzkt other version are < 0.1% 16:29 VanessaE so between 0.4.10 and the "gUnknown" category, that's nearly 2/3 of the userbase (in nrz's graph that is) 16:30 est31 sfan5, am I right that 0.4.6 had no custom user agent at all? 16:30 est31 or 0.4.7 16:30 sfan5 0.4.7 or older had no user agent 16:30 est31 ok. 16:30 nrzkt yes and gunknown is 0.4.10 + 16:31 VanessaE 6 times as many users in that range as 0.4.12 16:31 nrzkt it's not single user here 16:31 nrzkt it's single connection 16:32 nrzkt i don't group by IP because i don't have IP 16:32 est31 I just want to find out whether nrzkt is right or sfan5. 16:32 nrzkt sfan5 do'nt give me 16:33 est31 because sfan5 says that 25% have o.4.7 or older 16:33 sfan5 ? 16:33 est31 and nrz says its only 5% 16:33 sfan5 then something is wrong with how nrz parses the data 16:34 nrzkt http://pastebin.com/UqCP1YUx 16:34 nrzkt it's the logstash filter 16:34 nrzkt and here is the regexp used 16:35 nrzkt MINETESTCLIENTLOG \[%{HTTPDATE:timestamp}\] "%{DATA:client_name}/(?:%{DATA:client_version}(?:-%{DATA:git_rev}(?:-%{DATA:git_rev2}(?:%{DATA:git_rev3})?)?)?)? \(%{DATA:os_name}(?:/%{DATA:os_version} %{DATA:os_architecture})?\)" 16:35 sfan5 cat game.log | cut -d '"' -f 6-7 | grep -E "^(Minetest|-)\"" | wc -l 16:35 sfan5 i use that one for 0.4.7 or older 16:36 TBC_x It would be very useful to log ip though 16:37 nrzkt it's only in sfan5 results 16:37 sfan5 the ip is logged 16:37 sfan5 but i did not give nrzkt the data with ips 16:38 sfan5 because privacy 16:38 nrzkt but i could be good to have better stats for me to have the ip, because i could merge version/ip in one result 16:38 sfan5 i doubt that it gives better stats 16:38 nrzkt and don't count 10 users for a signle user launch 10 times its client 16:38 sfan5 well i could do that before giving you the logs 16:38 TBC_x if you run it through uniq, you can get more realistic stats 16:38 nrzkt i need the date 16:39 sfan5 you have the date 16:39 nrzkt with a stupid bash you cannot have uniq entries 16:39 nrzkt if you give me the date 16:39 sfan5 I'm pretty sure you have the date 16:39 nrzkt your regexp didn't match on the first file you give me 16:39 nrzkt yes i have it 16:39 nrzkt but i need to have date + ip + client version to have a single user entry for an interval of time 16:40 nrzkt atm i cannot merge per version/ip 16:40 TBC_x you can get grep output with unique lines 16:40 nrzkt no TBC 16:40 nrzkt you are using a stupid bash 16:40 nrzkt to have a single user over a time interval you need: time, IP and useragent 16:41 nrzkt uniq is stupid, if i have time i cannot merge for a amount of time 16:41 sfan5 wtf 16:41 sfan5 who even said I need to use bash? 16:41 nrzkt you use bash atm :p 16:41 nrzkt and i use a professionnal tool to do stats 16:41 sfan5 actually i use zsh 16:41 sfan5 well ok 16:41 nrzkt yes shell, i use zsh too on my server 16:41 sfan5 are my stats inferior now? 16:42 sfan5 just because i don't use a "professional" tool? 16:42 nrzkt they are not correct because they don't take all the arguments to have real stats. 16:43 sfan5 lolwat 16:43 nrzkt we use kibana at work to do business intelligence based on server logs. This permit us to have some ROI 16:43 sfan5 ofc they are correct 16:43 sfan5 they just count each GET /list as one 16:43 nrzkt i should take my train, we will talk later, have a nice day 16:43 TBC_x define correct data 16:43 sfan5 and dont do any merging 16:43 nrzkt yes, but you don't merge signle user into one entry 16:43 nrzkt this is the most important thing 16:43 sfan5 correct 16:43 sfan5 not really 16:43 est31 no 16:44 nrzkt i can ddos with my 0.4.12 to have 1M entries :p 16:44 nrzkt ++ 16:44 sfan5 well thats nice 16:44 est31 I agree with sfan5 we should value users after how much they use minetest 16:44 sfan5 problem is we can't 16:44 est31 if they use it once a month, they should be valued less than if they use it daily 16:44 sfan5 all we can do is merge entries in a time frame by ip and user agent 16:44 est31 sfan5, we do if we dont count users but requests to /list 16:45 sfan5 we do what? 16:45 est31 then those who make many requests to /list get valued more than those who make few 16:47 sfan5 correct 16:48 TBC_x sfan5: how many unique requests from IPv6 do you get? 16:48 sfan5 http://meow.minetest.net/tmp/serverlist_stats_2015-03-14.txt 16:48 sfan5 the value there should be pretty accurate 16:52 TBC_x can you send me all acesses from 2a00:1028:96c0:4482:21d:72ff:fe09:636e ? 16:53 hmmmm [12:44 PM] I agree with sfan5 we should value users after how much they use minetest 16:53 hmmmm "we" should value users? 16:53 hmmmm we do, those two people do not 16:53 TBC_x sfan5 16:53 sfan5 what 16:53 sfan5 oh 16:53 sfan5 why 16:53 TBC_x Need sample data 16:54 sfan5 for what 16:54 TBC_x for my generator 16:54 hmmmm needless to say i won't approve of any kind of change that breaks compatibility. we should also be on the lookout for sneaky surprises in commits from either of them 16:54 est31 ↑ this 16:56 sfan5 TBC_x: there are none 16:57 TBC_x then send me cat log | tail, including IP, but annonymized to 255.255.255.255 16:58 TBC_x please 16:58 sfan5 haha no 16:58 sfan5 actually i found a few 16:58 sfan5 2a00:1028:96c0:4482:21d:72ff:fe09:636e - - [15/Jul/2015:18:28:17 +0000] "GET /list?proto_version_min=13&proto_version_max=25 HTTP/1.1" 200 140322 "-" "Minetest/0.4.12-dev-8ac0cf5-dirty (Linux/4.0.5-1-ARCH x86_64)" "-" 16:58 TBC_x thanks 17:01 TBC_x are there any line breaks? 17:07 rubenwardy I just did a git bisect, but it didn't work. 17:08 rubenwardy https://gist.github.com/rubenwardy/19a6e332f86431e6f45e 17:08 rubenwardy this doesn't work but it should 17:08 rubenwardy not in 0.4.12-dev or 0.4.12 17:08 rubenwardy it works with minimal, but not with MTG 17:09 rubenwardy and I've had enough of Minetest for today 17:15 nrzkt sfan5: please replace your log parsing with faster parsing by removing this ugly pipes 17:15 nrzkt cat game.log | cut -d '"' -f 6-7 | grep -c -E "^(Minetest|-)\"" 17:15 sfan5 how about you don't tell me what to do 17:15 nrzkt grep -c count lines 17:16 nrzkt this is the first optimize 17:16 sfan5 please 17:16 sfan5 i don't want your lecture about optimizing command 17:16 sfan5 i don't want you lecture about using professional tools 17:16 sfan5 i don't want your lecture about how my statistics are wrong 17:16 sfan5 thanks 17:17 nrzkt then continue to give stats for non unique clients 17:17 kilbith i don't want to learn 17:17 sfan5 you could have told me "you should only count unique clients" 17:17 sfan5 but no 17:17 sfan5 you needed to tell me that the tools i use suck 17:18 nrzkt it's not adapted to a such usage 17:18 sfan5 that i should be using professional tools like you 17:18 est31 bash/zsh/even sh are professional tools too btw. 17:18 nrzkt no 17:18 nrzkt this is a good for basic toolkits 17:18 nrzkt not for parsing millions of lines of log to do stats 17:19 est31 remember the donald e knuth story where he had a page long pascal example, then a guy came around crunching it to a few lines of bash? 17:19 nrzkt this is deprecated and indications are not really correct because shell doesn't use stats algorithms 17:19 est31 pages long* 17:19 est31 nrzkt, either way algorithms are only about speed, not about actual output 17:20 est31 perhaps its faster to have dedicated code, but its not "more correct" 17:22 nrzkt the kibana code talk to elasticsearch to give it stats about entries in the index 17:23 nrzkt for your script sfan5: to win some time you should do : awk -F '"' '{print $2}' game.log | grep -c -E "^(Minetest|-)\"" 17:23 sfan5 stfu 17:24 nrzkt this read one time stdout and one time the file 17:24 nrzkt with your stats it's very slow because you read 1 time the file and 3 times stdout :) 17:24 TBC_x it doesn't matter 17:25 TBC_x there's no random access 17:25 nrzkt and next, improve awk to count the entries itself and do stats :) 17:25 nrzkt okay TBC_x have you parsed 800Mb files with shell ? 17:25 TBC_x not yet 17:25 TBC_x writing generator 17:25 init if we actually an actual performance improvement, we could rewrite it into another language with a compiler that can generate fast binaries 17:25 nrzkt okay, do it and compare 17:26 nrzkt awk can do the action i show you in one call, but i'm not very aware with awk 17:26 init so can perl 17:26 nrzkt you don't need to re-invent a binary which does everything for u 17:26 TBC_x gonna do my own shell hacking 17:26 init and shell, too 17:27 TBC_x btw, where are the useragent strings? 17:27 TBC_x in the source 17:27 sfan5 httpfetch.cpp i think 17:27 sfan5 os version is done in porting.cpp 17:27 TBC_x ok 17:33 VanessaE bbl 18:10 * est31 brings this pr to attention, it gives servers the option to display a "reconnect" button if they restart or crash: https://github.com/minetest/minetest/pull/2922 18:13 nrzkt always +1 for me 18:14 est31 I guess this is a larger change, so it will need 2 +1 18:15 nrzkt ok 18:17 TBC_x theoretically, it could be added to every disconnect 18:17 est31 well, discuss that with nrz, he was against 18:17 est31 original design included that, yes. 18:18 nrzkt if i disconnect from the server i doesn't want to have the "please reconnect" form :) 18:19 nrzkt include that for unneeded reboot is good, for other cases it's just anoying 18:19 TBC_x well... for every involuntary disconnect 18:20 nrzkt crash, server reboot or kick is good 18:20 est31 (aka TOCLIENT_ACCESS_DENIED) 18:20 nrzkt but propose me te reconnect if my password is wrong ? 18:20 nrzkt i will re send the same password 18:20 TBC_x good point 18:20 TBC_x well... for every involuntary disconnect from game world 18:21 nrzkt it's cases covered by this pr 18:21 TBC_x k then 18:21 TBC_x damn... I hate my tension to overengineer everything 19:04 nrzkt that's better :p 19:07 Krock #2948 19:07 ShadowBot https://github.com/minetest/minetest/issues/2948 -- Fix MSVC number conversion warning by SmallJoker 19:18 paramat nore, Tesseract, any comments on adding steelblock to stairs mod, when i add acacia? game#581 19:18 ShadowBot https://github.com/minetest/minetest_game/issues/581 -- New stairs and slabs discussion 21:06 RealBadAngel VanessaE, here? 21:08 RealBadAngel VanessaE, the bug you have blamed me for, was here before 21:09 RealBadAngel wield3d mod is glitchy propably because attachements code is glitchy too 21:11 est31 ^ 21:18 RealBadAngel i think that attached objects lighting never gets updated at all 21:19 RealBadAngel enviroment.cpp line 2357, theres light updated for all the active objects 21:20 RealBadAngel but i cant see how then attached ones could be updated 21:40 RealBadAngel ok, it does update them too 22:05 RealBadAngel est31, here? 22:05 est31 why 22:05 RealBadAngel i know the reason for that glitch 22:05 est31 whats the reason 22:06 RealBadAngel https://github.com/minetest/minetest/blob/master/src/environment.cpp#L2380 22:07 RealBadAngel somehow on VE server pos is always "not ok" 22:07 RealBadAngel in else there ive put dstream<<"Pos not ok, "<