Time Nick Message 00:10 PilzAdam kahrl, https://github.com/PilzAdam/minetest/commit/b94f18298d61bd0a55c042a3becd2f63b2e9c6b0 00:30 kahrl PilzAdam: seems to work well 00:32 PilzAdam is the code acceptable too? 00:32 kahrl the sort_mod_list logic looks reasonable, yeah 00:33 kahrl I think this can be pushed 00:34 PilzAdam https://github.com/minetest/minetest/issues/814 the list is getting shorter 00:37 PilzAdam guiFormSpecMenu.cpp:448, is this supposed to be infostream? 00:38 kahrl or just remove it and guiFormSpecMenu.cpp:428? 00:39 kahrl hmm, some of the other formspec elements print to infostream, some don't 00:39 PilzAdam what about 558? 00:39 kahrl I think all those can be removed 00:39 PilzAdam just greping for std::cout and it seems that sapier use it in more places (mainly in pathfinder) 00:41 kahrl ugh 00:41 PilzAdam guiFileSelectMenu.cpp:107 00:41 PilzAdam that could be removed too, I think 00:42 kahrl yeah all listed in http://paste.dy.fi/JTE can be removed, imo 00:42 kahrl I didn't look at pathfinder yet 00:43 harrison http://youtu.be/3W8cOo_yQtY 00:43 kahrl it even has INFO_TARGET etc. that aren't used 00:50 PilzAdam https://github.com/minetest/minetest/commit/1e4e64f83195ed1be0aa0c4a237a50de6dd132a9 00:50 PilzAdam I think its better to not touch pathfinder.cpp now and ask sapier about it 00:51 kahrl ah callerId wasn't even used... looks good to merge 00:52 kahrl hmm let me think about pathfinder 00:52 kahrl I think those cout uses outside of print_cost, print_type, print_pathlen, print_path, print_ydir, dir_to_name should be changed 00:54 kahrl ERROR_TARGET for all of them? does that spam errors when e.g. mobf vombies are used? 01:06 PilzAdam errorstream should be good, mobf would have bugs if they print all the time 09:19 kahrl what are the plans for the serverlist flags? 09:27 kahrl (I mean the *, C, D, P letters in the public serverlist in the main menu) 09:28 kahrl well I think they are ugly, and not explained anywhere 09:28 kahrl I realize that this would be hard to code but I'd prefer something like http://i.imgur.com/BhlkmWg.png 09:29 kahrl preferably with a tooltip when hovering these icons 10:31 Calinou looks good 10:31 Calinou but padlock makes it sound like server is private 10:31 Calinou I'd remove entirely "empty passwords disallowed" and instead, tell the user to input a password when selecting a password-only server 10:48 kahrl Calinou, yeah 10:48 kahrl showing that flag in the list seems bad for two reasons 10:49 kahrl 1. it encourages players to leave the password blank for servers without the icon 10:50 kahrl 2. it gives incentives to "hackers" to try to find players without passwords on such servers 10:55 Exio4 https://github.com/minetest/minetest/commit/e6687be4933e5115d31ade014300648051af5047#L2R4647 10:55 Exio4 i don't understand why isn't possible to "delete the client" just there 10:57 Exio4 i don't know anything about the network code btw :P 10:57 celeron55 maybe you shouldn't say anything about then 10:57 celeron55 +it 10:58 Exio4 ew :P 10:59 celeron55 if a client is deleted right there from the connection, the access denied message won't generally be sent at all 11:00 Exio4 ah 11:00 celeron55 because of how MT's network layer queues and throttles packets and generally does stuff 11:00 celeron55 and especially it won't get anywhere if the packet gets dropped on the way 11:01 Exio4 ah, thanks 13:08 nore ok, I know where the empty mod pages come from in modstore 13:09 nore it is because they do not have an approved version yet 13:09 nore so engine is not happy 13:10 nore 15:04:20: ERROR[main]: readModStoreModDetails: not a single version specified! 13:10 PilzAdam http://irc.minetest.ru/minetest-dev/2013-08-15#i_3264857 13:27 nore PilzAdam: https://github.com/Novatux/minetest/commit/f73445418c89bf4b0981664f83e880e373b2886a 13:27 nore What do you think of that? 13:31 PilzAdam whats the point of showing mods there that you cant install? 13:32 nore you can get them on the forum 13:33 nore + else it is blank pages 13:33 sfan5 nore: don't forget to remove this debug output: https://github.com/Novatux/minetest/commit/f73445418c89bf4b0981664f83e880e373b2886a#L0R178 13:33 nore https://github.com/Novatux/minetest/commit/f870dcae5bf633ef4b99e4263d2fb6fe2809a55a 13:34 nore I saw it 2 minutes ago... 13:34 PilzAdam if you want to install mods from the forum then you go to the forum or the web frontend of moddb 13:35 PilzAdam the point of the in-game modstore is that you have 1 click installation 13:35 PilzAdam there wont be any blank pages if you just ignore the mods without a version 13:35 nore so the code should check if mods in modlist have a version before reporting them 13:36 PilzAdam should it be checked in core or in Lua? 13:37 nore In core, I think 13:37 nore because it is core that reports error 13:41 nore Or even in the code of mmdb 13:41 nore because the problem comes from here 13:41 nore mmdb should not send mods that have no version 13:45 nore but that should be merged: https://github.com/Novatux/minetest/tree/modstore-pagecount 13:45 nore It fixes pagecount, which is a problem when numbers of mods is a multiple of mods per page 13:46 nore In which case there is one additionnal page, which will be blank 13:48 nore with this patch there isn't that problem 13:48 PilzAdam is the "search" tab supposed to be blan? 13:48 PilzAdam +k 13:49 nore I think it is not yet implemented 13:49 nore yes it is blank 13:50 nore this is normal, since there isn't any code for it 13:50 nore modstore.lua, L64 14:18 nore https://github.com/minetest/minetest/pull/872 14:18 nore a fix for modstore 14:18 nore could someone look at it? 14:18 PilzAdam how do I test it? 14:19 nore You can't ;) 14:19 nore not could I, but you can hand-test it 14:20 nore i.e., if there are 20 mods, you would get 5 pages before where you get 4 now 14:20 nore since math.floor(20/5)+1 = 5 (old way) 14:20 nore and math.ceil(20/5)=4 (new way) 14:20 nore s/=/== 14:27 nore so? 14:44 nore sfan5: anything about #872? 14:45 VanessaE 872? 14:46 nore fix of pagecount in modstore 14:46 VanessaE oh there it is. 14:46 nore currently, it numbers of mods is a multiple of mods per page, one page too much will be shown 14:46 nore s/it/if 14:47 nore the problem is that it cannot be tested 14:49 nore except hand-testing 14:51 thexyz why? 14:51 nore why what? 14:52 thexyz > cannot be tested 14:52 nore sapier, what about #872? 14:52 nore thexyz: need to modify number of mods in mmdb 14:52 nore wait, I know what to do 14:52 nore use mods_per_page = 3 14:53 nore you should have 6 pages, without the pull you get 7 14:53 nore the last one being empty 14:53 thexyz nore: you can just use some fake data 14:53 thexyz or write your own modstore 14:53 thexyz and test against it 14:53 nore I could, but it works, that is sure 14:54 thexyz okay 14:54 thexyz it's not like we have exceptional coverage anyway 14:54 nore retested-it 14:54 nore it works 14:55 nore s/-/\ 14:55 nore ^ this is "\ " 14:57 sapier callerId was relic of a not working signaling method tried in one of the previous versions 14:58 nore sapier: mmdb needs to be fixed so that it does not send mods that do not have a version 14:58 nore because you get tons of blank pages else 14:58 sapier exactly iqualfragile is informed about this 15:01 sapier what's 872 good for? 15:02 nore this fixes the fact that there s one additional page if number of mods is a multiple of modsperpage 15:02 sapier "box[11.6," ..scrollbarpos .. ";0.28,0.5;32CD32]" ok now I understand why color difference between #AAAAA and AAAAAA was mentioned yesterday ;-P 15:02 nore and the fact that pagecount should be the number of pages, not number of pages-1 15:03 sapier ok I guess you tested this 15:03 nore yes 15:03 nore I set modsperpage=3 15:03 nore and it needs the pull to work correctly 15:04 nore because there are 18 mods in mmdb currently 15:05 nore so 7 pages without the patch, 6 with 15:05 sapier ok seems to be fine then ... the empty page problem is to be fixed by iqualfragile ... yes of course we could build a shadow copy of modlist within menu but that's a lot of code and lots of additional bugs 15:07 nore btw, do you know if my 5 commits are enough to go in "contributors"? 15:08 PilzAdam sapier, is it really that complicated to simply ignore mods with no version? 15:08 sapier I don't know of any rules when you're called contributor ... but I guess it's more "continued contribution" than puree number of commits 15:08 nore rubenwardy: you will have to rebase you mainmenu change 15:09 nore PilzAdam: the problem is that now we need to get modstore details 15:09 nore for each mod 15:09 sapier pilzadam problem is that mod details and mod list is separated 15:09 sapier modlist doesn't contain information about versions 15:09 nore because else you don't know what version 15:09 nore and it is long to get mod details 15:10 nore that's why modstore takes long to load 15:11 sapier of course this is nothing complicated but quite some work to get it right, while it's as simple as skipping those without version in moddb 15:11 nore so the change should be done in mmdb 15:11 sapier so either about 500 lines of code in minetest or 20 lines in moddb ... guess what is more likely to have bugs ;-) 15:12 nore sapier: is the mod search not finished yet? 15:12 nore because its tab is empty 15:12 sapier it's not even started ;-) 15:12 PilzAdam so we dont check the validity of data but rather just rely that everything we get is good? 15:13 sapier pilzadam we check it and print an empty element 15:13 nore validity is checked when getting details 15:13 nore but not when getting the list 15:13 sapier you don't read whole minetest map on startup and check for block errors .. do you? 15:13 nore of all mods 15:13 PilzAdam invalid data causes visual bugs, though 15:13 sapier invalid map data causes minetest to crash 15:14 sapier it's a tradeoff do you want a menu that is slow or a menu that is terribly slow? 15:14 sapier or a menu showing outdated information 15:15 nore what should be done is that minetest sends its version to mmdb, and mmdb returns only compatible mods 15:15 nore that way, so outdated info, all mods compatible 15:15 nore s/so/no 15:16 sapier iqualfragile suggested that crazy http header addon because of mmdb was able to render information different for different clients, so here's the usecase for that feature 15:16 celeron55 what 15:16 sapier mmdb uses a special http header 15:16 celeron55 no, minetest versions isn't a thing that mimetypes should be used for 15:17 celeron55 mmdb uses mimetypes to know whether a client wants html or json, nothing more than that 15:17 sapier not exactly 15:17 sapier there's aversion too 15:17 celeron55 what you want is just a ?mtversion=0.4.7 to the GET url 15:17 sapier not same as minetest version of course 15:17 sapier no a json client version 15:17 celeron55 lol 15:17 celeron55 it's a protocol version 15:17 celeron55 not client version 15:18 celeron55 it describes the format of the data, not what the client is 15:18 sapier call it what you want it's a version specifying how information is to be presented 15:18 celeron55 yes; and that means you don't use it for what you said 15:18 sapier why? 15:18 celeron55 _how_ information is to be presented; not _what_ information 15:20 sapier the difference between _how_ and _what_ is something very artificial 15:20 sapier structure of data is information too 15:21 sapier but if you insist on every information beeing used we need to update protocol version and add number of versions to overview data too thus everything will work too 15:21 celeron55 you need to maintain artificial limits in order to preserve sane structure and separation 15:22 sapier don't talk about sane structure the information presented by overview and details doesn't fit at all to what information is needed for minetest 15:23 sapier maybe a sane new protocol version would be an option too instead of adding workarounds everywhere 15:23 celeron55 asking for mods for a certain version of minetest is clearly a "search field" kind of thing; a query parameter 15:24 celeron55 i'll not argue more because you simply can't say anything that would make me think otherwise 15:24 sapier you are aware that overview currently isn't enough to tell if that mod could be displayed in minetest or not? 15:25 nore you could try to load the webpage and parse it ;) 15:26 sapier modstore uses overview information to count how many pages of mods are available 15:26 nore but that is definitely too complicated 15:26 sapier details are only downloaded for current page 15:27 nore or the other way could be to modify overview 15:27 sapier if we need to download full moddb prior showing anything you're goona have to drink a lot of coffe ;-) 15:27 sapier what do you suggest nore? 15:27 sapier show a mod but hide the download button? :-) 15:28 nore either modify mmdb so you ask for your version, either supported versions are in overview 15:28 nore sapier: I did that, but PilzAdam said it was useless 15:28 sapier I don't like to tell that as usually but pa is right about that 15:29 nore I know... :) 15:29 nore but #872 needs to be merged nevertheless 15:29 sapier does mmdb have a interface to get overview limited to a passed version? 15:29 nore whatever is made 15:29 nore dunno, I did not do it 15:30 sapier celeron55 the I call it finished method didn't work for german politician profala too ;-) 15:32 PilzAdam sapier, can you start using punctuation for other things than smileys? 15:32 sapier As celerons point of view "don't limit information to what can be understood by client" isn't generally wrong there are only two options 15:32 sapier 1 completely new proctocol version including version count of mod in overview 15:32 sapier pa I once tried to use punctuation but got it wrong everytime 15:33 sapier or 2 add support for querying mmdb for mods containing a special version 15:33 PilzAdam sapier, btw, http://irc.minetest.ru/minetest-dev/2013-08-16#i_3265384 15:34 sapier already seen it obviously I forgot to replace std::cout by those targets defined above in some places 15:34 PilzAdam would you agree that errorstream should be used? 15:35 sapier I'd suggest using ERROR_TARGET for those not beeing within the debug ifdefs 15:37 sapier do you wanna do it yourself or shall I supply a pull request? 15:37 PilzAdam I can do that 15:38 sapier but plz don't remove the debug code it's going to be usefull if there are bugs in ;-) 15:38 nore https://github.com/minetest/minetest/pull/873 15:38 nore we spoke of this a few days ago 15:39 sapier lol I guess that one was written before get_texturepath was available 15:39 nore yes, it was... 15:40 nore get_texturepath was with TP selection 15:40 PilzAdam sapier, what about the one in #ifdef PATHFINDER_CALC_TIME in line 344? 15:41 sapier it's a debug feature to so no use to be enabled if not debugging 15:41 nore PilzAdam: can you merge #872 and #873 since those are only bugfixes? 15:42 nore or is a second dev needed? 15:42 sapier I guess 344 can be changed but imho its no difference 15:42 PilzAdam https://github.com/PilzAdam/minetest/commit/75a139e62cccd7c2b053a1ae6ed7f0fed5abdf63 15:42 PilzAdam did I miss anything? 15:43 sapier I don't think so, seems to be fine. 15:44 sapier ^used punctuation ;-) 15:44 nore PilzAdam: when will https://github.com/minetest/minetest/pull/856 be merged? 15:45 nore about 10 mods need it... 15:45 nore if not more 15:45 PilzAdam 873 seems good 15:47 PilzAdam Ill merge 872 too, since sapier agrees on it 15:48 nore and do you agree for 688? 15:48 sapier PA be carefull I haven't tried 872 15:48 nore RealBadAngel was ok for it, IIRC 15:49 sapier 688? it's a feature addon not a bugfix right? 15:50 nore yes it is, but if PA agrees, that makes 2 core devs agreeing 15:51 sapier I'm not sure about how this is done if after_use returns false nothing is done at all? 15:51 sapier Is this same as for other similar things in mt? 15:52 nore yes, I was asked to change that, see the discussion 15:52 sapier oops sorry didn't read that 15:52 nore but yes, it should return an itemstack that will replace the tool, or nil to make the tool not wear out 15:53 nore and if you wan to wear the tool as it would have, you can use the dp parameter 15:54 nore s/wan/want 15:54 sapier as long as it keeps compatible ... that thing was already changed some time ago ... the new version wasn't quite easy to understand ;-) 15:55 PilzAdam re 856: wasnt there something with entities too? 15:55 nore with entities? 15:56 nore do you mean, allow or not entities to interact? 15:56 PilzAdam yea, entity interaction should call can_interact() too 15:56 PilzAdam -yea 15:56 nore entity on_punch, etc 15:56 nore that would need to modify the engine, no? 15:56 PilzAdam btw, minetest.can_interact() should be moved to item.lua 15:57 PilzAdam or not.... I think it can stay in misc.lua if its for entities too 15:57 sapier I don't like the idea of calling from core to lua only to check for interact 15:58 PilzAdam sapier, and why? 15:58 sapier you do at least two stack changes everytime while it's an empty callback most of time 15:59 sapier as long as only slow users cause this it most likely won't be an issue but if this is done more frequent it's something different 16:02 sapier kahrl I'd prefere your version of server flags too .. I guess it's not that difficult to add if you use images for all combinations of flags you may be able to extend textlist by icon support and set the icon according to that flags 16:04 PilzAdam sapier, is that really a problem? on_punch is called too 16:06 nore we should need a separate callback for entity interaction, no? 16:06 nore not the same as node insteration 16:06 PilzAdam nore, why? 16:06 sapier I don't know if it's a problem in this special case in general I'd avoid doing calls from lua <-> c ... but of course sometimes there's no other chance to do it I haven't thought enough about this special case to be sure about it 16:08 PilzAdam nore, you would pass pointed_thing to the callback 16:11 PilzAdam re 688: all the other tool callbacks use default functions (like minetest.item_place) in the default definition table, that should be done with after_use() too 16:11 PilzAdam on the other hand we want to move away from redef_wrapper() 16:38 sokomine thanks for adding the information message when the map fails to save due to sql problems (in this case: disk full). that helped me a lot and alerted me to the problem. i'm glad that it has been added! 17:28 PilzAdam sapier1, I get "18:27:21: ERROR[main]: Jsonreader: https://forum.minetest.net/mmdb/mods/ not found (SSL connect error)" in a win build in wine 17:28 sapier1 does it crash? 17:29 PilzAdam no, just shows no mods 17:29 sapier1 I'd guess a wine ssl error 17:29 sapier1 maybe certificate check not ignored or something like that? 17:30 sapier1 or crypto algorithm not implemented 17:30 nore IIRC, I had a similar bug when compiling against libcurl4-nss-dev 17:30 nore and that got fixed when compiling against libcurl4-openssl-dev 17:30 sapier1 it's a curl error so nothing to fix in mt 18:58 Tesseract Selection boxes longer than four or so don't seem to work. Known bug? 21:42 sokomine who had the brilliant idea to loose air when there's a torch? 21:44 PilzAdam sokomine, what? 21:46 sokomine i was standing on a server...idling shortly due to rl...and when i came back i was dead and my stuff was gone. apparently this was due to a torch that suffocated me 21:47 kaeza lol 21:48 PilzAdam sokomine, that happens if you use the unstable dev version 21:48 PilzAdam the server had an outdated 0.4.7 git version installed, while you have a newer client 21:51 kahrl people die of smoke inhalation in RL... 21:51 sokomine yes, they do so rl. but hardly when there's only a torch nearby and plain air all around 21:53 Tesseract It seems that loading a area with vm:read_from_map() every step isn't enough to keep it loaded. I suppose that the server unloads it as soon as it notices that there are no players around. 21:53 PilzAdam sokomine, basically its this commit: https://github.com/minetest/minetest/commit/7b13d119ed917c137fc375eff790a7754fd93386 21:53 PilzAdam if a server prior that commit and a new client connect then the client drowns in all nodes except ignore and air 21:54 PilzAdam (both, server and client must be > 0.4.7) 21:59 sokomine ah. did take a look now and almost drowned while standing at the same node as a sign 21:59 sokomine and that's repaired now? 22:00 Tesseract sokomine: Yes. 22:01 PilzAdam what do you mean by "repaired now"? 22:01 PilzAdam its a incompability between 2 unstable dev versions, we dont really care about that 22:03 sokomine well, if it doesn't happen anymore if client and server are newest i do consider that repaired :-) it would have been bad if it had been intentional 22:04 PilzAdam Tesseract, the vmanip is not meant to force load parts of the map 22:11 Tesseract PilzAdam: I know, but it is the only method we have. 22:25 kahrl Tesseract: maybe you could position a bot client (which does nothing but stay connected) in the mapblock you want to keep loaded 22:28 kahrl huh, did github change their IP address to 192.30.252.131? 22:28 kahrl git push is warning me 22:29 Tesseract kahrl: Oh, yes, that sounds like a easy thing to do that isn't hacky at all. 22:29 PilzAdam kahrl, no warnings here 22:31 kahrl https://twitter.com/search?q=192.30.252.131&mode=realtime&src=typd 22:41 thexyz check whois 22:41 thexyz OrgName: GitHub, Inc. 22:48 kahrl maybe they added some servers because of the ddos 22:49 kahrl updated diaganim: https://github.com/kahrl/minetest/commit/575e42ec4318d966aaa2616d54341c16f922c1d7 22:50 kahrl this now properly sets the texture coords so that adjacent nodes that flow in the same direction tile 22:55 PilzAdam that looks good, especially with lava flowing down a hill 22:59 kahrl I'll put up a pull requests and get some more opinions. Or merge it right away? 22:59 kahrl -s 22:59 PilzAdam I dont see any reason to not merge it 23:00 kahrl alright 23:03 PilzAdam what about https://github.com/minetest/minetest/pull/871 ? 23:07 kahrl looks reasonable 23:13 kahrl tested it in singleplayer and it doesn't break anything from what I can see 23:14 kahrl I assume you tested it in multiplayer as well? 23:14 PilzAdam of course 23:14 kahrl I think it can be merged then 23:14 PilzAdam hm, if a player punches an object then the client would apply the [brighten as prediction, and when recieving the punch command then it would apply that [brighten again, wouldnt it? 23:16 kahrl yeah but that shouldn't be a problem because updateTextures always starts with m_prop.textures 23:17 kahrl and that isn't changed 23:17 PilzAdam yes, but on laggy servers you would see 2 seperated flashs, the second one without any visible reason 23:17 kahrl right 23:18 proller to discussion: http://dev.minetest.net/Formspec_json 23:18 PilzAdam it could be solved by using m_hp - result_hp for damage, instead of the damage that is send 23:21 kahrl PilzAdam: that should work, yeah 23:24 PilzAdam I merge it then 23:25 * kahrl approves 23:27 PilzAdam proller, why dont we write a parser for Lua tables to current format, so it is readable and backwards compatible? 23:29 PilzAdam btw, the current format is way better readable than json (in your example) 23:29 proller because current format is positioned and bad 23:30 PilzAdam "positioned"? 23:30 proller in json you can specify random number of keys everywhere 23:31 proller image[5,6;7,8;b.png] -- depends on number in "array" 23:31 proller you cant skip for default value 23:32 PilzAdam if you want to break compatibility then use BKVL 23:33 kaeza proller, that is not exactly... readable. sorry 23:33 PilzAdam kaeza, a Lua table parsers needs to be done either way, so thats not a problem 23:34 proller it must read cpu, not you 23:34 kaeza PilzAdam, see thexyz's minetest-stress 23:37 proller kaeza, json or lua call can be formated and multiline 23:37 proller and looks like here https://github.com/xyzz/minetest-stress/blob/master/tests/test_formspice.lua