Time Nick Message 00:00 VanessaE (that just doesn't have the same ring as "over 9000", does it?) 00:35 kahrl sapier: minor notes about #1054 (had a quick look at it now) 00:35 ShadowBot https://github.com/minetest/minetest/issues/1054 00:35 kahrl the comment "// Wait result for a second" is wrong now 00:37 kahrl about the MutexedQueue::getList and getMutex methods, would it be possible to make them private and RequestQueue a friend class? 01:02 sapier kahrl I tried to make getList and getMutex private but I can't because of RequestQueue directly messing around in there 01:05 kahrl that's why I suggested making it a friend class 01:06 kahrl another question, are the exception-throwing methods still needed or can they be removed? 01:07 sapier I haven't done a closer investigation on this 01:07 sapier friend class ... of course :-) sorry, need to sort my mind again after counting hundreds of packets 01:08 kahrl hehe 01:08 sapier there's a bug in client connect if packets are received out of order ... I dared to use more then channel 0 so I seem to cause it way more often then before 01:09 kahrl hrm 01:09 VanessaE and OF course it's my servers that trigger the bug more often also, since they have more packets to send, too :_) 01:09 VanessaE :) 01:09 sapier you're the one sending item/nodedefs/announce media messages > 300 packets ;-) 01:09 VanessaE yep :) 01:10 kahrl well it hasn't really been a bug per se, since data inside one channel is sent in order (if sent reliably) 01:10 sapier well ... sending non related data through different channels shouldn't cause a lock up on propper design ;-) 01:15 sapier kahrl any idea how to make a template friend of another template it's derived from? :-) 01:15 sapier stop not derived from 01:16 sapier forget about it wrong class 01:17 sapier but I'd still need to make a template friend of another template ... I'm not that used to templates to know if this is even possible 02:13 kahrl sapier: this should work: template 02:13 kahrl friend class RequestQueue; 02:15 VanessaE kahrl: I gave him a new challenge: duplicate VE-Survival complete with VanessaE_Game + my worldmods (minus the IRC mods) and run and recreate the itemdef/nodedef hang that on server branch :) 02:15 VanessaE right about now, his PC should be a smouldering puddle of steel, fibreglass, and lead. 02:15 VanessaE :) 02:15 sapier doesn't seem to hang for me ... at least not completely ... but client still takes a break of about 8 seconds prior requesting media 02:16 sapier 8 seconds ... that's ages 02:16 VanessaE that explains some of the sign-on lag I always see. 02:17 VanessaE though on really lean servers, that "break" never happens, so I always assumed the client is busy crunching the data it recevied. 02:17 sapier only thing beeing done withing those 8 seconds is ping/pong between client and server 02:17 VanessaE received* 02:18 sapier yes but 8 seconds is quite a lot of time for crunching 02:18 VanessaE true 02:18 kahrl is the client checking for cached media during that time? 02:18 sapier maybe 02:18 VanessaE especially if you're spinning a 2.8 GHz processor doing it. that's a lot of data to chew on then. 02:19 VanessaE kahrl: could be, idk for sure 02:19 sapier as next thing to happen is requesting media it's very likely checking before 02:20 VanessaE even then, 02:20 VanessaE computing hashes really takes this long? 02:20 VanessaE when the hashes are the filenames themselves? 02:20 sapier no but reading from disk 02:20 VanessaE meh 02:21 VanessaE it takes a split second to get a file listing from disk 02:21 VanessaE that can't be it. 02:21 kahrl also sending all those textures to the GPU 02:21 VanessaE kahrl: that's certainly possible. 02:21 kahrl well actually no, that should happen later 02:22 kahrl not sure what exactly happens in IVideoDriver::createImageFromFile 02:23 VanessaE but even then, according to my calculations, it would take 10 times as much data being read from disk as there is data to be read to account for a delay like that. 02:23 VanessaE (at the speeds my particular box is capable of) 02:23 VanessaE so I doubt it's disk read delay 02:27 VanessaE so I assume kahrl's guess is at least partly right - it's the client chewing on textures or hashes something. 02:28 VanessaE kahrl: which reminds me, that no-extrude patch works wonderfully. any chance you can expand that to include a settings menu checkbox and get it pushed to master? it's too good not to have as an option for slower machines I think. 02:31 kahrl for me, Client::loadMedia takes about 1.5ms on average (minetest_game with a few mods) 02:31 kahrl so with 7000+ media files, an 8 second wait to load media is not too surprising 02:32 VanessaE I think he's saying the delay happens after the download is done. 02:32 VanessaE ? 02:32 kahrl no, before requesting media 02:32 VanessaE oh ok 02:34 kahrl I'll have to think about the extrude thing 02:34 sapier at least I discovered a possible memory leak ... guess I'm gonna look for this first 02:35 VanessaE please do. A lot of HD texture pack authors will thank you :) 02:37 VanessaE sapier: oh yeah? 02:39 sapier ok maybe not it's using that amount of memory right after startup 02:39 sapier 2gb memory usage ... crazy game you created vanessae ;-) 02:40 VanessaE lol 02:40 VanessaE it uses only about 150 MB on the server. 02:40 VanessaE and about...oh... well ok on the client it uses a fair bit :) 02:40 VanessaE try it with HDX and your PC will surely melt down :D 07:21 thexyz hmmmm: note that enet is already here and it works fine and integrating it to current codebase was not hard at all 07:23 RealBadAngel hi guys 13:50 thexyz john_minetest: i think pa himself said it was too hacky 21:13 specing Okay, why did you hardcode the -j make option in README? 21:13 specing Should do 21:13 specing make "-j$(($(grep processor /proc/cpuinfo | wc -l) + 1))" 21:13 specing :D 21:15 troller wat? 21:15 troller (grep processor /proc/cpuinfo 21:16 troller $ grep processor /proc/cpuinfo 21:16 troller grep: /proc/cpuinfo: No such file or directory 21:20 Calinou in my script I use something like: "cat /proc/cpuinfo | grep vendor | wc -l" 21:22 Exio4 useless use of cat and wc 21:22 Exio4 $ grep -c processor /proc/cpuinfo 21:22 Exio4 6 21:22 * Calinou puts a cat in a WC 21:22 Calinou :'( 21:23 specing lol 21:44 thexyz no 21:59 thexyz maybe 21:59 thexyz I think someone from here was going to work on it 21:59 thexyz hmmmm, probably 22:01 hmmmm yes 22:01 hmmmm this comes with switching to TTF by default 22:03 troller https://github.com/freeminer/freeminer/commit/75e5895afaec95c530d68847a058aac5338e3c81 22:05 thexyz no one bothered to fix it, if I had to guess 22:05 thexyz do that then? 22:11 thexyz see, it's not that easy 22:23 thexyz yeah, I'm sure there is 22:24 thexyz do that! 22:25 thexyz close! 22:26 thexyz anyway, I think that you'll have to examine FREETYPE_FOUND 22:26 VargaD I'm working on a small lua script that reads all the mods and makes some diagrams: http://althea.no-ip.info/recipe.dot.png 22:26 VargaD What do you think? 22:26 thexyz well you can't just enable it because what if there's no freetype? you don't want it to always fail 22:27 thexyz although maybe this is a good idea actually 22:27 thexyz to encourage package maintainers to depend on freetype 22:27 thexyz VargaD: this is interesting but I'm not sure how could that be useful 22:28 thexyz something that makes it easier to provide documentation for a mod (i.e. by generating craft recipes) would be good though 22:29 VargaD Yes I would like to also generate recipes in some fancy html 22:31 thexyz john_minetest: I don't understand you 22:31 thexyz the Irrlicht font (the small one) is only used for nicknames, no matter whether you use freetype or not 22:31 thexyz *irrlicht builtin 22:31 thexyz when built with Freetype the font size is the same 22:32 VargaD it would be nice to make nicknames bigger 22:32 thexyz so I fail to understand how's it small; still I agree that freetype should be default 22:32 thexyz it does 22:32 VargaD it is so difficult to read it 22:32 Calinou font_size = ? 22:32 VargaD oh thanks Calinou 22:32 VargaD :) 22:46 VargaD it seems I have to parse depends.txt because some mods (mesecons) defines global variables that needed in other submodules... 22:58 thexyz right 22:58 thexyz it could be a better idea to just write a patch to the engine 22:58 thexyz this was already done once 22:58 thexyz (or a mode) 22:58 thexyz s/mode/mod/ 22:58 VargaD hmm, maybe it is possible to do it in a mod 22:59 thexyz https://github.com/cornernote/minetest-game_wiki 22:59 VargaD but you have to query the recipies somehow 22:59 thexyz you can redefine register_craftitem 23:00 VargaD but then you have to be the first mod 23:01 thexyz right, that's why I think this should be a core patch 23:01 thexyz just modify builtin 23:01 thexyz or wait 23:02 thexyz there's minetest.get_all_craft_recipes 23:03 VargaD it works but then other query methods needed to get other information 23:05 VargaD thexyz: thanks game_wiki might be useful 23:06 VargaD I prefer not to use php, static html is safer and less demanding (for my cubieboard) 23:13 iqualfragile VargaD: have you created that pull request yet? 23:14 VargaD I haven't created any pull request 23:14 VargaD Which pull request? 23:17 iqualfragile VargaD: those performance improvements 23:18 VargaD xyzz's ABM improvements? 23:25 iqualfragile VargaD: yes 23:31 VargaD No I haven't, I just finished christmas and got back to my fast computer 23:32 VargaD I would like to make a template ActiveABM class so it's size is conpile time constant, but I would like to see that it is faster... 23:35 VargaD anyway what is the proper way to make that pull request to give credit to xyzz? 23:53 VargaD Hi sapier 23:54 sapier hello