Time Nick Message 00:01 VanessaE (theinvsblman is who found it) 00:14 kahrl must be a really crappily written window manager if it eats that much time 00:15 VanessaE it's xfce. f 00:15 VanessaE -f 00:15 VanessaE I use it also, but I don't see nearly the load that he did. 00:15 VanessaE still moving it out of the title bar is probably a good idea, since the title of the program is not "27fps". 00:15 VanessaE (e.g. don't misuse the title bar) 00:15 Exio4 where is the .patch to try it? 00:16 VanessaE Exio4: [08-25 19:49] lines 3357 to 3366 in game.cpp 00:16 VanessaE comment those out. 00:16 Exio4 ah 00:22 kahrl not going to use them for the mentioned reasons, but I was curious: here are the results of testing bloom filters: http://codepad.org/NPIDm0ta 00:22 kahrl (a very trivial implementation of them) 08:09 sfan5 why didn't we do anything about #881 yet? 10:28 thexyz sfan5: lol 10:28 thexyz meh, I think I shouted at somebody for merging it 10:28 thexyz yet it's here 10:58 Calinou make[2]: *** No rule to make target `po/du/minetest.po', needed by `locale/du/LC_MESSAGES/minetest.mo'. Stop. 10:59 sfan5 run cmake again? 10:59 Calinou works 12:07 proller https://github.com/minetest/minetest/pull/892/files https://github.com/minetest/minetest/pull/883/files https://github.com/minetest/minetest/pull/882 13:00 celeron55 hmm, apparently there is a wiki that automatically scrapes articles marked for deletion in wikipedia 8) http://speedydeletion.wikia.com/wiki/Minetest 18:11 kahrl any idea how the file format for the sha1 set should look like? 18:12 Exio4 hm 18:13 Exio4 why should it have a way to look, isn't it is just a "bunch of hashes"? with a fixed length? :P 18:13 kahrl I'm thinking of [u32] signature [u16] version [u8*20] hash over the rest of the file [u32] number N of following hashes [u8*20*N] hashes 18:14 Exio4 oh 18:14 kahrl am I thinking too complicated? 18:15 kahrl I guess the third field isn't really needed 18:16 proller maybe human-readable? with spaces, one hash per line 18:16 kahrl N can be inferred from the file size 18:17 kahrl proller, I think for something like minetest-cdn.net that would be too big 18:17 proller or.. json 8) 18:17 Exio4 nice joke man! 18:17 kahrl oh god no :P 18:17 proller define "too" 18:17 proller 10G ? 18:17 kahrl proller, the full file is downloaded every time someone is missing some media file 18:18 Exio4 [u32 signature] [u16 version] [u32 N-number-of-hashes] [u8*N*20 hashes] 18:18 proller why download full file? 18:18 Exio4 kahrl: ^? 18:18 kahrl proller: suggest something else? 18:18 proller why not request by hash ? 18:19 kahrl because when many files are missing there are lots of failed requests 18:21 kahrl see the discussion yesterday with thexyz, and he's the one hosting this stuff 18:21 proller if connection gzipped - not big difference in transferred size in bimary vs plain text 18:21 proller ok 18:22 * VanessaE mumbles something about rsync ;) 18:22 Exio4 imagine sending 2000 requests to some server and then sending other 2000 because the first didn't have the media 18:22 Exio4 1000 nodes with two textures each, for example ^ 18:23 proller not very big problem with keepalive 18:24 kahrl not sure if httpfetch works well with keepalive when there a multiple servers 18:24 kahrl are* 18:24 thexyz was my suggestion with asking server what files it has ignored? 18:24 proller maybe compress mods ? 18:24 kahrl thexyz: huh? 18:24 Exio4 wouldn't it need some server-side script? 18:25 Exio4 proller: hmm? mods? how is this related 18:25 thexyz kahrl: http://irc.minetest.ru/minetest-dev/2013-08-25#i_3287247 18:25 kahrl I didn't see that 18:26 proller Exio4, this 1000 nodes lives in ~50 mods, -> download and unpack mods 18:26 proller by hashes too 18:26 thexyz welp 18:27 kahrl in any case there should be a way to host a remote media server without having to run scripts on it 18:28 kahrl could be done by ignoring the POST data and returning the full list 18:28 thexyz true 18:30 kahrl so I think every server but minetest-cdn.net will probably use a static file, because it's easiest 18:30 thexyz why do you need [u8*20] hash over the rest of the file [u32] number N of following hashes then? 18:30 kahrl as I said, I don't 18:31 thexyz you only said you don't need third field 18:32 kahrl N can be inferred from the file size 18:33 kahrl what do you think about binary vs. text? 18:34 thexyz binary 19:08 kahrl http://paste.dy.fi/gXI some leaked code 21:20 harrison http://youtu.be/kJd096BFFUQ 21:41 thexyz harrison: where do I download that? also, how's that related to minetest development? 22:31 PilzAdam proller, have you written this: http://dev.minetest.net/TODO#Rewrite_falling_code_to_c.2B.2B ? 22:31 proller нуы 22:31 proller yes 22:31 PilzAdam how does a queue in the core help here? isnt the problem that there are too many entities? 22:32 proller processing them every step by 1000 in core will 10x+ faster 22:33 proller step queue + update around queue 22:46 thexyz proller: any idea https://forum.minetest.net/viewtopic.php?id=7039 22:48 proller disable GLES ? 22:49 harrison thexyz do not worry your experience is valid 22:49 thexyz do you know why it says -- Found system opengles2 library /usr/lib64/libGLESv2.so;/usr/lib64/libGLESv2.so;/usr/lib64/libSM.so;/usr/lib64/libICE.so;/usr/lib64/libX11.so;/usr/lib64/libXext.so 22:50 thexyz and then make[2]: *** No rule to make target `/usr/lib64/libGLESv2.so', needed by `/var/tmp/portage/games-action/minetest-0.4.7/work/minetest-0.4.7/bin/minetest'. Stop. 22:50 thexyz harrison: hm? 22:50 harrison namaste 22:54 thexyz harrison: o-okay 22:55 proller thexyz, strange, something in cmake