00:40 < eeeee> is ddnet server valgrind-clean atm? 00:44 < Savander> goood night 00:44 < Savander> ;) 00:54 < Spyker> deen, here is a nuub called teeworlds admin who spam and change ip all time, he is insulting artkis dyslate france etc.... can you block him? impossible to play... too much spam 00:54 < Spyker> now like 10 teeworlds admin in the server 00:55 < Spyker> teeworlds admin in nickname 00:56 <@deen> Spyker: hi 00:56 < eeeee> if somebody there has rcon you can just password the server for like 10 minutes, he'll likely get bored and not come back 00:58 <@deen> looks like that guy has a lot of ips 00:59 < Spyker> thanks deen, nice admin... i have to go now bb! 01:00 <@deen> eeeee: ehm, good question. last time i checked was long ago, and I only checked for memory leaks I think. what do you have in mind? 01:00 < eeeee> deen: can you look up some of that guy's ip in http://www.spamhaus.org/lookup/ ? 01:01 < eeeee> i'm interested in how often such people use proxies as opposed to just reconnecting and getting different address from the isp 01:01 <@deen> he's just reconnecting 01:01 < eeeee> how long does it take him to reconnect? 01:02 <@deen> just a second 01:03 < eeeee> about valgrind, it reports some still reachable stuff on mac os (but most of it from *Mach* libs), i was wondering if it's a mac os thing or a known tw issue 01:06 <@deen> how do you run it? 01:06 < eeeee> just valgrind --leak-check=full ./DDNet-Server_x86_64_d 01:07 <@deen> I get a few "Conditional jump or move depends on uninitialised value(s)" 01:08 < eeeee> yeah that thing has been in tw forever 01:08 <@deen> and "Use of uninitialised value of size 8" 01:08 <@deen> in SendPacket 01:08 <@deen> nothing else 01:09 < eeeee> when you stop the server it will report the blocks that weren't freed 01:09 <@deen> oh, right^^ 01:09 <@deen> getaddrinfo losing a lot^^ 01:10 <@deen> only getaddrinfo 01:10 < eeeee> okay good to know 01:10 <@deen> wait, do they actually get lost or was that a misreport by valgrind? 01:10 <@deen> because leaking 6 KB in a few minutes already would be bad 01:10 < eeeee> i can take a look 01:11 < eeeee> but have to filter out all this macos crap 01:11 <@deen> I remember that I took a look before, just don't remember my conclusion^^ 01:11 <@deen> net_host_lookup (system.c:922) 01:12 <@deen> http://ddnet.tw/valgrind.txt 01:13 <@deen> should probably call freeaddrinfo() earlier if result != 0 01:13 < eeeee> yep 01:15 <@deen> nope, doesn't change anything 01:16 <@deen> https://bugzilla.redhat.com/show_bug.cgi?id=116526 01:17 <@deen> hm no, looks fixed^^ 01:17 <@deen> it also leaks on mac osx or is it my weird compile of glibc? 01:17 < eeeee> would hope so, it's been >10 years :D 01:17 < eeeee> i also found that in my mac os report 01:17 <@deen> this looks more recent: https://bugs.launchpad.net/ubuntu/+source/eglibc/+bug/1364584 01:17 < eeeee> will try the freeaddrinfo thing 01:21 < eeeee> yeah doesn't help on mac os as well 01:21 < eeeee> still leaking 01:26 <@deen> it's weird 01:26 <@deen> we use getaddrinfo and freeaddrinfo in the exact same way as in the example in the man page 01:26 <@deen> and the example from the man page does NOT leak according to valgrind 01:26 < eeeee> huh 01:27 < eeeee> maybe it leaks only on specific types of addr 01:27 <@deen> or valgrind is misreporting? 01:28 < eeeee> that's easy to check 01:28 < Learath2> remove all addr related stuff check if it sill leaks ? 01:29 < eeeee> call getaddrinfo and freeaddrinfo over 9000 times and watch the mem usage :P 01:29 < Learath2> or that :D 01:30 < Learath2> for(;;){getaddrinfo(); freeaddrinfo();} 01:30 <@deen> it couldn't have to do with being C code inside of C++? 01:31 <@deen> I don't know how valgrind works and if it could have problems with that 01:31 < Learath2> YAY autoupdate works finally 01:31 < eeeee> can't you compile your manpage example with c++ and check it? 01:32 <@deen> I did, works as well 01:32 <@deen> Learath2: great! 01:32 < Learath2> deen: needs some tidying up and removing of weird debug messages 01:33 <@deen> getaddrinfo is really weird 01:33 < Learath2> dbg_msg("FUCK", "THIS SHIT %s", aVersion); 01:33 <@deen> if I do "while(1) e = getaddrinfo(host, NULL, &hints, &result);" 01:33 <@deen> it actually leaks memory 01:34 <@deen> if I also call freeaddrinfo it seems to leak as well, for about 30 seconds, and then stops leaking 01:35 < Learath2> thats really weird 01:35 < eeeee> yeah whatever. doesn't sound like something worth wasting time on 01:36 <@deen> I guess that was my old result as well :P 01:37 < eeeee> here's what it looks like on mac os http://pastie.org/private/kzcvgl90henpxuv80unaa 01:40 < ddnet-commits> [ddnet] def- pushed 1 new commit to DDRace64: http://git.io/vxnggw 01:40 < ddnet-commits> ddnet/DDRace64 55bfd2c def: Always call freeaddrinfo 10:55 < eeeee> got websocket protocol working in tw server \m/ 10:56 < eeeee> still has to figure out the net_socket_read_wait part, otherwise ping to localhost is around 35ms :S 11:03 < eeeee> haha also serverbrowser is kinda useless because it takes like a second to negotiate websockets session 13:07 < Nimda> Generic World by Loop just released on Brutal at 2014-12-28 13:07 13:15 <@deen> eeeee: definitely need a cached serverbrowser of ddnet servers :P 17:23 < Learath2> deen: should i restart the client automatically ? 17:23 < Learath2> after an update 17:23 <@deen> Learath2: yes, but give a warning before, as it happens now 17:23 < Learath2> Do you want to restart now ? (Yes) (No) 17:23 < Learath2> good enough ? 17:24 <@deen> No, before the update even, I think 17:25 <@deen> "An update to DDNet client is available. Do you want to update now? This will restart the client." 17:25 < Learath2> I do already ask before the update deen 17:26 <@deen> ok, good 17:26 <@deen> that should be good enough i think 17:56 <@deen> I made http://ddnet.tw/stats a bit faster 18:00 <@deen> hm, and less accurate^^ 18:12 < Nimda> #wontfix by Waschlappen & Broken just released on Moderate at 2014-12-28 18:08 18:54 < Savander> maybe, you should add possibility to choose the date ( e.g. from 2014-12-25 t o 2014-12-28 ) with more accurate stats :P? 19:38 < Muttley> Deen, any interrest for a DDNet NL addition ? 22:43 <@deen> Muttley: hm, ping from Netherland to GER should be pretty good I guess 22:45 < Muttley> deen i know, but just as an extra ;-) 22:45 <@deen> can you guarantee good ddos protection? 22:45 <@deen> otherwise it would be more of a pain than help 22:45 <@deen> european ddnet servers get ddos attacked a lot 22:49 <@deen> So, 2 people have their computer beeping at random times in DDNet client 22:49 <@deen> anyone know what that is? 22:53 < Savander> keyboard failure? 22:55 < eeeee> bell character in console? 23:06 <@deen> eeeee: aaaah, that would be fun! 23:07 <@deen> eeeee: Maybe someone is running around and spamming bell character on all servers! 23:07 <@deen> should probably just filter that out 23:27 <@deen> Savander: ideally ddnet.tw/stats would use highstock with dynamically loaded data. not sure if that's worth all the work though^^ 23:27 < Savander> in counter-strike 1.6 you was able to open cdrom 23:27 <@deen> Savander: haha 23:27 < Savander> rlt 23:27 < Savander> rly 23:27 < Savander> via server 23:27 < Savander> :P 23:28 <@deen> something like this: http://www.highcharts.com/stock/demo/lazy-loading 23:28 <@deen> you can zoom in and it loads more data 23:28 < Savander> yeee 23:28 < Savander> in google stats it looks simillar ;d 23:29 <@deen> if you can translate ddnet.tw/stats to work like that I'd be glad^^ 23:30 < eeeee> bell character might not be malicious, maybe unicode support is just broken in terminal and it sees 0x07 (or whatever the BEL code is) which is actually part of some unicode character 23:30 <@deen> sigh, right 23:30 <@deen> we *did* change the unicode stuff in windows recently 23:30 <@deen> there was a pull request by bear 23:31 <@deen> ah no, by heinrich5991 23:31 <@deen> https://github.com/def-/ddnet/commit/92cdb156794496b27a709face5df28f7090ee0fc