09:11 < ackpacket> What reason would a 1.7.x server have to use it anyway?
09:14 <+sadimusi> construct looks a lot nicer than I remembered it
09:17 <+sadimusi> SpaceManiac: how's glowstone coming along btw? do you think it will ever be a viable replacement for the vanilla server?
09:18 < ackpacket> I don't suppose there's a way to ignore encryption altogether if the client chooses to do so?
09:18 <+SpaceManiac> Pretty decent. I've been a little slow working on it because almost all the easy problems are solved
09:18 <+sadimusi> ackpacket: no, but the server can
09:19 < ackpacket> right, in offline mode
09:19 < ackpacket> I think it also happens with local servers
09:19 < ackpacket> should be up to the client haha
09:23 < ackpacket> Yeah something looks fishy about this packet
09:25 <+SpaceManiac> sadimusi: there's a lot of situations (like PvP, mobs, and the stupidly obscure adventure world features) where Vanilla will always be stronger, but I'd say it's possible
09:26 <+sadimusi> SpaceManiac: "Somewhat-complete inventory support." doesn't exactly sound confident ;)
09:28 <+SpaceManiac> the click handler logic is complicated! :P
09:28 < ackpacket> hmm
09:33 < ackpacket> Can anyone see this?
09:34 <+SpaceManiac> Yes.
12:10 < ni291187> hello
12:11 < ni291187> i want to write a custom client and I am gathering information
12:11 < ni291187> http://wiki.vg/Protocol_FAQ
12:12 < ni291187> but, Sends a position + look packet to confirm the spawn position
12:12 < ni291187> how do I know the player' ss pawn position?
22:31 < benbaptist> Did the Minecraft server query change somehow in the most recent snapshots? My server script works just fine when querying on a 1.7.2 client, but fails on any recent snapshot.
22:31 < benbaptist> such as 14w25b
23:24 < benbaptist> Alright, nevermind. I went and extracted a real response from a 14w25b server, and it worked fine when I printed that out. Apparently the example response on http://wiki.vg/Server_List_Ping#Client_-.3E_Server is not enjoyed by 14w25b. I was just using that as a test. 
23:24 < benbaptist> so that was the issue.
23:25 <+SpaceManiac> benbaptist: that's the old, pre-1.7 response format. It's possible the snapshots have stopped understanding it
23:25 < benbaptist> SpaceManiac, No, there's 1.7 packets at the beginning.
23:26 < benbaptist> you scrolled down too far, to the 1.6 and before area.
23:26 <+SpaceManiac> You linked to that section
23:26 < benbaptist> oh, whoops.
23:26 < benbaptist> That wasn't what I intended to link to - odd.
23:26 < benbaptist> I was working with the 1.7 selection, though. 
23:27 <+SpaceManiac> Oh, it's possible something was crazy with the favicon, I suppose, since it's not real data in thst example
23:27 < benbaptist> yeah. I tried removing it, but I actually had other issues to fix before that, and so maybe removing the favicon would've helped.
23:27 < Thinkofdeath> benbaptist: fixed the example, the uuid used didn't have the '-' which causes issues with newer clients
23:28 < benbaptist> Thinkofdeath, Ahh, nice!
23:28 < Thinkofdeath> 1.7.5< had no '-', 1.7.9 added them back
--- Day changed mer. juin 25 2014
00:12 < benbaptist> are there any Python-based SMP projects around that I could take a peek at the code for the encryption? 
00:12 < benbaptist> I'm a bit stumped with how to do protocol encryption. :P
00:19 < dx> there's a few in http://wiki.vg/Client_List and http://wiki.vg/Library_List and one server https://github.com/bravoserver/bravo/
00:20 < dx> the encryption code hasn't changed a lot since 1.3 afaik
00:20 < dx> so even if they are marked as outdated it's going to be useful
00:22 < dx> and this is probably the only python 1.7 implementation https://github.com/nickelpro/spock
00:26 < benbaptist> yeah
00:26 < benbaptist> thanks man!
02:00 < benbaptist> Okay, ASN.1 scares me. Am I forced to use encryption as the server, or can I somehow bypass it, but still use authorization?
02:02 < dx> benbaptist: nope
02:03 < dx> benbaptist: but nobody's forcing you to rewrite all the code from scratch either
02:03 < benbaptist> dang. I'm going to need to search further in the code I've looked at, to find the encryption stuff.
05:47 <+sadimusi> benbaptist: everything you need should be in here https://github.com/sadimusi/mc4p/blob/master/mc4p/encryption.py
06:05 <+sadimusi> benbaptist: although I'd suggest you use tlslite instead of pycrypto
06:06 < benbaptist> sadimusi, Ah, okay. Thanks man! Is tlslite a non-standard module?
06:06 <+sadimusi> yes, but unlike pycrypto it's pure python
06:06 < benbaptist> Ah, awesome! 
06:06 <+sadimusi> and it will use pycrypto if it's available to speed things up
06:09 < benbaptist> sadimusi, should I just utilize the whole encryption.py file, since it's under the wtfpl?
06:10 <+sadimusi> sure, if you want to
06:10 < benbaptist> Thanks so much man, you're a life saver!
06:11 < benbaptist> I've been so stumped on this for weeks. :P
06:11 <+sadimusi> just be warned that installing pycrypto on windows is a major PITA
06:12 < benbaptist> yeah, I'm okay with that. My project, Wrapper.py, doesn't even work on Windows anyways thanks to the standard Python bundle for windows not having pkg_resources which I use for reading HTML files out of the package for the web server aspect.
06:23 < benbaptist> What's the "Verify Token" field on the Encryption Request packet? 
06:23 < benbaptist> this page partially confuses me, as it seems to be written from the client's POV as opposed to a neutral POV. http://wiki.vg/Protocol_Encryption
06:24 <+sadimusi> what makes you think so?
06:27 < benbaptist> I might just be overlooking something, but it seems like it skips over important information about how the server does tasks. I don't get where the Verify token field comes from. 
06:27 <+sadimusi> it's random
06:28 < benbaptist> ah, okay. I can just throw random characters at it?
06:28 <+sadimusi> but you're right, that isn't explained anywhere
06:29 <+sadimusi> random bytes even
06:29 <+sadimusi> the vanilla server uses 4 iirc
06:29 < benbaptist> okay
06:29 <+sadimusi> it's just to prevent replay attacks
07:34 < benbaptist> wait, so since the server ID string is empty, do I even need to worry with doing anything with sha1.update()?
07:35 <+SpaceManiac> You need to update with the server ID string, since some servers do use it
07:35 < benbaptist> I'm the server though, so do I really need to do anything?
07:36 <+SpaceManiac> Oh, in that case I imagine it's fine
17:06 < shoghicp> hmm, what is the difference between UUIDv2 and UUIDv3?
17:07 < humerusj> shoghicp: already have your test?
17:07 < shoghicp> yeah :)
17:13 < Not-cdc9> [mcprotocol] thinkofdeath pushed 2 commits to master [+0/-0/±107] http://git.io/xKcVgg
17:13 < Not-cdc9> [mcprotocol] thinkofdeath bcc948b - 14w26a
17:13 < Not-cdc9> [mcprotocol] thinkofdeath 922cc39 - 14w26b
19:42 < benbaptist> Where does the private key in protocol encryption even come from? This doc seems to breeze over that. http://wiki.vg/Protocol_Encryption#Server
19:44 < shoghicp> benbaptist: it's generated by the server
19:52 < benbaptist> Apologies. My system just had a freak out.
19:53 < benbaptist> Screen went black and when it came back, there were visual glitches and no internet connection was available. :P
19:54 < shoghicp> did you get my reply?
20:27 < mbaxter> In case you missed it, shoghicp replied that it's generated by the server, benbaptist 
20:27 < benbaptist> Oh!
20:28 < benbaptist> I'm writing the server-side portion, so what precisely do I generate? Just random bytes?
20:29 < shoghicp> benbaptist: a 1024-bit RSA private key
20:29 < shoghicp> and the public key :)
20:32 < benbaptist> ah, okay. I have the public key already
23:25 < mpa1212> Does anyone happen to have a lua version of the examples seen on http://wiki.vg/Server_List_Ping ?
--- Day changed jeu. juin 26 2014
03:15 < benbaptist> Hmm, I don't mpa1212. I might try to figure it out later though if you're still here. :P
03:26 < mpa1212> benbaptist: :P I'd love it if you would try! I threw my novice programmer self at it for a while earlier today, but got in trouble with lua's lack of explicit data types
03:29 < Altenius> Is anyone here good with the network protocol?
03:30 <+SpaceManiac> A lot of people, probably. What's the question?
03:31 < Altenius> I'm having trouble reading the packets.
03:32 < Altenius> I made a listener on port 25565, and I tried to connect to localhost.
03:32 < Altenius> It gets three bytes, but the bytes are "254", "1", "250"
03:33 <+SpaceManiac> Sounds like the status ping for 1.6 and older
03:33 < Altenius> Yeah
03:33 < Altenius> I'm testing on 1.6.4
03:33 < Altenius> It's different on 1.7?
03:33 <+SpaceManiac> That's right
03:33 <+SpaceManiac> 1.7 is actually a massive change in the protocol structure
03:34 < Altenius> So http://wiki.vg/Protocol#Handshaking is all 1.7?
03:35 <+SpaceManiac> Yeah, the whole Protocol page is for the current version
03:35 < Altenius> Alright, thanks.
03:47 < mpa1212> benbaptist: If you want to look at (the terrible) code I've smashed together (if you do attempt it) its here https://gist.github.com/mpa1212/12e31f8c046335406978 
03:56 < mpa1212> \
04:00 < greatman> For a client to show a scoreboard, do I need to create a team with the player in it first?
04:01 < greatman> or just sending objective, score, display does the trick?
04:01 <+SpaceManiac> I don't think a team is needed
04:01 < greatman> hmm
04:05 < greatman> This is basicly what I do, could it be CB overriding my change?
04:05 < greatman> https://www.irccloud.com/pastebin/BaNcop0F
04:06 < greatman> Because if I resend my objective packet, the client crash with a already exist
04:14 <+SpaceManiac> If you're already using Bukkit, why are you sending packets instead of using the scoreboard API features?
04:15 < greatman> I got HealthBars using the main scoreboard so I can't create a unique scoreboard for everybody
04:15 < greatman> or im doing something wrong
05:48 < Azelk> What would a response packet look like?
05:49 <+sadimusi> response to what?
05:49 < Azelk> When pinging a server.
05:49 <+sadimusi> http://wiki.vg/Server_List_Ping#Ping_Process
05:49 < Azelk> It sends something in json but what
05:49 <+sadimusi> there's an example
--- Log closed jeu. juin 26 20:48:14 2014
--- Log opened jeu. juin 26 20:48:25 2014
20:48 -!- Irssi: #mcdevs: Total of 137 nicks [1 ops, 0 halfops, 11 voices, 125 normal]
--- Log closed ven. juin 27 10:49:03 2014
--- Log opened ven. juin 27 10:49:27 2014
10:49 -!- Irssi: #mcdevs: Total of 135 nicks [1 ops, 0 halfops, 12 voices, 122 normal]
10:55 -!- Irssi: Join to #mcdevs was synced in 344 secs
11:07 < Not-cdc9> [mcprotocol] thinkofdeath pushed 1 commit to master [+0/-0/±77] http://git.io/Wz60pw
11:07 < Not-cdc9> [mcprotocol] thinkofdeath d7370e9 - 14w26c
20:05 < Altenius> In the MC protocol, with the ping, it's supposed to only send a long.
20:05 < Altenius> But it sends 8 bytes instead of 4
20:05 < Altenius> What am I missing?
20:05 < MrARM> Isn't a long 8 bytes? 
20:06 < Altenius> Is it?
20:07 < Altenius> I thought it was 4
20:07 < Altenius> Oh, it is!
20:07 < MrARM> in Java it looks like it is
20:15 < merlish> yep long means 64-bits -> 8 bytes
20:15 < merlish> short 16 bits, int 32 bits
21:08 < Drainedsoul> this is why
21:08 < Drainedsoul> the documentation shouldn't use "short", "int", and "long" when it means integers with precise widths
21:13 < dx> while i don't disagree with you, the real problem here is people not reading this damn page http://wiki.vg/Data_types
21:33 < Not-cdc9> [Glowstone] SpaceManiac pushed 3 commits [+11/-0/±6] http://git.io/h2V8Pw
21:33 < Not-cdc9> [Glowstone] SpaceManiac 69fd08d - Changed offline-mode UUID to match Vanilla.
21:33 < Not-cdc9> [Glowstone] warriordog 4c6cd40 - Added drop information for blocks with non-standard drops (see #98).
21:33 < Not-cdc9> [Glowstone] SpaceManiac 22cd9e7 - Properly implemented experience/levelling math (fixes #117).
21:48 < dx> i kinda wish i could get a proper diff before submitting stuff to the wiki
21:48 <+SpaceManiac> "Show changes" button not do it?
21:49 < dx> ehh, well, yes, that's not really a proper diff, i don't want to screenshot it
21:50 < dx> http://wiki.vg/index.php?title=Protocol&diff=prev&oldid=5873
21:50 < dx> http://wiki.vg/Protocol
21:50 < dx> opinions?
22:00 <+SpaceManiac> looks pretty good
22:39 < benbaptist> Does anyone have any clue how Encryption Request is sent on the latest of snapshots? (14w26c and such). I haven't quite been able to figure it out myself. I might do packet dumps of real server activity to figure it out if nobody else knows.
22:41 < Altenius> I'd use wireshark
22:41 < Altenius> That's what I've been using.
22:41 < benbaptist> I suppose I could, I was just going to slightly modify my Wrapper's proxy mode to do nothing but read packets.
22:55 < benbaptist> I think I might've figured it out. The bytearrays are now prefixed with varints as opposed to shorts. 
22:55 < benbaptist> Otherwise, the packet structure is the same.
22:55 < benbaptist> *might* be. This is just what I've figured out so far.
23:13 < benbaptist> Hooray! So, during one of the snapshots, the encryption request packet (sent by the server) changed the shorts to varints. Now I need to figure out how to actually encrypt a packet.
23:13 < benbaptist> Are the lengths (prefixing the packet) unencrypted?
23:14 < shoghicp> no
23:14 < shoghicp> everything is encrypted
23:14 <+SpaceManiac> no, the whole TCP stream's encrypted
23:16 < benbaptist> Okay. To be honest, I'm not certain how much data I'm supposed to read to feed it to the decrypter.
23:17 <+sadimusi> generally you can read the length field one byte at a time
23:17 <+sadimusi> in 90% of the cases one byte should be enough
23:18 < benbaptist> Wait, the length field is decrypted? Or I have to decrypt every byte of it as I read it?
23:18 <+sadimusi> you have to decrypt everything, but I'd suggest you put the buffer after the decryption
23:19 <+sadimusi> (obviously there's also a buffer before the encryption, I just mean the one you're actively working with= 
23:20 <+sadimusi> you're writing a server, right?
23:20 < benbaptist> Okay, that's where I'm confused then. I don't know how far "everything" is. I'm using your encryption.py from this URL ( https://raw.githubusercontent.com/sadimusi/mc4p/master/mc4p/encryption.py ). But I don't know what to feed the .decrypt() and .encrypt() functions (I'm using the AES128CFB8, unless that's not correct :P )
23:20 < benbaptist> Yup!
23:20 < benbaptist> all the code I'm working on is server-side
23:22 <+sadimusi> you just have to run all the data you get through the same object's decrpyt method, the way you portion it does not matter
23:23 <+sadimusi> and make sure to use a separate instance for sending
23:24 <+sadimusi> https://github.com/sadimusi/mc4p/blob/master/mc4p/proxy.py#L332
23:24 <+sadimusi> this might help
23:24 <+sadimusi> note that mc4p is meant to run on the client side, so it's not optimized to handle many connections at once
23:25 <+sadimusi> you probably want to use gevent or twisted or something
23:26 <+sadimusi> oh, mc4p already uses gevent. I didn't even remember
23:26 <+sadimusi> but it still does a lot of unnecessary copying
23:30 < dx> just checked the source code of bravo, it does authentication without implementing encryption
23:30 <+sadimusi> o.O
23:30 < shoghicp> how o.O?
23:30 < dx> oh wait
23:30 < dx> nevermind
23:30 < shoghicp> ah
23:30 < dx> "Online authentication was broken and has been disabled."
23:30 < dx> lol.
23:31 <+sadimusi> is bravo even still alive?
23:31 < dx> kinda misleading, there are a bunch of other mentions of authentication but it doesn't work
23:31 < mathuin> I spent some time with encryption in bravo, but it was harder than it looks.
23:31 < dx> sadimusi: last protocol update was 1.4
23:31 < mathuin> It was for 1.4
23:31 < dx> sadimusi: last commit may 15, 2014, so not so bad
23:31 < shoghicp> I played with auth a few days ago for a Minecraft: PC to MCPE bridge
23:32 < shoghicp> and each crypto lib calculated the continous buffers in a different way
23:32 < benbaptist> Sorry, I just netsplitted from the channel. What did I miss while I was gone? :P
23:32 <+sadimusi> dx: the contribution graph doesn't look promising https://github.com/bravoserver/bravo/graphs/contributors
23:32 <+sadimusi> benbaptist: nothing
23:34 < benbaptist> Did my last messages go through or did the netsplit cut them off?
23:34 < benbaptist> this is a screenshot of what I last saw http://s.benbaptist.com/PpV6BA8kz8.jpg :P
23:34 < dx> we missed the last message there
23:34 < shoghicp> I missed all your messages
23:34 < benbaptist> Ah, dang.
23:34 < dx> shoghicp: what
23:35 < shoghicp> dx: yeah, on one I fixed encryption
23:35 <+sadimusi> benbaptist: http://d.pr/i/f2C1
23:35 < shoghicp> well, I'm just bundling mcrypt into the default binaries
23:35 < shoghicp> it seems to calculate them as MC wants
23:35 < dx> shoghicp: i'm watting at "I missed all your messages"
23:36 < shoghicp> dx: yeah, no idea why
23:36 <+sadimusi> shoghicp: they're just earlier than you think
23:37 < benbaptist> ah, okay.
23:37 < dx> yeah, scroll 20 minutes up
23:37 < dx> 17
23:37 < dx> there's no way irc is so broken that you don't see that
23:38 < shoghicp> oh, the line length was lot longer here and it mixed with sadimusi messages
23:57 < benbaptist> Failure for me. http://s.benbaptist.com/c2FsEd6eKv.jpg :P
--- Day changed sam. juin 28 2014
00:39 < Altenius> Is every packet encrypted?
00:40 < benbaptist> Yup. After Encryption Response is received, I run everything through the cipher.
00:41 < Altenius> aw
00:41 < Altenius> I'll have to do the same :(
00:42 < benbaptist> I don't really know if I'm parsing anything properly though. I just pack my packet normally, get the length, id, and additional payload, and then run it through the cipher. As for parsing, I read one byte at a time and decipher, which is probably not even remotely close to correct as I'm supposed to probably buffer a lot more before running it through the cipher. I just don't get how much I run through the cipher though, because
00:42 < benbaptist>  I don't know the length of hte packe.t
00:42 < benbaptist> Altenius, Are you working on server-side stuff right now?
00:42 < Altenius> I'm making a custom server with a friend.
00:42 < Altenius> With C++
00:43 < Altenius> Why can't you find the length of the packet?
00:46 < Drainedsoul> benbaptist: Why don't you just run what you have through the cipher, and then proceed with processing the plaintext
00:49 < Altenius> Is everything after the encryption key request encrypted, even the login success packet?
00:49 < benbaptist> Yes, I believe os, Altenius .
00:49 < benbaptist> so*
00:49 < Altenius> or everything after the encryption key response.
00:50 < Drainedsoul> http://wiki.vg/Protocol_Encryption maybe read the wiki?
00:50 < Altenius> Reading that now.
00:50 < benbaptist> Drainedsoul, What I mean, is I don't understand how much I'm supposed to read ahead from the socket, to send to the cipher for decryption.
00:50 < Drainedsoul> sounds like you have some tight coupling going on
00:51 < Drainedsoul> just read from the socket until you get EWOULDBLOCK or EAGAIN and then proceed
00:51 < Drainedsoul> if you decrypt and there's not a whole packet, owell
00:51 < benbaptist> As for sending packets, I believe I have that down. I just pack my length, ID, and payload as I normally would, but at last minute, I run it through the cipher.
00:51 < Drainedsoul> if you decrypt and there's more than a whole packet
00:51 < Drainedsoul> owell
00:51 < benbaptist> Okay. :P
00:51 < Drainedsoul> but your networking code shouldn't know about packet decryption
00:51 < benbaptist> so it's safe because the server doesn't send more than one packet per chunk?
00:51 < Drainedsoul> and your packet decryption code shouldn't know about your networking code
00:51 < Drainedsoul> wut
00:51 < Drainedsoul> TCP/IP is a stream
00:52 < Drainedsoul> you'll always read it in order
00:52 < benbaptist> erm
00:52 < mathuin> Yeah, that's what stream means.
00:52 < benbaptist> that's not what I meant to say. I meant, the vanilla client won't send more than one packet before an EWOULDBLOCK?
00:53 < benbaptist> if so, I guess that's what I'll do.
00:53 < Drainedsoul> why would taht matter?
00:53 < Drainedsoul> the client is the remote end
00:53 < Drainedsoul> it has no control over EWOULDBLOCK/EAGAIN/etc.
00:53 < Drainedsoul> that's between you, the OS, and the networking stack of the OS
00:53 < Drainedsoul> maybe you get 1 packet
00:53 < Drainedsoul> maybe you get 1/2
00:53 < Drainedsoul> maybe you get 14
00:53 < Drainedsoul> why does it matter
01:31 < barneygale_> From a client, how often do I need to send a Position-y packet? Once per tick, so every 50ms?
01:31 <+sadimusi> benbaptist: it doesn't matter how much you read at a time
01:31 <+sadimusi> woops, was scrolled way up
01:31 < Drainedsoul> I was about to say, I thought I just went over that :P
01:32 < benbaptist> sadimusi, Ah, okay.
01:32 < Drainedsoul> so I wasn't believable?
01:33 < benbaptist> Oh, no, I 
01:33 < benbaptist> LOL, I didn't mean it in that sense. I just mentioned sadimusi as I read his message last. :P
01:33 < Drainedsoul> ;P
12:27 < mindcat> so I have question, when appear cluster level minecraft server software?
12:27 < mindcat> or why I didn't see that now?
12:29 < mindcat> I mean...
12:30 < mindcat> why I didn't see a server could do 2000+ players in a world?
12:30 < mindcat> if we have 2000+ players
12:30 < mindcat> .
12:31 < mindcat> a server, 2000+ online players in a world without any lag
12:35 <+sadimusi> mindcat: what exactly is your question?
12:40 < mindcat> sadimusi: maybe my "question" isn't question :S
12:40 < mindcat> sadimusi: did you hear EVE?
12:41 < mindcat> EVE Online
12:42 <+sadimusi> I know eve, but you can't really compare it to minecraft
12:43 < mindcat> sadimusi: :S
12:54 < mindcat> wellll
12:55 < mindcat> oh, I still have some jobs need process
15:10 < Altenius> How does the encryption request packet work?
15:11 < Altenius> It says the server generates a keypair on startup, does that mean 3 keys?
15:18 <+Fador> private and public RSA keys derived from two prime numbers
15:35 < barneygale_> Hi all, I'm writing a command-line client. After the login sequence, I wait for a Player Position & Look and send another Player Position & Look back, then send Player packets every 50ms. I don't get kicked by the server, but I do see the server sending Player Position & Look packets to me every second or so, which makes me think it's not accepting my position updates and continually correcting them. Any hints on getting this working?
15:36 < barneygale_> The server sends on_ground=false as well, which seems weird to me
15:36 < barneygale_> I logged off from a vanilla client with my player on the ground
15:37 < barneygale_> If I don't do the 50ms Player packet loop, the server doesn't correct my position, but I do get kicked for not sending anything to the server
15:43 < Altenius> So, you're flying?
15:43 < Thinkofdeath> barneygale_: Are you sending back the position and look exactly as you got it from the server?
15:44 < barneygale_> lemme gisthub code
15:47 < barneygale_> one moment, firefox is being a derp
15:51 < barneygale_> Hm, I may have at least one other issue here. I'm getting nonsensical coordinates :/
15:51 < Altenius> hmm
15:51 < Altenius> well
15:51 < barneygale_> yeah derp I'm decoding them wrong
15:51 < Altenius> How'd you do the encryption part?
15:51 < barneygale_> Let me see if that fixed it..
15:51 < barneygale_> Altenius: a lot of trial and error :D
15:52 < barneygale_> But I used PyCrypto
15:52 < Altenius> ah
15:52 < barneygale_> was originally using M2Crypto but it's a bit horrible
15:52 < Altenius> I'm using Crypto++
15:53 < barneygale_> hah yeah that was pretty bad of me. I was reading the coordinates long longs and sending them as doubles
15:53 < Altenius> lol
15:59 < Altenius> What's the difference between a public and private key?
16:19 < barneygale> Altenius: server has private+public key. It sends the public key to the client, and the client uses it to encrypt a couple of things. It sends back the encrypted data to the server
16:20 < barneygale> The idea of public/private keys is that things encrypted with the public key can only be decrypted by the private key, and vice versa (correct me if I'm wrong)
16:20 < barneygale> The client generates a symmetric key, encrypts it with the server's public key, and sends it back
16:20 < Altenius> oh
16:20 < barneygale> the server then decrypts the encrypted symmetric key with its private key
16:21 < barneygale> now both client + server have the same symmetric key
16:21 < barneygale> so they can move on to using AES or whatever it is
16:21 < barneygale> It's the same as the SSL handshake, kinda
16:22 < barneygale> but on top of the SSL-like handshake you also have the communication with the mojang session server
16:22 < Altenius>  hmm
16:22 < Altenius> I might be lazy and use the same key for every client :P
16:22 < barneygale> Keeping the same private/public key between clients is fine
16:23 < barneygale> As long as your private key is not available then you're fine
16:23 < Altenius> oh
16:23 < Altenius> So all the clients can have the same public key?
16:23 < Altenius> and private
16:23 < barneygale> Only the server has a public/private key associated with it
16:23 < barneygale> The client does not have its own keypair
16:23 < barneygale> The client DOES receive the server's public key and then use it to encrypt a message
16:24 < barneygale> the client also generates a symmetric key, which is different to the server's keypair
16:24 < Altenius> okay
16:24 < barneygale> Go look at a youtube video on SSL :D
16:24 < Altenius> What's the point on encrypting this kind of data anyway?
16:25 < barneygale> Altenius: before this, you could run a "rogue server" and social-engineer a victim to connect
16:25 < barneygale> by doing so you get their serverId token and can connect as them to another server
16:25 < Altenius> Yeah
16:25 < Altenius> But why?
16:25 < Altenius> Who would want to do that...
16:26 < barneygale> well
16:26 < barneygale> people did do that
16:26 < barneygale> various servers got hacked iirc
16:26 < Altenius> Pointless imo
16:26 < barneygale> I don't see how you can say that lol
16:26 < Altenius> What's the point in hacking a server? What would you benefit from it?
16:26 < barneygale> It was insecure, it had an obvious security hole that had been solved decades prior
16:27 < barneygale> Altenius: that's like asking why people rob and murder
16:27 < barneygale> just because you wouldn't do it, doesn't mean no-one else would
16:27 < Altenius> I don't do anything not beneficial.
16:27 < barneygale> It's not the most gaping security problem the protocol has had
16:27 < Altenius> Couldn't they still get past this though?
16:28 < barneygale> If they can, no-one has done so afaik
16:29 < Altenius> So people used to set up fake servers that just logged the ID?
16:30 < Altenius> The server send the client the public key. The client encrypts the ID. The server decrypts the ID.
16:30 < barneygale> There's another player in this all: serverId
16:30 < barneygale> Which is different from the keypair and the symmetric key
16:31 <+sadimusi> barneygale: it's no longer used
16:31 < Altenius> The access token is all you need isn't it?
16:31 <+sadimusi> but the public key is enough
16:31 < Altenius> Then you use their access token to log into another server.
16:31 <+sadimusi> you never get the access token
16:31 < barneygale> sadimusi: oh, the server sends a blank one?
16:31 < Altenius> Oh.
16:31 <+sadimusi> barneygale: yes
16:31 < Altenius> Yes barneygale 
16:31 < barneygale> and now it's just symmetric + public key?
16:31 < Altenius> As of 1.7.2
16:31 < barneygale> Ah cool
16:31 < barneygale> That makes sense
16:32 < barneygale> It's still called "serverId" in the HTTP request though
16:32 < barneygale> That's what I was referring to, not the thing the minecraft server sends
16:33 <+sadimusi> that one's now just the hash of the public key + shared secret
16:33 < barneygale> yup
16:33 <+sadimusi> well, the client still supports server ids and some servers do still send one
16:36 < barneygale> hoping to get this project released today
16:37 < barneygale> :-)
16:37 < Altenius> So, the server has ONE public key and ONE private key. It sends the same public key to the clients.
16:37 < Altenius> ?
16:38 < barneygale> Yes
16:38 < barneygale> The thing that's unique to each connection encyption-wise is the symmetric key (also called shared secret)
16:39 < barneygale> Altenius: https://www.youtube.com/watch?v=3QnD2c4Xovk
16:45 < barneygale> actually that video might not be very helpful
16:45 < barneygale> it's been a while since I did crypto at uni
16:53 < barneygale> This one's better. It doesn't really cover what to do with the symmetric key, simply how it's exchanged to begin with: https://www.youtube.com/watch?v=wXB-V_Keiu8
--- Day changed dim. juin 29 2014
01:03 < Drainedsoul> When reading entity metadata, can Slots be empty slots? (i.e. just -1)
01:04 < shoghicp> 0xff
01:04 < shoghicp> so yeah
01:04 < shoghicp> or wait, entity metadata?
01:04 <+SpaceManiac> I would expect that a slot in an entity metadata could also be empty
01:04 < shoghicp> I don't know if it will use zero or -1 there
01:11 < Drainedsoul> thanks
09:28 < eddyb> hi
09:28 < eddyb> I'm seeing something odd with the 1.8 snapshots: level.dat doesn't contain the seed (and it's missing other usual data), but the game knows the seed
09:29 < eddyb> it does show up in level.dat_mcr, but I can move that away and it still shows the right value for /seed
10:36 <+sadimusi> eddyb: I just checked one of my saves and the RandomSeed is still in there
10:38 < eddyb> sadimusi: try creating a brand new world
10:38 < eddyb> at least that's what I did
10:38 < eddyb> (in 14w26c)
10:39 <+sadimusi> I created mine in 14w26b, maybe c introduced the issue
10:39 <+sadimusi> let me check
10:39 < eddyb> I don't really understand where the game gets the data
10:40 <+sadimusi> yes, still in there
10:40 <+sadimusi> maybe you are looking at it while the server is running
10:41 < eddyb> sadimusi: the game is closed
10:41 < eddyb> let me try gunzip it again
10:42 <+sadimusi> can you send me your level.dat?
10:42 < eddyb> RandomSeed�Bmy_�3j
10:42 < eddyb> ugh :S
10:42 < eddyb> sadimusi: it's my decoder :(
10:42 < eddyb> is it normal for a List to have TAG_END as its element tag type?
10:43 <+sadimusi> I can't really remember
10:43 <+sadimusi> I used the one I wrote about 4 years ago and it still worked fine :)
10:43 < eddyb> I looked at it in a hex editor and decided that the only meaningful interpretation is an empty List of Compounds
10:43 < eddyb> I guess I was wrong :S
10:44 < eddyb> sadimusi: this was at 5am, mind you :)
10:45 < eddyb> http://www.reddit.com/r/rust/comments/29buxw/spinning_trianglthon_today_at_12pm_pdt_come_to_sf/
10:45 < eddyb> the other dev was trying (and failing) to render a cube :P https://github.com/PistonDevelopers/hematite
10:46 < eddyb> sadimusi: something else I wanted to ask... do you know where the block ID -> block state mappings are held right now?
10:46 < eddyb> at least the "default" ones. is it Java code, some (JSON?) file in the .jar, or a part of the world data that I didn't stumble upon yet?
10:46 <+sadimusi> I haven't really looked into the snapshots all that much
10:46 < eddyb> aww
10:46 < eddyb> there is a bunch of state data in the assets folder
10:47 <+sadimusi> let me take a look
10:47 < eddyb> if I had the mappings, I could lazy mode my way through rendering most blocks :P
10:51 < eddyb> the Inventory List looks like: 09 00 09 49 6E 76 65 6E 74 6F 72 79 00 00 00 00
10:51 < eddyb> well, there's an 00 before that, sorry
10:52 < eddyb> so TAG_List("Inventory") and then 00 00 00 00
10:53 < eddyb> there should be no TAG_End in there, since the next property is in the same Player Compound (Sleeping)
10:53 <+sadimusi> I don't think item stacks use the new state data
10:54 < eddyb> a normal list has 1 byte of tag type and 4 bytes of length. this one has 4 zero bytes
10:54 <+sadimusi> can you give me that level.dat, I'd like to know if my parser fails as well
10:54 < eddyb> my decision was okay, I guess (I just forgot to skip 3 of those zeroes), but the data still makes no sense
10:54 < eddyb> sadimusi: how do you handle a list like this in your code?
10:55 <+sadimusi> https://github.com/SimpleServer/SimpleServer/blob/master/src/simpleserver/nbt/NBTList.java#L154
10:55 < eddyb> wait, no, there's 5 zero bytes, I forgot the tag is only one byte
10:55 < eddyb> sadimusi: I knew it
10:55 <+sadimusi> so you just have an empty inventory?
10:55 < eddyb> yay non-type-safety
10:56 <+sadimusi> :D that library is ugly
10:56 < eddyb> sadimusi: Java implementations don't distinguish between lists of different element types. I do
10:56 < eddyb> which makes my dynamic storage of a Byte List just as efficient as a Byte_Array
10:57 < eddyb> sadimusi: the perks of a language with ADTs :P
10:57 <+sadimusi> that would certainly be possible in java as well
10:57 <+sadimusi> but efficiency definitely wasn't the focus of my implementation 
10:57 <+sadimusi> neither was type safety :D
10:57 < eddyb> it is possible, nobody does it
10:58 < eddyb> (that's why Byte_Array exists in the first place. silly notch :()
11:02 < eddyb> sadimusi: sadly I would have to use "reflection" and transmutation to get from ByteArray to byte vectors without wasting some space
11:02 < eddyb> someone had a serialization redesign in mind that would be less "pull what you need" and more "get a token" that could alleviate this
11:03 < eddyb> sadimusi: in other news, I have an idea about eliding rendering for most chunks
11:03 <+sadimusi> how so?
11:04 < eddyb> using a worst-case ray trace
11:04 < eddyb> first pass: compute chunk sides that are not completely opaque
11:05 < eddyb> second pass: compute minimal visibility angles (in the case of  a vertical tunnel, that would be 30° from one corner to the opposite one, I believe)
11:06 <+sadimusi> the mapping from damage value to state data seems to be defined in cea.class
11:06 < eddyb> you can turn that into a simpler constraint on the player's position, relative to the chunk
11:07 < eddyb> the more flexible angle computation is limited to immediately visible chunks
11:08 <+sadimusi> sounds like that might work
11:08 <+sadimusi> the models fore each block state is indeed defined in a json
11:08 < eddyb> and the results should stay fairly constant over the course of a second of movement
11:08 <+sadimusi> but the states themselves are hardcoded
11:08 < eddyb> sadimusi: thanks, I was afraid of that. hopefully that will be alleviated before 1.8 is released
11:09 < eddyb> and propagation should stop fairly early
11:09 < eddyb> think of looking down into a tunnel that curves sideways after one chunk
11:09 <+sadimusi> it seems every possible block state is resolved to a string, the mapping to the different attributes must be happening someplace else
11:10 <+sadimusi> "Fair warning, this format is highly likely to change even more in the future!"
11:10 <+sadimusi> sounds promising
11:10 < eddyb> passing by the surface-facing chunk will toggle the visibility of the second chunk below it 
11:10 <+sadimusi> got to go
11:10 < eddyb> but the next chunk in the tunnel won't be visible
11:11 < eddyb> and you don't have to propagate at all, to it
11:11 < eddyb> or if you do, it's just one chunk and not the next, etc.
11:11 < eddyb> and the space requirements are joke, a few bytes per 16x16x16 chunk will handle most of this :D
11:33 < eddyb> sadimusi: okay, I solved that issue, and made the output cleaner: https://gist.github.com/eddyb/61867cc4eabb9f94560c
20:48 < Altenius> What's the format of the encryption request packet?
20:51 < Altenius> ^
20:53 < Altenius> :/
21:13 < Altenius> How long are UUIDs?
21:24 < Lucariatias> Altenius: 8-4-4-4-12
21:36 < Altenius> What happens after the login packets?
21:36 < Altenius> Is the client supposed to send a packet?
21:38 <+SpaceManiac> After you send the login response you switch to the PLAY protocol and send stuff like the JoinGameMessage, spawn location, chunks, and other stuff
22:04 < Altenius> Well, my server crashes the client when logging in.
--- Day changed lun. juin 30 2014
20:20 < benbaptist> dang, altenius left the server.
20:20 < benbaptist> or left the channel, I mean.
20:21 < benbaptist> If he's on the snapshot, the encryption request/response packets changed to using variable integers instead of shorts.
--- Day changed mar. juil. 01 2014
02:29 < Drainedsoul> "The class has an array of strings linked to reason codes 0, 1, 2, and 3 but only the codes for 1 and 2 are null." What's this note on "Change Game Start" (Play/Clientbound) supposed to mean?
02:34 < Drainedsoul> *State
03:39 <+SpaceManiac> Drainedsoul: it's some kind of weird note about the way the client shows messages for codes 0 and 3
05:57 < Drainedsoul> SpaceManiac: So it doesn't mean anything about how the packet is serialized/deserialized?
05:57 <+SpaceManiac> right
--- Day changed mer. juil. 02 2014
13:50 < HisShadow> Hello. I've been sniffing some packets between my minecraft client and some server and I have a question: are packets encrypted after the initial login sequence or not?
13:51 < shoghicp> yes, they are
13:51 < shoghicp> except in offline mode
13:52 < HisShadow> what exactly is offline mode?
14:00 < l4mRh4X0r> It's a server setting, it does not check with the session servers whether you are who you're saying
14:01 < HisShadow> ah
14:03 < l4mRh4X0r> And since the encryption needs Mojang's servers as well, you can't use that either when you're offline
14:05 < shoghicp> well, encryption could be done without mojang servers, but it would not be secure because of the public key check
14:12 < l4mRh4X0r> HisShadow: if you're feeling adventurous, you *could* figure out a way for the server to dump its encryption key and use that to decrypt the traffic, but it's more of a hassle than it's worth from own experience
14:12 < shoghicp> or the client
14:12 < shoghicp> the client is the one that generates the key ;D
14:13 < l4mRh4X0r> Right. I'm more of a server-side guy :P
14:13 < shoghicp> you might be able to patch it if needed, setting it to a constant value
14:14 < shoghicp> HisShadow: there is also a way to dump packets fromt the client directly
14:14 < shoghicp> but I don't remember it
14:17 < HisShadow> well, I'm trying to make a simple bot for a server with forge, I've found a nodejs(!) code that works in vanilla, but not with Forge, and I'm trying to figure out what additional packets I need to send
14:28 < l4mRh4X0r> HisShadow: the custom payload packet
14:28 < l4mRh4X0r> You need to let forge know what mods you support.
14:36 < l4mRh4X0r> To find out the format, I suggest running your own forge server with the same mods in offline mode, and then sniffing the packets
14:36 < l4mRh4X0r> Would probably be the easiest way, unless you're really into Java
15:43 < HisShadow> l4mRh4X0r: weird, I've made my own server and online-mode is set to false in properties, yet I still see some random strings instead of the decoded data
15:56 < HisShadow> strange, even though it is in offline mode, my client still sends 0xFC instead of 0xCD
15:57 < shoghicp> hmm... could it be that it uses encryption even when in offline mode?
15:58 < shoghicp> that's strange
16:16 < HisShadow> yes, I guess it uses encryption all the time, independant of authorisation, damn :(
16:17 < shoghicp> HisShadow: you might be able to run a proxy in the middle
16:18 < shoghicp> or check if packets can be dumped by the client itself
16:19 < HisShadow> shoghicp: how would the proxy help?
16:20 < shoghicp> well, it can handle auth and allow dumping packets
16:20 < shoghicp> it needs a session token :(
16:22 < Not-930a> [mcprotocol] thinkofdeath pushed 1 commit to master [+0/-0/±105] http://git.io/gQV4RQ
16:22 < Not-930a> [mcprotocol] thinkofdeath 003e81f - 14w27a
16:31 < HisShadow> so yeah, client generates some random string, then encrypts it with a public key and sends it to the server
--- Day changed ven. juil. 04 2014
20:03 < Eloston> I have two questions about the 1.7 protocol. First, do Login packets stop working after a Login Success packet? Secondly, is the Login 0x00 Disconnect packet the same format as the Play 0x40 Disconnect packet?
20:03 < Eloston> wiki.vg/Protocol does not make it very clear
20:04 < Thinkofdeath> Yes and yes
20:04 < Eloston> Thanks
20:09 < Eloston> Does the Login Serverbound 0x00 Login Start packet contain the player's in-game username or UUID?
20:10 < barneygale> Eloston: username
20:11 < Eloston> barneygale: So I would have to ask the Mojang servers for the player's UUID, correct?
20:12 < barneygale> Yeah
20:12 < Eloston> Thanks
20:12 < barneygale> You get it from the "hasJoined" HTTP req
20:12 < barneygale> I have an implementation of the login if you're interested
20:12 < barneygale> https://github.com/barneygale/quarry/blob/master/quarry/net/server.py#L70
20:12 < barneygale> ... and I just spotted a mistake >.>
20:19 < HisShadow> hmm, I've been trying continuosly to make a bot, and I've encountered a weird thing. The server is modded with forge, so I send 0x01 packet and get and send 0xFA stuff, but right after I send mod list from my side, I start to receive gibberish, I can see 0xFA packet starting somewhere in there, but the whole thing starts with 0x02 0x00 0x23 and then 0xFA. Is it valid behaviour or is there some flaw somewher?
20:34 < barneygale> HisShadow: 0xFA? Is this an old server?
20:34 < HisShadow> barneygale: oh, forgot to mention it's 1.6.4
20:35 < barneygale> right
20:39 < barneygale> So the sequence in 1.6.4 starts with 0x02 handshake
20:39 < barneygale> did you mean 0x02 rather than 0x01?
20:39 < HisShadow> barneygale: okay, let me show you what I've sniffed with smproxy
20:40 < barneygale> It also might be that as soon as you send your mod list, the server things you understand some forge-specific protocol
20:41 < barneygale> idk anything about forger
20:41 < barneygale> forge*
20:42 < HisShadow> well from what I could see in the source and in wiki.vg forge uses 0xFA packet to communicate, and forge client also sends out a 0x01 packet in the beginning
20:44 < HisShadow> http://bpaste.net/show/435153/
20:44 < HisShadow> here's the whole thing
20:52 < barneygale> Pretty strange
20:54 < barneygale> What's important is that smproxy is failing to decode the server-sent packets, and your packets seem reasonable
20:54 < barneygale> and you shouldn't be able to get into a state like that
20:54 < HisShadow> that was default minecraft client too
20:54 < barneygale> The default client? Are you injecting in the plugin messages?
20:55 < HisShadow> err, I meant to say modded client specifically for this server
20:55 < barneygale> Ah
20:55 < barneygale> Well it suggests to me there some forge-specific goodness here somewhere
20:56 < barneygale> double-check you're decoding length fields correctly and whatnot, otherwise go decompile forge
20:56 < barneygale> (is it open-source?)
20:56 < HisShadow> yeah, it is
20:57 < barneygale> Yeah
20:57 < barneygale> So I reckon after the server sends you a 0x01, you need to do something forge-specific to continue reading packets
20:58 < HisShadow> it sends you the list of the mods and you need to respond, but all that happens over packet 0xFA, I dunno why the server is sending me an 0x02
20:58 < barneygale> It's not *really* 0x02
20:59 < barneygale> forge is doing its own encryption or someshit
20:59 < barneygale> idk
21:07 < barneygale> HisShadow: this might be it: https://github.com/MinecraftForge/FML/commit/d0dd05a15c2eca9eabd308319c2ed85cb632922b
21:09 < HisShadow> barneygale: um, how so?
21:11 < barneygale> Oh, sorry I got my packet ids mixed up ;D
--- Day changed sam. juil. 05 2014
03:08 < Eloston> Hey guys, I got a little problem here. I designed a custom server so a disconnect packet will be sent and the connection closed right after a Handshake packet with state 2 (login). Sometimes the client will display the disconnect message, and other times it will print "Internal Exception: io.netty.handler.codec.DecoderException: java.io.IOException: Bad packet id 0"
03:08 < Eloston> This is 1.7.10
03:08 < Eloston> Anyone have any ideas why this could be happening?
03:23 < Drainedsoul> You're just sending a disconnect, nothing else?
03:23 < Eloston> Yes
03:24 < Eloston> and then immediately closing the connection
03:24 < Drainedsoul> when are you sending it?
03:24 < Drainedsoul> what do you mean "closing the connection"?
03:24 < Eloston> close the socket
03:24 < Drainedsoul> like close(fd);?
03:24 < Eloston> Well in Python sockets are objects and they have a close() method
03:25 < Eloston> https://docs.python.org/3.4/library/socket.html?highlight=socket#socket.socket.close
03:25 < Drainedsoul> okay you're still calling close(fd);
03:25 < Drainedsoul> yeah it doesn't matter how Python does it, at some level the close or closesocket (depending on the OS) function is getting called
03:26 < Drainedsoul> anyway, when you close a socket I'm p. sure the OS doesn't make guarantees about what has been sent/received and what hasn't
03:26 < Drainedsoul> you should call shutdown first, unless you actually want to brutally close the connection
03:26 < Eloston> I had shutdown originally, but still the same issue
03:26 < Drainedsoul> why'd you change it?
03:26 < Drainedsoul> anyway, when are you sending this packet?
03:27 < Eloston> I'm sending it right after the handshake packet
03:27 < Drainedsoul> so you're pulling the handshake packet off the wire?
03:27 < Eloston> Yeah I parse the handshake packet first
03:27 < Drainedsoul> have you Wiresharked it to see what's actually getting sent?
03:27 < Eloston> Also I'm using sendall() to send my data to the client (if that bit of information is of any use)
03:28 < Eloston> and no I haven't
03:28 < Drainedsoul> why would you use sendall
03:29 < Eloston> It said it would block until all of the data is sent
03:29 < Eloston> https://docs.python.org/3.4/library/socket.html?highlight=socket#socket.socket.sendall
03:29 < Drainedsoul> yeah, why would you want that
03:29 < Drainedsoul> I thought you were writing a server
03:30 < Eloston> Not quite, it's more like an inetd for Minecraft servers than a server
03:31 < Drainedsoul> so you're saying it listens for/accepts client connections, but isn't a server?
03:31 < Eloston> Yes
03:31 < Drainedsoul> http://en.wikipedia.org/wiki/Client%E2%80%93server_model
03:31 < Drainedsoul> you have some reading to do
03:31 < Drainedsoul> if it listens for/accepts client connections, it is by defn a server
03:31 < Eloston> Oh by server I thought you mean Minecraft server
03:32 < Eloston> Then yes it is a server
03:32 < Drainedsoul> blocking is evil, esp. in a server context
03:32 < Drainedsoul> but that's really beside the point atm, you should Wireshark and see what's actually going over the network
03:35 <+SpaceManiac> Uh, sendall is probably preferable.
03:36 < Drainedsoul> Why, so you can sit around waiting?
03:36 <+SpaceManiac> The alternative is manually making sure send() sent all the bytes
03:36 < Drainedsoul> okay, and
03:36 < Drainedsoul> what's your point
03:36 <+SpaceManiac> why manually do it if sendall() is already there
03:36 < Drainedsoul> because send doesn't block
03:36 < Drainedsoul> (at least if the socket is in non-blocking mode)
03:37 <+SpaceManiac> Then neither would sendall?
03:39 < Drainedsoul> Assuming that's true, what happens when the underlying call to send returns EAGAIN or EWOULDBLOCK.  sendall's documentation says it throws on error and doesn't notify you of how much data was sent, if any at all
03:39 < Drainedsoul> how do you know where to pick up next time
03:39 < Drainedsoul> besides, if sendall behaves like that, it literally has no advantages over send
03:42 <+SpaceManiac> Okay, so sendall would block if the whole data can't fit into the kernel buffer
03:43 <+SpaceManiac> If it's just a simple thing like it sounds like is the goal here it doesn't matter
03:43 <+SpaceManiac> If you need to do other tasks while trying to stuff your data into the buffer then you need to track stuff manually and at that point you might as well use twisted or whatever
03:44 < Drainedsoul> unless Python exposes epoll/kqueue/iocp somehow, using some kind of library is definitely the best option
04:07 < Eloston> Is it normal to see java.nio.channels.CancelledKeyException error messages?
04:07 < Eloston> It happens sometimes after I connect/disconnect to my test server after a while
04:16 <+SpaceManiac> I don't think that's normal
04:29 < Eloston> SpaceManiac: This warning message shows up: WARNING: Selector.select() returned prematurely 512 times in a row; rebuilding selector.
04:29 < Eloston> (I capture the stdout of the Minecraft process)
04:31 < Eloston> Here's the messages before the CancelledKeyException occurs:
04:31 < Eloston> io.netty.channel.nio.NioEventLoop select
04:31 < Eloston> WARNING: Selector.select() returned prematurely 512 times in a row; rebuilding selector.
04:31 < Eloston> io.netty.channel.nio.NioEventLoop rebuildSelector
04:31 < Eloston> INFO: Migrated 1 channel(s) to the new Selector.
04:31 < Eloston> [Server Connector #41/ERROR]: Couldn't connect to server
04:31 < Eloston> I'm on Debian Jessie on OpenJDK-JRE 7 64-bit if that helps
04:34 <+SpaceManiac> Does this happen on any servers other than the one you're developing?
04:34 < Eloston> Let me try
04:42 < Eloston> SpaceManiac: Yes I get it after a while
04:43 < Eloston> But I think I connected/disconnected too many times because my login session got cancelled
04:51 < Eloston> SpaceManiac: Minecraft 1.7.10 uses netty 4.0.10. I downloaded the latest netty from netty.io (4.0.21) and it seems to fix the issue
04:52 < Eloston> Although I haven't tested to see if anything else broke
04:54 < Eloston> The same messages appear, but I can join a Bukkit server just fine
06:07 <+Amaranthus> Eloston: It seems like every release of netty fixes/breaks that for different groups of people and no one can figure out wtf is going on
06:08 <+Amaranth> I think the netty folks have decided its a bug in openjdk's nio implementation on linux but I'm not sure anyone reported it
--- Day changed dim. juil. 06 2014
17:44 < eddyb> sadimusi: hi, I've been meaning to ask you: when you mentioned the .class file for the hardcoded block id -> block state mappings, did you actually get some form of decompiled Java code out of it?
21:06 <+sadimusi> _eddyb_: turns out some blocks define their states in their own class
21:06 <+sadimusi> _eddyb_: for example redstone wire
21:06 <+sadimusi> most of them seem to be aggregated in cea.class though
21:06 <+sadimusi> https://gist.github.com/sadimusi/f56d67b385a5ba5b8da6
21:06 < eddyb> https://i.imgur.com/2wYpRwR.png btw :D
21:07 < eddyb> waait
21:08 < eddyb> how come redstone has more block states than the 16 representing its power?
21:09 < MrARM> when it's on a side of a block?
21:09 < eddyb> sadimusi: thanks a lot, I guess that + aop.class would be enough to automatically create mappings for most blocks :D
21:09 < eddyb> MrARM: yeah, but... I thought only 16 states were stored for redstone
21:10 < shoghicp> it's block render state too
21:16 < Thinkofdeath> eddyb: Nice what are you coding it in?
21:16 < eddyb> Thinkofdeath: Rust
21:17 < Thinkofdeath> nice, I still haven't played around with that yet
21:17 < eddyb> the hacky code used to render that chunk can be found here https://github.com/eddyb/hematite/tree/4844cefc812dfee18f28b077aafcd7c808d806b9
21:18 < eddyb> will most likely force push over it. still wondering how to allow vanilla renders without copyright issues
21:18 < eddyb> requiring the user to unpack the latest snapshot's jar somewhere is likely fine, isn't it?
21:18 < Thinkofdeath> I have mine download the client jar and unpack it for them
21:18 < eddyb> can probably get everything I need automatically from ~/.minecraft/{versions,saves}
21:20 < eddyb> Thinkofdeath: direct AWS links are likely considered "piracy"
21:21 < Thinkofdeath> I guess :/ Although they never see it, I unpack it in memory
21:21 < eddyb> but do you ask for username/password and check it's premium?
21:22 < Thinkofdeath> Nope its just a mapviewer plugins, seemed overkill/asking for issues
21:24 < eddyb> sadimusi: is it just me, or are states supposed to be different from what I see in that file?
21:24 < eddyb> I mean, I remember seeing like 4 different "properties" at once
23:00 <+sadimusi> eddyb: yeah, I'm not sure where that mapping is happening
23:03 <+sadimusi> seems to be in the block classes
--- Day changed lun. juil. 07 2014
14:55 < akiba_> hey guys
14:55 < akiba_> does anyone understand what this code does?
14:55 < akiba_> http://pastie.org/9364082
14:59 < rom1504> missing context
15:19 < akiba_> hey rom what context do you need?
15:23 < merlish> which file it's in, other functions in the file to help infer the meaning (especially named functions!), rest of the function if any (missing close bracket on last line)
15:24 < merlish> this said, maybe it's something to do with the hotbar...? it looks for stack ids through 10 items
15:26 < merlish> that's the only 10 long list exposed to the player i can think of
15:26 < merlish> i don't know what the recursing into the same function on another object is about
15:26 < merlish> but it's looking for the first stack that has a non-zero amount that is less than the amount given as an argument
15:27 < merlish> -> one of the crazy inventory sorting things? shift-click functionality in inventory?
15:39 < akiba_> shift-click doesnt do anything
15:39 < akiba_> right click does
15:40 < akiba_> only thing i can think of
15:40 < akiba_> is the client reading packets wrong?
16:25 < Fenhl> merlish: isn't the hotbar only 9 slots?
16:26 < merlish> err it could well be
16:26 < merlish> :p
16:28 < akiba_> merlish: is the client reading packets wrong?
16:28 < merlish> i don't know what you mean :s
16:28 < akiba_> cause thats the only thing i can think of for making stackamounts null above 255 from changing writeint to writelong
16:28 < merlish> i have only seen the function, i don't know of the error
16:28 < akiba_> merlish can i show you?
16:29 < merlish> sorry, i don't think i can help
16:29 < merlish> i'm sure someone here can
16:31 < akiba_> these were all the changes i made to the code http://pastie.org/9359399
16:59 <+sadimusi> akiba_: what exactly are you trying to achieve?
18:14 < Bored> Is there anyway from the raw packets/protocol to tell if a player has hit another player without performing a calculation of distance/look/etc ? 
18:18 < akiba_> hey guys
18:18 < akiba_> can you tell me a good reason to use a long instead of an int in a video game
18:19 <+sadimusi> akiba_: one simple example would be the player position in minecraft
18:21 <+sadimusi> Bored: I think the client sends some kind of entity hit packet, the only thing you have to do is to check the distance if you want to prevent cheating
18:22 <+sadimusi> Bored: http://wiki.vg/Protocol#Use_Entity
18:23 < akiba_> anything other than location?
18:23 < akiba_> like cash stack or anything?
18:28 < Bored> Oh sorry, i was looking for a client bound way of doing this.
18:31 <+sadimusi> akiba_: anywhere you need big or precise numbers (or both). 
19:45 < TkTech> akiba_: That is an odd question. If your values won't fit inside the bounds of an int, what else would you use (while still using native types)?
19:46 <+sadimusi> oh looks like I mixed up long and double once again *blush*
19:52 < TkTech> sadimusi: Tisk tisk.
19:53 < TobiX> Well, this should make Pocket Minecraft development easier: https://github.com/OculusVR/RakNet ;)
19:54 < MrARM> well, shoghicp's RakLib
19:58 < MrARM> but for people using other languages
20:04 < TkTech> go TkTech 
20:04 < TkTech> Whoops
21:00 < Drainedsoul> TobiX: Ugh VS 2005 support, why
21:18 < TobiX> Drainedsoul: Because commercial? Big $$$ in VS support ;)
--- Day changed mar. juil. 08 2014
04:05 < akiba_> hey guys
13:48 < bored> I have a MITM proxy script using the Node-Minecraft-Protocol Lib, it has major issues with clientbound 0x0C ( Spawn Player). It seems to be the texture and signature fields. What should the signature contain ? 
13:51 < bored> Hmm larger than expected packets...
14:51 < bored> So I'm sending an empty data field for the spawn player packet which stops the client from crashing but creates a lag for when a player spawns..
14:51 < bored> and the other players don't seem to appear until and entity_teleport packet is received..
--- Day changed mer. juil. 09 2014
06:03 < jflory7> Does anyone know how to pass a command to a Minecraft console via tmux in a bash script?
06:07 <+SpaceManiac> jflory7: not really the best place to ask, but look at "tmux send-keys"
06:09 < jflory7> Oh, oops. Wasn't sure where else to ask about this, sorry. Anyways, SpaceManiac, I have done that, however, my problem is that my startup script seems to be randomly terminated once I run my new script, and THEN it sends the command, e.g. "stop" is run as a Linux command instead of a Minecraft command
06:18 < Not-930a> [fCraft] fragmer * r2340 2 files : Improved compatibility with noncompliant CW map files.
17:59 < Aaron1011> Does anyone know what the Camera packet is used for in the pre-release protocol?
18:07 < Aragas> Maybe new entity?
18:08 < Dinnerbone> Sets the camera of spectator-mode players.
18:17 < dx> dirty-room reverse engineering
18:20 < dx> whoa, the pre-release protocol page is full of Thinkofdeath edits
18:20 < dx> Thinkofdeath: you're awesome.
18:20 < Thinkofdeath> :)
20:15 < Aaron1011> Is there a separate packet for the Guardian attack?
20:22 < Thinkofdeath> Grum: Am I missing something or are you not compressing bulk chunk packets anymore?
20:35 <+Amaranth> Thinkofdeath: Time to readd the hack to only use single chunk packets again :P
20:36 < Thinkofdeath> Amaranth: Looks like those arn't compressed either :)
20:45 < Thinkofdeath> Amaranth: actually they may be compressing everything now
20:45 <+Amaranth> Just compressing the stream?
20:46 <+Amaranth> I suppose that compression happens on the netty threads too?
20:46 < Thinkofdeath> yeah by the looks of it
20:46 <+Amaranth> I hope they're compressing then encrypting and not the other way around :P
20:47 < Thinkofdeath> this.i.pipeline().addBefore("encoder", "compress", new gm(paramInt));
20:47 < Thinkofdeath> also looks optional, wonder how the client knows if its being using or not
20:51 < Thinkofdeath> ah packets are prefixed
21:02 < Dinnerbone> See www.mojang.com snapshot blog post
21:03 < dx> "Compressed space and time" - sweet!
21:04 < dx> Thinkofdeath: there's your issue, they made chunk compression optional since it's better to compress space and time instead.
21:05 < dx> thanks dinnerbro
21:09 < Dinnerbone> "Servers can now customise network compression in server.properties"
21:09 < Dinnerbone> :p
21:09 < Thinkofdeath> whoops kinda assumed that was just for chunks for some reason
21:15 < barneygale> ooh cool!
21:15 < dx> compressed json! finally!
21:19 < barneygale> wait rly
21:21 < barneygale> oh wow, all the config files are json
21:21 < barneygale> neat!
21:35 < Dinnerbone> As of 1.7.something :p
21:37 < barneygale> oh hah
21:37 < barneygale> i've been out the game :P
21:43 < Dinnerbone> I'd appreciate comments/suggestions about how the compression is handled
21:43 < Dinnerbone> (And the protocol in general, for that matter. You guys never complain anymore. :( )
21:49 < barneygale> i'm fishing around in jd-gui currently
22:52 < Thinkofdeath> Dinnerbone: Suggestion: remove gzipping from nbt tags now that large packets are gzip'd
22:53 < Dinnerbone> Thanks Thinkofdeath, forgot about nbt gzipping
22:53 < Thinkofdeath> np :)
22:54 < dx> will dinnerbro be able to remove the gzip without feeling tempted to remove nbt too?
22:55  * Dinnerbone removes dx
22:56 < dx> ;_;
22:56  * Dinnerbone readds dx with improved everything
22:57 < dx> :D
23:05 < HisShadow> Is this the real Dinnerbone ? Like, the Dinnerbone?
23:06 < dx> yes and he just implemented me as a minecraft feature. are you jealous?
23:06 < dx> (also, don't say his name, even if he's right here. use "dinnerbro" instead)
23:06 < HisShadow> okay
23:08 < HisShadow> dx: what is your function as a feature?
23:09 < dx> HisShadow: i don't know, but i'm probably better than nbt
--- Day changed jeu. juil. 10 2014
00:40 < xthexder> Hey guys, anyone know what the uuid response should look like for the Login Success packet?
00:40 < xthexder> It's supposedly supposed to contain a - now, which I'm a little confused about
00:42 < Thinkofdeath> xthexder: Like 45fb032e-8f20-3512-91d7-f973f6eee980
00:43 < xthexder> Alright, thanks
00:49 < xthexder> The code for going from the http response to that is a little ugly.... =/ Why did they even bother changing that
01:12 < Drainedsoul> So what's happening with compression now?
01:14 < yawkat> AFAIK the entire protocol is compressed now
01:14 < yawkat> not just stuff like chunks
01:15 < xthexder> Oh, I guess that would be why everything is breaking for me...
01:16 < xthexder> Maybe I should just go back to 1.7.9 then
01:23 < xthexder> Is there an in-progress page for the 1.7.10 protocol?
01:24 < xthexder> I could probably contribute a bit
01:24 < yawkat> xthexder: i think .10 protocol is the same as .9
01:25 < shoghicp> ^
01:25 < xthexder> Whats the deal with compression then?
01:25 < Thinkofdeath> compression is 14w28a snapshot
01:25 < shoghicp> that's for the snapshots
01:26 < xthexder> ah, ok nevermind then
07:33 < Grum> <Thinkofdeath> Grum: Am I missing something or are you not compressing bulk chunk packets anymore? <-- depends on your settings now
07:59 <+Amaranth> Grum: Why would you want adjustable compression settings?
09:18 <+md_5> Dinnerbone being sneaky and looking at my timeline :p
09:19 <+md_5> It's probably not that bad, although you did skew my wish a little :p
09:19 <+md_5> Apologies if I was a bit rude
09:23 < Dinnerbone> Because I thought I asked you for feedback about it and wanted to see if you replied, over 12 hours ago = lost in my feed!
09:23 < Dinnerbone> (Though it turns out I didn't explicitly ask for feedback!)
09:24 <+md_5> Anyway, thanks, I guess. We'll make it work.
09:26 < Dinnerbone> I actually do want feedback and suggestions and complaints about the protocol and efficiency, people complain about stuff but never to me so I don't get to fix their worries.
09:28 <+md_5> Dinnerbone no, just to Grum who says protocol is -7 on the priority list :p
09:29 <+md_5> Dinnerbone can you please fix the bloody race condition on sub protocol switch
09:29 < Dinnerbone> Tell me what it is and how to reproduce it!
09:29 <+md_5> Easy
09:29 <+md_5> Switch protocol from handshake to login, then send a kick packet
09:30 <+md_5> You'll get disconnected with XXX cannot be cast to YYY
09:30 <+md_5> where XXX is the Login packet handler and YYY is the handshake packet handler
09:31 <+md_5> I think in 1.7.10 its: bll cannot be cast to fv
09:32 <+md_5> lemme just chec
09:32 < Dinnerbone> Will look into it when I get to the office!
09:32  * Dinnerbone afks for an hour
09:33 <+md_5> sorry, in 1.7.9 its bll cannot be cast to fv
09:33 <+md_5> I have no idea what it is in 1.7.10
09:35 < Grum> md_5: quote me on that then please
09:35 < WizardCM> dem sneaky devs xD
09:35 <+md_5> <Grum> it really is the last priority
09:36 <+md_5> <Grum> also not remotely a priority
09:36 < Grum> and now with more contxt
09:36 <+md_5> <Grum> you guys are only talking about protocol shit
09:36 <+md_5> <Grum> i can tell you now, we will not fix that any time soon
09:36 <+md_5> <Grum> it really is the last priority
09:36 < Grum> again more context, 'that' is whaT?
09:37 < Grum> do i really have to search in my own logs? :(
09:37 <+md_5> I believe that was referring to the encoding of 3d coordinates
09:38 < Grum> which channel you getting these logs from? saves me some grepping time
09:38 <+md_5> here
09:38 <+md_5> 26th June last year
09:38 < Grum> ah
09:38 < Grum> lol mcdevs_20130626.log
09:38 <+md_5> grep priority log/freenode_#mcdev* -C 10
09:40 < Grum> and then it was indeed the last priority
09:40 < Grum> meh i cant even remember what we were doing around that time
09:40 < Grum> i think it was even before we used netty
09:40 < Grum> so yes, last on the list of 'stuff to look at'
09:40 <+md_5> making horses is what you were doing
09:40 < Grum> no, jens did that
09:40 < Grum> i just cleaned it up as much as i could without breaking them
09:41 < Grum> i haven't added a single 'fun feature' beyond sheep-breeding mixing their colors
09:41 < Grum> so feel free not to go point that sort of bullshit at me
09:42 < Grum> Also, as Dinnerbone will attest to, i did bring up this 'inconsistency' in our protocol
09:42 < Grum> it was deemed not super important because its impossible to replicate using vanilla and trivial to work around
09:42 < Grum> that said, it's a bug and it should be fixed
09:45 < Grum> and for more context to that discussion
09:45 < Grum> "[09:34:01] <Grum> SirCmpwn: we're not just doing these mini incremental changes to the protocol"
09:45 <+md_5> Dinnerbone here is one ticket: https://bugs.mojang.com/browse/MC-37203
09:45 < Grum> and this was before we started using netty
09:46 <+SirCmpwn> ?
09:46 < Grum> SirCmpwn: someone (md5) is digging up year+ old logs
09:46 <+SirCmpwn> ah.
09:46 <+SirCmpwn> I was going to say, you haven't uttered my name on freenode in months, what are you quoting
09:46 < WizardCM> would have been long before a plan to change to netty, i'd assume
09:51 <+md_5> Dinnerbone here is some more info about the race condition (12th December 2013): http://paste.md-5.net/qayehaxiqu.xml
09:51 < Grum> no we had it planned we just didn't do it yet
09:51 < Grum> because it was rather major
09:51 < mrapple> Dinnerbone: I know last_username sent you a message about the tab list a week or so ago on skype, did you ever get a chance to look at that?
09:52 < Grum> hehe you want to use it again for 'random information display'? :P
09:53 < mrapple> yes <3
09:53 < mrapple> and it's not random, it's very useful
09:53 < WizardCM> i've always found the vanilla tablist fairly dull, i like seeing servers spice it up
09:54 < WizardCM> i've never been on a server where a modified tablist showed something i didn't want
09:54 < WizardCM> actually, speaking of tablist - what's with the lack of padding on the right of skin icons? (unless that was fixed recently, only just updating my snapshot build)
09:56 < WizardCM> also spacing between name and hearts if they're enabled in the list, should be right aligned tbh
09:57 < mrapple> the tab list is super useful http://i.oc.tc/image/1b360A2O2G2h
09:58 < mrapple> we can show flair, friendship status (italics), match info, team sizes, etc etc all with one button :D
09:59 < WizardCM> see, if i go on a minigame server, and i don't see UI being well used (like that screenshot) it dulls the experience
09:59 < WizardCM> being able to access that data without a command is fantastic
10:00 < WizardCM> "61 unread appeals"
10:00 < WizardCM> phew xD
10:03 < Grum> mrapple: i kinda dislike the abuse of the list just because you have no other way of displaying information :/
10:03 < Grum> but maybe we should allow the server to set a string that gets displayed above the list orso
10:03 < WizardCM> Grum: that's what the bossbar is for ;)
10:04 < mrapple> my point is that a lot of servers use it to display helpful information since our ways to show information to the client are limited
10:04 < Grum> md_5: btw please experiment with the compression stuffs in the latest snapshot
10:04 <+md_5> We've done spigot, havent done bungee yet
10:04 < mrapple> and removing it will be sad :( just asking for some sort of way for it to be feasible in 1.8
10:04 < WizardCM> a more controllable alternative would be nice ^
10:05 < mrapple> agreed, but at the very least allowing us to still use the tab list would be very nice
10:05 <+md_5> Grum what exactly did you want me to do....
10:06 < Grum> md_5: find the right balance for the 'minimum compressed size' and 'cpu-usage' :)
10:06 < Grum> also its hilarious if you compress nothing at all (set it to -1 for example)
10:06 < Grum> mc has LOTS OF DATA :P
10:06 < Thinkofdeath> yeah tried that already :P
10:06 <+md_5> honestly I'll probably default to 1024ish
10:06  * md_5 shrug, Thinkofdeath does all the protocol work these days
10:07 < Grum> smaller might be usefuller, not sure
10:07 < Grum> mrapple: but it would be nice to just have a single textcomponent to display above the list not? I mean stuff like that would be doable-- assuming its multiline ofc
10:07 < mrapple> we already have the bossbar which is great
10:08 < mrapple> I mean any client display you're willing to add for us is awesome
10:08 < mrapple> I think the tab list is perfect because it allows for tables
10:08 < WizardCM> hiding the bar portion would also be nice, if we're throwing ideas out there
10:08 < mrapple> the rows and columns of the tab list make it excellent for display information
10:08 < mrapple> IMHO it's cleaner and nicer than just throwing a text component up there
10:08 < mrapple> also, the user can easily toggle it
10:09 < Grum> my issue is you are trying to ask for the wrong thing
10:09 < Grum> you are asking "for abuse of the tablist" where you actually just want a way to "display extra information"
10:09 < Grum> I'd prefer adding a way to display extra information over abusing the tablist any day
10:10 < mrapple> abuse makes it sound bad :(
10:10 < mrapple> we're doing good things with it
10:10 < Grum> no, it is bad, why? because it breaks when we change something
10:10 < Grum> and then you guys are sad
10:11 < Grum> gotto run, alaedy late >.>
10:11 < WizardCM> stuff breaks all the time
10:11 < mrapple> there are a lot of things that break all the time
10:11 < WizardCM> most of the time there is an alternative fallback
10:11 < mrapple> that's what we're asking for
10:15 < WizardCM> visually though, i do like the improvementds in recent snapshots to the player list. it's just that they'll only affect smaller servers
10:15 < WizardCM> improvements*
10:15 < mrapple> indeed
10:16 < mrapple> the tab list has definitely seen some good improvements for vanilla
10:16 < mrapple> we just need some fallback support for display info :)
10:16 < WizardCM> mhm
10:16 < WizardCM> having a separate UI would also be a good thing
10:16 < mrapple> meh
10:16 < WizardCM> the ping bars can get in the way sometimes xD
10:16 < mrapple> an option to remove the ping bars would be good
10:16 < mrapple> having the tab list at all on a large server is pretty meaningless
10:16 < mrapple> we add so much more useful information
10:16 < WizardCM> mhm
10:18 < WizardCM> then again, you also have to realise
10:18 < WizardCM> a lot of people don't know of the tablist
10:19 < mrapple> really? are you sure about that?
10:19 < WizardCM> yes
10:19 < WizardCM> i've spoken to players and mentioned it
10:19 < WizardCM> and they've gone "oh wow, cool!"
10:19 < mrapple> hm I may have to run a poll
10:20 < WizardCM> whenever i try out a minigame server i've never been on before i always open the tablist to see what interesting thing they're doing with it
10:20 < mrapple> i feel like it's well known, probably just because I use it all the time
10:20 < WizardCM> another strange thing i learned: when players join a server they've never been on before they don't always look at their starting items
10:21 < WizardCM> even if it's a single book in their first slot
10:23 < WizardCM> which is the opposite of what i'd do
10:23 < WizardCM> starting items? especially a book? i'm sure it's important or interesting
10:23 < Grum> Things tend to break more when you depend on things to be used in non intended ways 
10:24 < WizardCM> this is true
10:24 < WizardCM> but when there is no alternative, you use what exists to its full potential
10:25 < WizardCM> you push the boundaries to make your server more unique and interesting
10:25 < Grum> So rather than breaking our system to allow random flexibility maybe adding ways to allow you to achieve the real goal is better
10:26 < WizardCM> absolutely
10:26 < Yogu> Did I understand it correctly that eventually, all the different blockstates will have IDs? And that these IDs will be auto-generated, not hard-coded in the code? Will the server then somehow tell the client which ids are which blockstate?
10:27 < Grum> So adding a 'display name' to the tablist would allow custom names/icons etc
10:27 < WizardCM> :O could you imagine server logos in the tablist
10:27 < WizardCM> that'd look pretty cool
10:27 < WizardCM> if done correctly, of course
10:27 < Grum> No because you have no control over how it sorts
10:28 < Grum> And you don't need to either
10:28 < WizardCM> for a basic player list, of course you don't need it
10:28 < Grum> It sorts by teams
10:28 < Grum> Then name
10:29 < Grum> Iirc
10:29 < Dinnerbone> Teams then UUIDs I think
10:30 < WizardCM> uuid? that's a strange thing to sort by
10:30 < WizardCM> i would think by score could be more interesting
10:30 < WizardCM> and if there isn't one, fall back to alphabetical
10:30 < Grum> If there are scores I think it does
10:30 < Dinnerbone> Yeah
10:30 < WizardCM> ah yeah
10:31 < WizardCM> any particular reason for sorting by uuid?
10:31 < Grum> Display name is a nice idea for some flair-like things 
10:31 < WizardCM> or rank
10:31 < Grum> Yeah
10:31 < WizardCM> a lot of servers are now using scoreboard prefixes and being able to put admins at the top natively would be pretty cool
10:31 < Grum> Or who has the flag
10:31 < Grum> Or similat
10:31 < WizardCM> mhm
10:31 < Dinnerbone> Sorry my mistake, it's team -> score -> name
10:32 < WizardCM> name is good, uuid would be very strange Dinnerbone :P
10:32 < Grum> Almost  at hornstull 
10:32 < Grum> Pfft some of these state bugs are horrid to fix and we have tens of them