19:03 < astradev> Alright.
19:03 < TheUnnamedDude> Holli sheet
19:03 < TheUnnamedDude> That renders wrongly on ubuntu
19:03 < TfT_02> Grum, you bored? :P
19:03 < Grum> ⃣o.o ⃣
19:03 < Grum> i think it is kerned far left hehe
19:03 < TheUnnamedDude> http://screencloud.net/v/zuTd
19:04 < astradev> TfT_02: That's an understatement.
19:04 < Grum> ⃝ ⃣
19:04 < Grum> :P
19:04 < Grum> ⃣ ⃣ ⃣ ⃣ ⃣ ⃣ ⃣ ⃣ ⃣ ⃣ ⃣ ⃣ ⃣ ⃣ ⃣ ⃣ ⃣ ⃣ ⃣ ⃣ ⃣ ⃣ ⃣ ⃣ ⃣ ⃣ ⃣ ⃣ ⃣ ⃣ ⃣
19:04 < Grum> aiaiaiai
19:04 < TheUnnamedDude> Inb4 ascii art in the vanilla server
19:04 < Grum> ⃝ ⃣ ⃝ ⃣ ⃝ ⃣ ⃝ ⃣ ⃝ ⃣ ⃝ ⃣ ⃝ ⃣ ⃝ ⃣ ⃝ ⃣ ⃝ ⃣ ⃝ ⃣ ⃝ ⃣ ⃝ ⃣ ⃝ ⃣ ⃝ ⃣ ⃝ ⃣ ⃝ ⃣
19:04 < Grum> uhoh :P
19:04 < Grum> . ⃝. ⃣. ⃝. ⃣. ⃝. ⃣. ⃝. ⃣. ⃝. ⃣. ⃝. ⃣
19:05 < Grum> http://i.gyazo.com/a30a3ec8ee082f836095d5406ac9743e.png O.o
19:05 < TheUnnamedDude> Nooo
19:05 < TheUnnamedDude> Not gyazo
19:05 < Grum> give me another sane tool that works sane on osx :)
19:05 < Grum> sane please! ;D
19:05 < TheUnnamedDude> screencloud
19:05 < TfT_02> Heh, it looks entirely different for me. Lemme make a screenie
19:06 < TheUnnamedDude> Well, I do not know how it works on OS X
19:06 < humerusj> Someone's having some fun.
19:06 < TheUnnamedDude> Make a script with screencaputure and bind it to a key? :D
19:07 < TfT_02> http://imgur.com/1CLBgd7.png
19:07 < TheUnnamedDude> s/caput/capt/
19:08 < astradev> https://st0rage.org/~astra/public.php?service=files&t=4ce2e51aec1518c6510c910b56daf1c0
19:08 < astradev> that's what it looks like here.
19:08 < shoghicp> I see a bunch of null characters here :P
19:08 < Grum> lol that doesnt work at all ...
19:08 < Grum> screencloud that is ;(
19:08 < TheUnnamedDude> Whaa
19:09 < astradev> androirc renders it as yellow squares. https://st0rage.org/~astra/public.php?service=files&t=4ce2e51aec1518c6510c910b56daf1c0
19:09 < Grum> it claims: shift-alt-2 ...
19:09 < Grum> does nothing
19:09 < Grum> for a 'region'
19:10 < Grum> its not doing anything! megarage
19:10 < astradev> ⃣ ⃣ ⃣ ⃣
19:10 < astradev> hm
19:10 < TfT_02> Grum if you have Dropbox for OSX, you can configure it so that it will save all the screenshots online. So you can use the native commands like CMD + Shift + 4 to take a regional screenshot
19:11 < astradev> I just use Owncloud. links all my storage services
19:11 < Grum> its really wtf broken
19:11 < Grum> oh it did something
19:12 < Grum> and said nothing
19:12 < Grum> and hid the window
19:12 < Grum> uninstalling
19:13 < Grum> that was surprisingly not working at all hee
19:13 < TheUnnamedDude> .-.
19:13 < TheUnnamedDude> Works well on windows
19:13 < TheUnnamedDude> Works ok on Ubuntu
19:13 < Grum> gyazo works nice .. but i hate the spam
19:16 <+Amaranth> Grum: So, just to make sure no one is missing anything, there is no way to make an NPC anymore unless you own the account for that UUID and have it log in once every 24 hours, right?
19:16 < Grum> and yeah i  know i can probably have the default screenshot stuff store it on dropbox; i want it pastable in my buffers =)
19:16 < Grum> Amaranth: i think so yeah
19:16 < TheUnnamedDude> Hate the fact that when you press the image it starts a download for gyazo
19:17 < Grum> wel you can have an unskinned npc =)
19:17 < TheUnnamedDude> Also the stupid long urls
19:19 <+Amaranth> Grum: But it makes the client disconnect
19:21 < Grum> you can ... prevent that but wont have a skin
19:21 < Grum> and that wont work in the future
19:22 <+Amaranth> I mean, the real solution here is actual NPCs as in custom entities
19:22 <+Amaranth> But it's going to suck between now and when that becomes a thing you can do
19:24 < Grum> abuse villagers? :D
19:25 < dividuum> About this disconnect thing. When does the client disconnect?
19:25 < Thinkofdeath> When it gets the Spawn Player packet
19:25 < TfT_02> Grum: It’s pastable with dropbox (?) Not sure I understand what you mean, but when it uploades the screenshot, the URL is on your clipboard. It does have ugly long urls though..
19:25 <+Amaranth> When you send it a spawn packet that doesn't have skin blob data (or has invalid data?)
19:26 < dividuum> And invalid is everything from expired to non-matching uuid?
19:27 <+Amaranth> I don't know about the invalid part, that's a guess
19:27 <+Amaranth> If you send 0 it definitely dies
19:27 < dividuum> Grum?
19:28 < Grum> pfft deobfuscate =D
19:28 < dividuum> i'd prefer a statement from the source :)
19:30 < Grum> https://gist.github.com/grum/bae8306e6a704be9d52a
19:30 < Grum> there u go
19:31 < Thinkofdeath> :P
19:31 < Thinkofdeath> http://hastebin.com/bamisejuju.java
19:31 < Thinkofdeath> and http://hastebin.com/hifoxuneko.java
19:31 <+Amaranth> hahaha that was great
19:32 < dividuum> nice statement :-}
19:32 < TheUnnamedDude> Wow
19:33 < TheUnnamedDude> Now I am going to hake minecraft with this code!!!
19:33 < Grum> you missed the important if statement before the first pastie Thinkofdeath
19:34 < Thinkofdeath> http://hastebin.com/duraduhasi.java
19:34 < Grum> and that is how you can have fake players on the server for now
19:34 < Grum> they will be steves
19:34 < Thinkofdeath> A non-version 4 uuid?
19:36 < Deaygo> Grum: the server can still hold hostage a users skin.  Can't it just do stupid stuff to make them render as Steve until they pay too.  Like change the name of the person when it gets sent back to the persons client?
19:36 <+Amaranth> Which is why he said that is going away
19:36 < Grum> right now? yes, later, no
19:36 < Deaygo> I didn't see the part of it going away, sorry.
19:37 <+Amaranth> Grum: So are you going to detect the server is in offline mode before allowing that? Because that could be faked too
19:38 < Grum> you cannot fake being in online mode
19:38 < Grum> if connection == crypted; you are in online mode
19:38 < Grum> no other way to get there
19:38 <+Amaranth> Oh right I remember how that handshake works now
19:39 <+Amaranth> The same protection against MITM of the handshake blocks pretending the server is offline mode so you can use steve skins
19:39 < Grum> yup :p
19:39 < Grum> right now you can forge a game-profile with a non-version-4-uuid for a steve
19:39 < Grum> 1.8 has code to simply mark the connection as crypted and flatout deny on that
19:40 <+Amaranth> So don't even bother
19:40 <+Amaranth> I wouldn't spend time updating NPCs for that if they're going to break in a month anyway
19:41 < dividuum> so basically I can shut down miners-movies :(
19:41 < dividuum> because all recordings become unplayable after 24h
19:41 <+Amaranth> Although considering the code isn't even stable enough for a snapshot perhaps a month is optimistic :D
19:41 < Grum> haha
19:41 < Grum> i can show you why Amaranth :p
19:41 <+Amaranth> dividuum: You'd have to run in offline mode and just show steves
19:41 < Thinkofdeath> dividuum: You could make a web based viewer
19:41 <+Amaranth> Grum: Because BlockState rewrite?
19:42 < Grum> $ git log -p snapshot-14w11b... | wc -l
19:42 < Grum>   141848
19:42 < Grum> not really :P
19:42 < dividuum> Thinkofdeath, true. But urgs :}
19:42 < Grum> dividuum: it'll work for old stuff and offline:(
19:42 < Thinkofdeath> dividuum: Its not /too/ hard :P
19:43 < dividuum> sigh
19:43 < Grum> we're slightly in a state of 'flux' atm =)
19:43 < dividuum> I was really pumped for the new camera thing
19:43  * Amaranth tries to remember what -p is
19:43 <+Amaranth> Oh right, to show the actual diff
19:44 < Grum> 141k lines of diff ....little bit much
19:44 < Grum> and not remotely done =)
19:44 < Thinkofdeath> Sounds like updating bukkit to 1.8 will be fun 
19:44 < Grum> much of the changes are clientside only
19:45 < Grum> but yes, we rip lots and LOTS of code out of all the block classes too
19:45 < Grum> and items too soon
19:45 < Grum> after i make another set of 'controlling objects' >.>
19:45 < Thinkofdeath> Btw how does the new block state system work internally?
19:45 < Thinkofdeath> e.g. for saving 
19:45 < Grum> pfft Dinnerbone can explain :p
19:45 < Grum> on
19:45 < Grum> oh
19:45 < Grum> still data
19:45 < Grum> we're not changing that
19:45 < Grum> we cant ... yet
19:46 < Grum> that requires a world format change ... aaaaand yeah no.
19:46 < Thinkofdeath> Maybe I changed my mapviewer away from block ids too soon then
19:46 < Thinkofdeath> Uses a string -> id map per a chunk
19:47 < Grum> we have some issues to solve with data-storage
19:47 < Grum> yeah, like that, is 'too high resolution' :P
19:47 < Thinkofdeath> Isn't too bad http://thinkofdeath.co.uk/mapviewer/ 
19:47 < Thinkofdeath> Comes with a small perf hit 
19:48 < dividuum> Thinkofdeath, hey. maybe web base playback isn't so bad after all :-)
19:48 < Grum> its not a small one
19:48 < Thinkofdeath> dividuum: :)
19:48 < Grum> that would create too much memory overhead
19:48 < Grum> keeping a map per chunk around
19:49 < Thinkofdeath> I guess
19:49 < Grum> but hell that might be good enough :P
19:49 < Grum> we dont know yet
19:49 < Grum> pro and cons to all the things
19:49 < Thinkofdeath> It isn't too bad, my mapviewer has the world loaded 5 times and the memory is still below ~200mb overall
19:50 < Thinkofdeath> (For the whole mapviewer not just the world)
19:50 < Grum> yeah but that is nothing compared to the area a server has to keep loaded hosting 100 people
19:50 < Thinkofdeath> Oh for server, true forgot about that
19:50 < Grum> per chunk would be nice, better would be per 16^3 ...
19:50 < Grum> but might be too expensive memory wise
19:51 < SinZ> why not have it per region
19:51 < Grum> per region might be nice ad those are the 'files on disk'
19:51 < Grum> but then you have to somehow manage to handle someone copy/pasting them into another world :P
19:51 < Grum> all doable
19:51 < Grum> not fun
19:51 < Grum> per world is going to be the minimum
19:51 < Grum> but you kinda dont just want to store it once
19:52 < SinZ> but you also kinda dont want repeated data everywhere
19:52 < Grum> we might do an inbetween; mega-map @ world-level which does index->string (which gets backed to index->block) and then from each chunk we'll have a 'pallet' of those indexes orso
19:52 < Grum> *baked
19:53 < SinZ> wc3 does the pallet system
19:53 <+Amaranth> Ooh ooh pallet, I did one of those
19:53 < Grum> pallet is awesome, most generated chunks can store blocks with 4bits/block
19:53 < Grum> and i mean 16^3 ofc ;)
19:54 < Grum> so that would just be 2k of uncompressed data
19:54 < SinZ> and chunks that are 100% smooth stone would actually be smaller size
19:54 < Grum> well, ideally we make 'different kinds of chunks'
19:54 <+Amaranth> https://gist.github.com/amaranth/10406779
19:55 <+Amaranth> I spent some time in here trying to get people to think up something better than that
19:55 < Grum> i have many kinda of formats we need to be able to try
19:55 < Grum> ideally i make a system that can mark chunks being of a certain kind
19:56 < Grum> and then we can start experimenting with what works best :P
19:56 <+Amaranth> Oh wait that's the wrong class
19:56 < Grum> and if we just have readers for all 'known formats' forever -- (should be small classes wrapping the data) -- then we can always upconvert to something else
19:57 <+Amaranth> There now it has both
19:57 < Grum> and 3d biomes ... stackable regions ...
19:57 < Grum> many things i need to add :p
19:58 < Grum> we have some nice ways to stop sending large amounts of data too
19:58 < Grum> but we have to see if its too 'heavy' for the server
19:58 <+Amaranth> You remember talking with Zeerix about doing something like this back in early 2012? That was my inspiration
20:00 <+Amaranth> The only representation I've found that would be better would be AABBs which is basically 2D RLE but has the advantage of saving you time on your physics engine
20:00 < Grum> yeah
20:00 < Grum> wont work so well because of custom-shapes
20:00 < Grum> we have no way of handling that yet
20:00 < Grum> going to be SO FRIKKINPAINFUL
20:01 <+Amaranth> Oh yeah, my thoughts on that was just if the block isn't using the default block model treat it like a tile entity
20:02 <+Amaranth> But this is for the server, not the client
20:02 < SinZ> but then there is silly blocks like the anvil which isn't a tile entity
20:02 <+Amaranth> I mean, it'd be nice to do something like this on the client too because a mesh falls out of an AABB with not much work
20:03 < Grum> right now we have definitions of the meshes @ client
20:03 < Grum> but only visual ones
20:03 <+Amaranth> Although this kind of setup is optimized for a world that has writes but isn't dominated by them
20:03 <+Amaranth> While minecraft writes like crazy
20:04 < Grum> yeah
20:04 <+Amaranth> The nice thing about the palette setup is it is still O(1) read (but with an extra level of pointer chasing) and O(log n) write
20:07 < Grum> i want to get to a point where we can experiment with different kinds of storage
20:07 < Grum> but not touching that before 1.9
20:08 < Grum> the amount of work that was needed to stop having Blocks decide their textures in code ...
20:08 < Grum> was .... incrediwtf
20:08 <+Amaranth> Grum: Now it's time for entities! :D
20:08 < Grum> no, items first
20:08 <+Amaranth> Items are just a texture?
20:09 < Grum> or a block...model
20:10 < Grum> also, the server knows the texture ... why? :)
20:10 < Grum> remember 'Icon' in Block? all gone.
20:11 <+Amaranth> Those kinds of things were useful for figuring out what classes were what sometimes
20:11 < Grum> Block.REGISTRY.get($BlockReference) gives you a "minecraft:stone" :P
20:14 < Thinkofdeath> I wish more of the packet classes had toString methods, would make working out the field names easier :)
20:27 < Grum> toostring is overrated
20:34 <+Amaranth> Some classes have methods that look a lot like a toString() but apparently weren't called toString() since that doesn't get obfuscated
23:07 < Thinkofdeath> Grum: If you want you can mark this one as my fault http://hopper.minecraft.net/crashes/minecraft/MCX-1506121/
23:10 < dx> Thinkofdeath: why would you send the same crash report 24351 times?!?!
23:10 < Thinkofdeath> dx: Spigot bug, so not me sending it :)
23:11 < Thinkofdeath> (I tried once for the hell of it, it wouldn't send :( )
23:11 < dx> Thinkofdeath: so, what happened there?
23:12 < Thinkofdeath> Player skulls need the correct case username
23:12 < Thinkofdeath> ^ That happens if its incorrect
23:12 < dx> oh
23:12 < Thinkofdeath> And in 1.7.5 it was valid to incorrectly cased skull owner
23:12 < dx> do the crash reports send the exact spigot version string?
23:13 < Thinkofdeath> I think they just send "Craftbukkit" currently
23:13 < Thinkofdeath> Should probably change that
23:14 < dx> i bet mbaxter would appreciate that
23:14 < Thinkofdeath> Yeah I know he would :P
23:18 <+Amaranth> :'(
--- Day changed ven. avril 11 2014
02:19 < astradev> Could someone help me with debugging network packets in the MC server? I started the 1.6.4 server up in offline mode, started it with:
02:19 < astradev> "java -Dlog4j.configurationFile=/home/box/Desktop/dbmc/config.xml -jar server164.jar nogui"
02:20 < astradev> And the contents of said config.xml are exactly as what's on wiki.vg
02:21 < NickG365> IIRC, that won't work prior to 1.7, since that's when log4j (and that parameter for a config file) was introduced.
02:22 < dx> astradev: what NickG365 said, use a proxy like https://github.com/sadimusi/mc4p or https://github.com/SirCmpwn/SMProxy
02:23 < astradev> Ooohhh. Well now I feel quite silly.
02:24 < astradev> Thanks.
02:30 <+SirCmpwn> if you plan on using SMProxy
02:30 <+SirCmpwn> you might want to update it first :P
02:31 < astradev> I just decided to stick to 1.7 for now, haha. 
02:31 < astradev> when I get around to 1.6.4, I'll have to, though.
03:01 < dx> SirCmpwn: not if it's going to be used with 1.6.4
03:02 <+SirCmpwn> oh, alright
03:02 <+SirCmpwn> yeah
03:08 < astradev> Do the logs not give the information for packets like Use Entity?
03:10 < astradev> [20:57:59] [Netty Client IO #1/DEBUG]: OUT: [PLAY:2] ja[]
03:10 < astradev> This is supposedly the "Entity Use" packet, but it doesn't return any data I can see.
03:35 <+Amaranth> Grum, Dinnerbone: How exactly do you expect to reduce the load on your web services when every call to /op, /deop, /whitelist, /ban, and (maybe) /pardon seems to require a network request to fully populate a GameProfile object just to use as an appropriate key in a HashMap?
03:36 <+Amaranth> Because those things are stored in HashMaps where the key is GameProfile.toString() and that toString() method includes the name, uuid, and properties
06:08 < mbaxter> Thinkofdeath: Much appreciated. I mean, it's not like non-CB mods claiming to be CB caused a minor smear campaign against me or anything...
06:57 < Grum> Amaranth: different webservice, that one is not taking any load at all
06:58 < Grum> Amaranth: of all the things that could have exploded, that is one that isn't atm :P
06:59 <+Amaranth> Grum: It's sad though, I have a UUID and I have to do a web request to get a worthless name and skin blob
07:00 < Grum> yeah that frankly makes no sense
07:00 <+Amaranth> Or write my own replacement for the reader/writer
07:00 < Grum> i dont see how any of the other information matters for those purposes
07:00 <+Amaranth> Perhaps it shouldn't use toString() :P
07:01 < Grum> please don't make me sad again before the workday starts
07:01 <+Amaranth> haha
07:01 < Grum> have had enough rage this week
07:01 < Grum> someone decided to do skull-lookups on the main thread
07:01 <+Amaranth> :D
07:01 < Grum> A R E Y O U F U C K I N G K I D D I N G M E
07:02 <+Amaranth> But it's just that one time, right?
07:02 < Grum> yes, for every skull.....
07:02 <+Amaranth> After that it stores the result in the skull forever?
07:02 < Grum> so load a chunk with 10000 skulls and everyone disconnects
07:03 < Grum> the original idea was: have it async with callback; send the data to the client as a 'steve skin' (so no skin) -- then when you have the data, simply update the client and bake it into the level
07:03 < Grum> but noooôôöòóœ
07:03 <+Amaranth> This is bad but it's just a one off
07:03 < Grum> yeah .... lets just say that i gave up on that note
07:03 < Grum> and its not a one-off
07:03 < Grum> its a 'foreverhappeningthisway'
07:03 <+Amaranth> I've got an OfflinePlayer interface that used to be based on names that I have to convert to UUID except I can't do anything with UUIDs either for some weird reason
07:04 <+Amaranth> :D
07:04 <+Amaranth> I already accepted skulls and moved on to something more tragic :D
07:04 < Grum> things would have been easier if someone hadn't decided that names were a good unique trait
07:05 < Grum> pff please relay all the tragic
07:05 < Grum> i'll go scream at some people today =)
07:07 <+Amaranth> Well that and the kick after 23h59m is all I've found so far
07:08 <+Amaranth> But I stopped there and decided to let someone else deal with it while I complained to everyone that could understand the explanation :P
07:09 < Grum> yeah ... meh
07:09 < Grum> in theory the 'kick' is not needed in the end
07:10 < Grum> right now, people would blame 1.7.7 as broken and not use it
07:11 < Grum> and obviously people werent listening when i said to remove 1.7.6 >.>
07:12 < Grum> we'll see
07:12 < Grum> its a bit annoying right now
07:12 < Grum> if we're showing tons of steves we might need to somehow make the client aware of the 'correct time'
07:17 < Grum> Amaranth: where is it doing this tostring?
07:17 <+Amaranth> Was going to say the expiry could be ticks since for some reason the server tells the client that but that ruins the signing part
07:17 <+Amaranth> Uh, we call it JsonList
07:18 <+Amaranth> It's got a HashMap<String, JsonListEntry> and to get the key it does Object.toString() using a method that grabs the GameProfile from the entry object
07:18 <+Amaranth> Let me get the obfuscated class name
07:18 < Grum> meh impossible to find
07:19 <+Amaranth> Well, you could just see what the parent class of the thing in PlayerList holding the whitelist is
07:19 <+Amaranth> Looks like the obfuscated class name in 1.7.7 is ol
07:20 < Grum> ah found i think
07:20 <+Amaranth> "Could not save the list after removing a user."
07:20 <+Amaranth> A string in the file
07:21 < Grum>     protected String getKeyForUser(K user) {
07:21 < Grum>         return user.toString();
07:21 < Grum>     }
07:21 < Grum> w...h...y
07:21 < Grum> so confused;
07:21 < Grum> public class StoredUserList<K, V extends StoredUserEntry<K>> {
07:21 < Grum> and then internally;     private final Map<String, V> map = Maps.newHashMap();
07:23 <+Amaranth> Yeah, we don't have the generics so it looked like someone was trying to make a list of things with an efficient index that wasn't the object the list is storing
07:23 <+Amaranth> And I can't explain that in a way that doesn't sound like a map
07:23 <+Amaranth> I blame mbaxter
07:23 < Grum> multiple things seem to be tostringged in there
07:24 <+Amaranth> Anyway, I think my solution is going to be to make GameProfile's toString only include the UUID and pray nothing falls over :P
07:24 < Grum> just the stringified uuid through the banlist and opslist ... and whitelist
07:24 < Grum> so why not make it uuid? :/
07:26 <+Amaranth> This skin stuff is a tough problem
07:26 <+Amaranth> Now that you've pointed out the possibility of holding skins hostage if you make it possible in any way people are going to do it
07:26 < Grum> those two are not connected
07:27 <+Amaranth> Oh I was just reading bug mail
07:27 < Grum> i mean, people would do this if we wouldn't do these changes
07:27 <+Amaranth> If you get rid of the expiry all people need to do is have someone login once and they can impersonate them forever
07:28 <+Amaranth> If you make it just fall back to a steve skin a server could use a stale skin blob on purpose until the user pays to get their skin on that server
07:28 <+Amaranth> I don't know that any of them would have done that before but after you pointed out it was possible people have the idea now :P
07:28 < Grum> which is why we're never going to remove expiry time
07:29 <+Amaranth> I could totally see them doing "$10 to look like dinnerbone" or something though
07:29 < Grum> which wont work; name wont match etc
07:29 <+Amaranth> Right, I'm saying if you didn't lock it down
07:29 < Grum> which is why we locked it down in the first place
07:29 < Grum> there is 'risk' in having servers distribute this data
07:30 <+Amaranth> I wish Riking would use the preview button and stop spamming edits to his comment on that bug
07:31 < Grum> those mails are going straight to devnull>.>
07:32 <+Amaranth> You could just not spawn the player but then someone would abuse that to be invisible in PVP or something
07:33 <+Amaranth> I think you're going to have to just fall back to Steve and not disconnect. And just mention brand guidelines to anyone who screws with it :D
07:37 <+AndrewPH> hey hi, important question here, how much deditated wam do i need for my minecraft suvruwr
07:37 <+SirCmpwn> 3
07:38 < Grum> if i could ban you i would have
07:38 < Grum> sad moment
07:38 <+AndrewPH> okay great thnksdf
07:38 <+Amaranth> Can I? No, I don't think so
07:38 <+Amaranth> -ChanServ- You are not authorized to perform this operation. :D
07:39 <+AndrewPH> so hi hey how's it going guys
07:39 < Grum> quite sure that kid has bigger balls than 75% of the internet
07:39 <+AndrewPH> probably
07:39 <+AndrewPH> he had the courage to ask so yeah
07:40 <+AndrewPH> my girlfriend slaps me when  I make fun of people in those awkward minecon videos
07:40 <+SirCmpwn> I feel like I missed something
07:40 <+Amaranth> Random topic change
07:40 <+AndrewPH> SirCmpwn: my question was asked by a guy at minecon
07:40 <+SirCmpwn> asked of whom?
07:40 < SinZ> Not as good as the autism question
07:40 <+AndrewPH> I don't know any minecraft or bukkit or anything devs by face
07:40 <+AndrewPH> SinZ: THAT kid has balls.
07:40 <+AndrewPH> he's going places
07:40 <+AndrewPH> maybe to timeout
07:41 <+Amaranth> We went using Object.toString() as a key to skull profile data getting loaded from the interwebs on the main thread to the client blaming Spigot for every mod/plugin that screws up skins in 1.7.7 :D
07:41 <+Amaranth> Oh, and then to dedicated ram
07:41 <+AndrewPH> deditated wam*
07:41 <+SirCmpwn> I feel like the mojang/bukkit guys are qualified to answer that question
07:42 < SinZ> without info like player count, its hard to say
07:42 < SinZ> and how much they want to explore in different directions and have 500 chickens in the space of one block
07:42 <+Amaranth> Let's just say 100MB/player and call it good
07:42 <+SirCmpwn> I would have made an educated guess that he hosts 1-30 players
07:43 <+SirCmpwn> wait, 100MB/player
07:43 <+SirCmpwn> jesus, is vanilla/bukkit still that bad?
07:43 <+AndrewPH> I'd say 200MB + (50MB*players)
07:44 <+SirCmpwn> I'm surprised MB is even involved
07:44 <+Amaranth> With zero overhead just the raw block data for a single player at the default view distance can be up to 83MB
07:44 <+SirCmpwn> C.N needs a handful of kilobytes per player on a bad day
07:44 <+Amaranth> Because you don't actually have any chunks
07:44 <+SirCmpwn> ?
07:44 <+AndrewPH> SirCmpwn: this is accounting for all the blocks, not just the player
07:44 < Eviltechie> guise, y don't u rewrit server in c++
07:44 < Grum> SirCmpwn: its not like some pissy counterstriek game
07:45 < Grum> you HAVE to have the world in memory to simulate it
07:45 <+AndrewPH> countershrek
07:45 <+SirCmpwn> well, presumably most players on a homey server play near each other
07:45 < Grum> and its just a fucktone of data
07:45 < Grum> yes, and 17x17chunks is how much data?
07:45 <+SirCmpwn> 3
07:45 <+Amaranth> 21*21 actually
07:45 < Grum> yeah, just downcasting it a bit
07:45 <+Amaranth> default view distance is 10 so (10+1+10)^2
07:46 < Grum> 441 chunks, at 16 sections at 4096 blocks, is how much?
07:46 < Grum> 28901376 ...
07:46 <+Amaranth> 83MB, like I said :P
07:46 < Grum> at ~3.5bits/block is ....
07:46 < Grum> indeed
07:46 <+SirCmpwn> you keep all of the sections in memory?
07:46 < Grum> right now, no choice
07:46 <+Amaranth> This is worst case
07:46 < Grum> worstcase per player yeah
07:46 <+Amaranth> In reality without mods you're only going to use 2.5 bytes per block, not 3 bytes
07:46 < Grum> and yes, do you want to have the added latency of having to load this of disk on the main thread somewhere?
07:47 <+Amaranth> And most chunks aren't built up to 256, probably more like 96 on average
07:47 <+SirCmpwn> latency doesn't really factor into memory needs, Grum 
07:47 < Grum> io-locking your system while having to wait for it blocking does >.>
07:47 <+Amaranth> Which is more like 26MB
07:47 <+Amaranth> But that's a perhaps somewhat optimistic average case
07:47 <+SirCmpwn> spawned a vanilla world, highest point nearby is Y=72
07:48 < Grum> we can greatly improve the memory footprint *IF* we get rid of random ticking
07:48 <+SirCmpwn> you only need a quarter of the sections loaded
07:48 < Grum> the 'pure air ones' are a special kind that take less memory atm
07:48 < Grum> but we can do better than that
07:48 < Grum> but ... we need to kill the random ticking
07:48 <+SirCmpwn> I will admit that C.N loads all 16 as soon as you put a block in the top section
07:48 <+Amaranth> But if someone builds a single block at y=256 all the sections below it are suddenly real things using 16*16*16*2.5 bytes each
07:48 <+Amaranth> To store air :P
07:49 < umby24> Nice sidenote to the conversation, should cut down on the constant movement updates every tick, just my opinion though.
07:49 < Grum> yeah ... for now >.>
07:49 < Grum> umby24: you are not being specific enough
07:49 <+SirCmpwn> wow, I haven't played vanilla in a while
07:49 < Grum> i assume you mean: 'sent from the client' ?
07:49 <+SirCmpwn> there are fancy flowers everywhere
07:49 < umby24> yes
07:49 < SinZ> umby24: yup, because having the client / server spam the crap out of movement packets is very helpful
07:49 < Grum> the client shouldn't send anything unless the client has input
07:49 < Grum> but yeah, someone didn't quite design this shit
07:49 < Grum> we all know that, dead horse, beat and rape it a bit more!
07:50 < SinZ> wasn't the 1.7 butthurt update supposed to fix this stuff
07:50 < Grum> no?
07:50 < Grum> we cannot touch this for a while
07:50 <+Amaranth> The networking protocol is just enough serialization to split the client in half and call half of it a server :P
07:50 < Grum> even now
07:50 <+SirCmpwn> "fix stuff" has been scheduled and delayed for quite some time
07:50 < Grum> not delayed at all
07:50 < Grum> just not at the top of the todo list
07:50 <+SirCmpwn> that's an odd todo list you've got
07:50 < Grum> because of this giant layer of yaks ontop of it
07:50 < Grum> not by our 'design'
07:50 < Grum> we get rather limited by which cruft sits in our way
07:51 <+SirCmpwn> I met some yaks IRL once
07:51 <+SirCmpwn> they were nice
07:51 < Grum> they are
07:51 < Grum> also very hairy
07:51 <+SirCmpwn> I don't know if hair is the right word for... that
07:51 <+Amaranth> Networking comes later is what he means, they're working on rewriting the inventory and rewriting the internal representation of blocks and rewriting the renderer (more refactoring really) to be data driven
07:52 <+SirCmpwn> didn't they already try to fix networking
07:52 <+SirCmpwn> a couple versions ago
07:52 < umby24> cough 1.7
07:52 < SinZ> <SinZ> wasn't the 1.7 butthurt update supposed to fix this stuff
07:52 <+SirCmpwn> I spent weeks cleaning up that mess
07:52 < umby24> ^
07:52 <+Amaranth> That was the first indication of just how deep the rabbit hole goes in the weird assumptions the client and server make about movement packets
07:52 < Grum> inventory is on a hold atm
07:52 < Grum> having some design issues >.>
07:52 < Grum> no idea how to fix yet
07:52 < SinZ> having made client bots, movement packets are the spammiest of all spam
07:52 <+SirCmpwn> I'm glad we have a stronger variety of flowers now, though
07:53 < Eviltechie> SirCmpwn: opium poppies?
07:53 < Grum> SirCmpwn: again? you do realize that jens adds shit like that?
07:53 <+Amaranth> Gotta put features out there. For one people demand them and for two working on boring yak shaving all day is a good way to burnout
07:53 < Grum> all the 'features' have been added by people basically in their spare time .. OR by being jens
07:53 <+SirCmpwn> I don't know the politics of mojang, I just deal with their decisions
07:53 < dexter0> And I bet it took all but a few hours.
07:53 < Grum> indeed dexter0
07:53 <+SirCmpwn> oh, hi Grum
07:53 <+SirCmpwn> just realized who I was talking to
07:54 < Grum> good
07:54 <+SirCmpwn> 1.7 made me sad
07:54 < Grum> just list all the 'gameplay' features i added
07:54 < umby24> just hire SirCmpwn to be your program designer
07:54 < Eviltechie> do we count bugs as features?
07:54 <+SirCmpwn> please don't 1.7 again
07:54  * Eviltechie ducks
07:54 < Grum> Eviltechie: sometimes :(
07:54 < Grum> SirCmpwn: in which way? be more specific
07:55 <+SirCmpwn> well
07:55 <+SirCmpwn> hmm
07:55 <+SirCmpwn> not sure where to start
07:55 <+SirCmpwn> how about the fact that the same packet can have different IDs and structures depending on which way it's going
07:55 < Grum> why would that be the same? lol
07:55 < Grum> it was retarded it was the same
07:55 <+SirCmpwn> what?
07:56 <+Amaranth> The ids are an implementation detail of their codebase at this point
07:56 < Grum> it was retarded that packets were reused and then mangled data into fields 'just to reuse it'
07:56 < Grum> i mean, look at the login packet before ...
07:56 < Grum> seriously overloaded to fuck
07:56 < Eviltechie> I design databases like that. I am not a clever man.
07:56 <+Amaranth> Although I'd feel better saying that if they weren't explicitly numbered in the code :P
07:56 < Grum> same
07:57 < Grum> i can remove that in 30 seconds
07:57 < Grum> we just didn't ... yet
07:57 < Grum> as it served no benefit right now
07:57 <+Amaranth> Could have sworn it was just a counter that incremented as you registered them
07:57 < Grum> nah not yet :)
07:57 < Grum> should be, decided not to right now
07:58 <+SirCmpwn> how about using VarInts in places where it doesn't make any sense, especially given that they take more resources to read/write
07:58 <+SirCmpwn> that irked me a bit too
07:58 < Grum> like where?
07:58 <+SirCmpwn> or the whole JSON chat nonsense
07:58 <+SirCmpwn> packet IDs, for one
07:58 <+SirCmpwn> or even packet length
07:58 < Grum> so what should packetids be?
07:58 < Grum> or the length?
07:58 <+Amaranth> 255 packet ids ought to be enough for anyone
07:58 <+SirCmpwn> both should be a uint16_t, imo
07:59 < Grum> ah, so even shorter packets?
07:59 <+SirCmpwn> or a uint8_t for packet IDs, Amaranth is right
07:59 <+Amaranth> The plugin channel packet in 1.7.7 can be bigger than that already
07:59 < Grum> you do realize its a singluar fucking bit-check you call 'overhead' favorable to wasting a byte extra on the protocol
07:59 <+Amaranth> And I was being sarcastic, that's a rip off of a famous old saying
07:59 < Grum> for EVERY PACKET
08:00 <+SirCmpwn> wow, a byte
08:00  * SirCmpwn whistles
08:00 <+SirCmpwn> it just adds complexity that doesn't need to be there imo
08:00 < Grum> also, right now we can grow to ... whatever the fuck complexity we want
08:00 < Grum> you need varint reading anyway
08:00 < Grum> doesn't matter where we use it then
08:00 <+Amaranth> On the other hand you save a byte there and include an extra kilobyte in the spawn player packet. Although if you brought back the byte too people would complain even more
08:00 <+SirCmpwn> haha, I just found this in my 1.7 networking implementation
08:01 <+SirCmpwn> private static readonly Type[][][] NetworkModes;
08:01 < umby24> hmm lets consult my comments real quick
08:01 < Grum> SirCmpwn: not something fromm our source
08:01 <+SirCmpwn> not saying it's your fault
08:01 <+SirCmpwn> but it is amusing
08:01 <+SirCmpwn> I am glad you finally figured out how to do pings
08:02 < Grum> you mean reply with the smallest packet possible?
08:02 < Grum> and then re-consult for the actual information?
08:02 <+SirCmpwn> no, I mean using JSON so you don't paint yourself into a corner
08:02 <+SirCmpwn> five times
08:02 < Grum> lol
08:02 < Grum> L O L
08:02 < Grum> so THERE we should use json
08:02 < Grum> but for chat we cannot?
08:02 <+SirCmpwn> well, I personally wouldn't have used JSON
08:02 <+SirCmpwn> but it serves your needs fine in that case
08:02 < Grum> you do realize that you can infinitely nest translatable statements using chat right?
08:03 < Grum> embed styles/colors/whatever?
08:03 <+SirCmpwn> the JSON chat is a horrific bastardization of JSON that is absolutely ridiculous
08:03 < Grum> and that we actually use that
08:03 < Grum> and that there is no fucking nice way of doing that beyond making your own parser for a custom thing?
08:03 <+SirCmpwn> man 3 printf
08:03 < Grum> if i could ban, i would.
08:03 < Grum> seriously retarded answer SirCmpwn
08:03 <+SirCmpwn> you asked me to tell you my complaints about 1.7
08:04 <+Amaranth> You could do an HTML-like syntax but you'd just be parsing it into something that looks like your JSON
08:04 < Grum> Amaranth: indeed
08:04 <+SirCmpwn> and that wasn't entirely serious, as I'm sure you could have guessed
08:04 < Grum> which is why we opted for json right now
08:04 < Grum> we had it lying around, we can control it *very* well
08:04 <+Amaranth> Whether or not your chat should be a JSON representation of an AST is another thing but whatever, it works
08:04 < Grum> and its not too expensive size-wise
08:04 <+SirCmpwn> I could probably have been convinced that BSON wasn't a terrible choice
08:04 <+SirCmpwn> why JSON?
08:04 <+Amaranth> Now, for consistency, get rid of NBT :P
08:05 < Grum> because we already have a json dependency?
08:05 <+Amaranth> Are you fucking serious? BSON is a joke
08:05 < Grum> msgpack is one i am considering
08:05 <+Amaranth> BSON is totally Web Scale(tm) so you should definitely use it
08:05 < Grum> and bson is poop
08:05 < Eviltechie> I vote we switch to RS-232.
08:05 <+SirCmpwn> BSON isn't great, but it isn't bad
08:05 < Grum> you can say the same for NBT
08:05 <+SirCmpwn> I probably would have just reused NBT since it was already stuffed into it
08:05 < Grum> but i'll try and murder you in your sleep
08:05 <+Amaranth> BSON is a binary format that uses _more space_ than JSON's textual format
08:05 < Grum> NBT SUCKS
08:05 < Grum> it has the most fucked up API (none) ever
08:06 <+Amaranth> BSON is in fact kind of similar to NBT :P
08:06 < dx> HEY EVERYONE. LET'S NOT TALK ABOUT THIS. THANKS
08:06 <+SirCmpwn> well, sure, but you're already stuck with it
08:06 <+SirCmpwn> unless you want to make a new level format again
08:06 < Grum> i want to
08:06 <+Amaranth> msgpack is cool but in practice it seems gzip'ed json is about the same size
08:06 < Grum> ideally i'd put the levels into a leveldb
08:07 <+SirCmpwn> Amaranth: but Mojang doesn't gzip json :/
08:07 < Grum> (yes that thing by google)
08:07 < Grum> SirCmpwn: we don't have to?
08:07 <+Amaranth> And then there is always protobuf if you don't need to ship the schema with the data
08:07 < Grum> yeah we looked at protobuf for 1.7 but it was 'too soon'
08:07 <+SirCmpwn> Grum: please don't assume that every time I speak I'm expressing criticism for minecraft
08:07 <+SirCmpwn> I was just responding to Amaranth 
08:08 <+Amaranth> I was talking about a replacement for NBT, not for JSON chat
08:08 <+Amaranth> They gzip NBT
08:08 <+SirCmpwn> can we work MathML into minecraft somehow
08:08 < Grum> MathML? :)
08:09 <+SirCmpwn> http://aiju.de/rant/XML/MathML
08:10 < umby24> lolz
08:10 < Grum> i found my new format
08:10 <+SirCmpwn> what have I done
08:10 < umby24> now if we can save levels in that format, uncompressed naturally...
08:10 <+Amaranth> I'm not sure I get the point of leveldb. It's just an on disk sorted KV store?
08:10 < Grum> we'll story it pretty-printed in UTF32
08:10 <+SirCmpwn> I don't like client-side physics
08:11 <+Amaranth> I mean, if you're needing something like that may as well use theirs but it doesn't seem all that impressive of a feat of engineering
08:11 < Grum> Amaranth: singular file, highly incorruptable, fast
08:11 <+SirCmpwn> so clearly we should have the server send the physics calculations to the client via MathML to make sure there's no lapse in communication
08:11 <+Amaranth> Oh does it actually do things for data integrity? I thought it was just a dumb store with on the fly compression
08:11 < Grum> it has 'transactions' iirc
08:12 < Grum> so it has fallback states iirc
08:12 <+Amaranth> Oh, bummer, it's in the wrong language for you though :P
08:12 <+SirCmpwn> does anyone remember why minecraft encryption isn't just TLS
08:13 <+Amaranth> Yep
08:13 < Grum> heartbleed SirCmpwn
08:13 <+SirCmpwn> I was thinking about that the other day when reading about heartbleed
08:13 <+Amaranth> That's not why
08:13 <+SirCmpwn> Grum: that's not why
08:13 < Grum> obviously not :)
08:13 < Grum> because TLS would be too simple! don't be silly now
08:13 <+Amaranth> it's because the auth service provides a trusted partner for key exchange so you get to ensure clients are premium at the same time you share a secret with them
08:13 <+SirCmpwn> oh, "the other day" was earlier today
08:13 <+SirCmpwn> https://github.com/SirCmpwn/Craft.Net/issues/210
08:14 < Grum> we have a custom-enough security scheme its hard to write up with TLS
08:14 < Grum> we could use it to START a TLS connection, but meh
08:15 <+Amaranth> Plus with TLS you'd need self signed certs and cert pinning like SSH (which users would ignore) or you'd need servers to pay a CA in order to run a server in online mode
08:15 <+SirCmpwn> or you could establish Mojang as a trusted CA for Minecraft alone
08:15 < Grum> oh and TLS is a pain in the ass to understand in general :P
08:15 <+SirCmpwn> I'd do the self signed route, though
08:15 <+SirCmpwn> the reason I wondered about it is because roll-your-own crypto is nearly always a bad idea
08:15 < Grum> though we're basically doing the same under the hood now >.>
08:16 < Grum> we dont actually care to have strong encryption
08:16 <+Amaranth> That reminds me, I would have thought the protocol break with 1.7.6/1.7.7 would have been a perfect time to slip in the protocol break planned for 1.7.4
08:16 < Grum> just strong enough to get rid of MITM
08:16 < Grum> which one Amaranth?
08:16 < Grum> i did add the custom packet thingie
08:16 <+SirCmpwn> well, in 20 years, Minecraft will be unsupported and TLS will be supported
08:16 < Grum> i must have forgotten something >.>
08:16 <+Amaranth> A security one I'm only vaguely familiar with
08:17 < Grum> SirCmpwn: errr there is very little to support
08:17 <+SirCmpwn> the implication being that Minecraft crypto will be broken and TLS will not be
08:17 <+Amaranth> Grum: That reminds me, what's with the counter while reading NBT data?
08:18 <+Amaranth> The only thing I can think of that could help with is too late to be useful, I think
08:18 < Grum> Amaranth: haha you don't want to know, really :(
08:18 < Grum> IntArray[maxsize] ....
08:18 <+Amaranth> Right
08:18 <+SirCmpwn> hey Grum, if you ever do redesign worlds, please don't do DIM-1 again
08:18 <+Amaranth> I just used a pokemon exception handler for that one :P
08:18 <+Amaranth> gotta catch em all!
08:20 < Grum> there is little you can do when you run out of memory right there and then
08:20 <+Amaranth> SirCmpwn: The Bukkit design for multiworld predates vanilla minecraft's design and uses named worlds, each with their own level.dat
08:20 < Grum> we will NEVER EVER have DIM + [-1,0,1] again >.>
08:20 <+Amaranth> Grum: You don't actually run out of memory though, it just hits an OOM-like exception trying to allocate too much at once
08:20 <+SirCmpwn> the C.N design uses one level.dat and named worlds with directories full of regions
08:21 < Grum> CN?
08:21 <+SirCmpwn> Craft.Net
08:21 <+SirCmpwn> third party server/client/etc
08:21 <+Amaranth> SirCmpwn: So you can't have a superflat world on the same server as a normal one?
08:21 <+SirCmpwn> Amaranth: yeah, I want to change that
08:21 <+SirCmpwn> you caaaaaan
08:21 <+SirCmpwn> but it's weird
08:21 <+Amaranth> Need a per-world level.dat to do it "right"
08:21 <+SirCmpwn> I'm trying to keep compatability with vanilla worlds
08:21 <+SirCmpwn> which makes it more difficult
08:22 < Grum> there are actually some other 'design' issues
08:22 <+Amaranth> And then you need to learn the trick that is sending two world change packets to clear the client's cache or whatever of what kind of world they are in
08:22 < Grum> for example, someone starts a game in 'offlinemode'
08:22 <+Amaranth> Otherwise going from normal to superflat doesn't get rid of the void fog
08:22 < Grum> now he gets a wolf-pet; how he sends this world to someone else ... 
08:22 < Grum> what happens with the pet? :P
08:22 < Grum> who owns it?
08:22 < Eviltechie> oh, so that's how those plugins change sky colors...
08:22 <+SirCmpwn> Amaranth: thanks for that tip, someone else was asking me to make multiworld work in C.N
08:22 <+Amaranth> Sky colors are based on biome
08:23 < Grum> the whole 'worldchange'-packet crap .. brrrrrrrrrrrrrrrrrrrr
08:23 <+SirCmpwn> what's this "worldchange" packet
08:23 <+SirCmpwn> there's probably some inconsistency between whatever you guys name it and whatever wiki.vg names it
08:24 <+SirCmpwn> http://wiki.vg/Protocol#Respawn
08:24 < Grum> no idea what it is called with us
08:24 <+SirCmpwn> a situation that is made worse by the 1.7 packet ID murder
08:24 <+Amaranth> The respawn packet
08:24 < umby24> (speaking of the above respawn packet, with only 3 possible values for Dimension, why is it an int? xd)
08:25 <+Amaranth> Sending the extra fake trip to the nether also fixes stuff like "Avoid changing player's dimension to same dimension they were already in, weird bugs can occur i.e. such player will be unable to attack other players in new world (fixes after their death and respawn)"
08:25 <+SirCmpwn> there are lots of similar problems with the protocol, umby24, I doubt they'll all be fixed soon
08:25 <+Amaranth> Just make them all varint, problem solved
08:25 <+SirCmpwn> they're too busy cramming UUIDs in
08:26 < umby24> lmao ok..
08:26 < umby24> occationally while working with the packets I'll hit one like that and go "wait.. wtf?"
08:26 <+Amaranth> UUIDs are actually important infrastructure work, even if they are a PITA
08:26 <+SirCmpwn> I am quite amused that they adopted varints to save a byte per packet, and swapped out entity IDs for UUIDs
08:26 <+SirCmpwn> UUIDs are fine, but sent as a string?
08:27 <+Amaranth> umby24: It's an int because notch assigned it to an int while writing single player :P
08:27 <+Amaranth> entity IDs are not swapped for UUIDs
08:27 <+Amaranth> Although that'd be nice
08:27 < umby24> Thanks, notch. :P
08:27 <+SirCmpwn> I thought they were working on that, maybe I was wrong
08:28 <+Amaranth> If they were UUIDs instead using a proxy to switch between worlds would be simpler because you wouldn't have to rewrite the entity ID
08:28 <+SirCmpwn> C.N does not allow the same entity ID to exist in two worlds
08:28 <+SpaceManiac> mm, backlogs
08:29 <+Amaranth> The worlds in this case are actually separate servers, thus the proxy
08:29 <+SirCmpwn> oh, right
08:29 <+SirCmpwn> I suppose that's true
08:29 <+SirCmpwn> but Mojang isn't really giving much thought to that use-case, I'd imagine
08:30 <+Amaranth> Anyway, back around to the original topic, StoredUserList makes me super sad and actually made me stop working on updating CraftBukkit
08:30 <+SirCmpwn> what's StoredUserList?
08:30 <+SpaceManiac> I guess I need to update Glowstone... I got hung up trying to fix NBT being really stupid and gave up
08:30 <+SpaceManiac> The thing used for bans/whitelists, I'm guessing
08:31 <+SirCmpwn> details?
08:31 <+Amaranth> SirCmpwn: In 1.7.7 bans, whitelist, and op are stored in json by UUID, that's the class that loads them
08:31 <+SirCmpwn> that doesn't sound like a bad thing to me
08:31 <+SirCmpwn> what do you dislike about it?
08:31 <+Amaranth> But it uses GameProfile.toString() as a key and the toString() includes a UUID, name, skin blob, and "legacy" field
08:31 <+SirCmpwn> :|
08:32 <+Amaranth> So I have a UUID and I still need to do a network request to get useless extra information just so toString() will match
08:32 <+Amaranth> Or I can just iterate all the entries and hope no one has a big ban list
08:32 <+SirCmpwn> there is an awful lot of "it's a decent idea but the execution is junk" in minecraft lately
08:32 < Grum> Amaranth: its not actually right
08:33 < Grum> it stores a srting
08:33 <+Amaranth> SirCmpwn: At least we're up to decent ideas? ;)
08:33 < Grum> but i dont see any code that goes and fetch all the information
08:33 <+SirCmpwn> Amaranth: haha, I guess so
08:33 <+Amaranth> Grum: Right, even worse, there is a chance these are being loaded with an incomplete GameProfile which means I have the opposite problem, they don't actually work for an online player without making an extra GameProfile with _less_ information
08:34 <+Amaranth> That's a better problem to have though but I think they all include a name which is the killer
08:34 < Grum> i am missing where you are having problems
08:34 < Grum> gameprofiles have an optional name
08:34 < Grum> it has to have either a name OR a uuid .. or both
08:34 < Grum> iirc if it only has a name it'll look up the uuid somewhere
08:35 <+Amaranth> I have a UUID, I have a StoredUserList, I would like see if this UUID is in there without doing a network request or iterating all the entries
08:35 <+Amaranth> As far as I can tell the solution is to toss StoredUserList and parse the files myself
08:35 < Grum> i am not sure where StoredUserList is used directly
08:35 < Grum> maybe it should just be abstract >.>
08:36 <+Amaranth> It's all subclasses but they all chain through to the GameProfile.toString() for a key
08:37 <+Amaranth> I guess I could just make it not use toString but then I need to patch GameProfile still to have a special equals that violates the java spec :P
08:37 < Grum> no they dont
08:37 < Grum>     protected String getKeyForUser(K user) {  return user.toString() }
08:37 < Grum> but its overriden in all subclasses
08:37 <+Amaranth> Yeah they call WhiteListUserEntry.whatever() which returns an Object they call toString() on and that Object is a GameProfile
08:38 < Grum> where does it do that? :/
08:39 < Grum> only two methods in the entry and they are serialize/deserialize
08:39 < Grum> https://gist.github.com/grum/46e045280503139b77e1
08:39 < Grum> i dont like the fact that this class looks like this
08:40 < Grum> serialization/deserialization should be handled through json and a registration of a serialization class .... not being done by this 'container class'
08:40 <+Amaranth> Sorry the method is actually on the parent class
08:40 <+Amaranth> It returns an Object
08:40 < Grum> 'it'?
08:40 < Grum> the parent of that class has 1 method 
08:40 < Grum>     T getUser() {
08:40 < Grum>         return user;
08:40 < Grum>     }
08:41 <+Amaranth> I've got 5
08:41 < Grum> https://gist.github.com/grum/95e8c8a62a63c7b68d04
08:41 < Grum> yes, and two are constructors, 1 is void and 1 is boolean
08:41 < Grum> so 1 method ;)
08:41 <+Amaranth> It stores an Object field, has two constructors, a getter for the Object, and some other crap
08:42 < Grum> i do not understand why on earth the secondary constructor exists OR even gets called
08:42 < Grum> because nothing overwrites it
08:42 <+Amaranth> Right so T is GameProfile and in the end it all ends up with Object.toString() for a key
08:42 < Grum> so ... JsonObject just gets thrown away
08:42 < Grum> where does it do a Object.toString()?
08:43 <+Amaranth> Back in StoredUserList the add method
08:43 <+Amaranth> I've got this.d.put(a(paramJsonListEntry.f()), paramJsonListEntry); where f is the getUser() and a() is the method that returns Object.toString()
08:43 < Grum>         map.put(getKeyForUser(infos.getUser()), infos);
08:43 < Grum> no
08:43 < Grum> a is overloaded
08:44 < Grum> StoredUserList should have had 'a' as abstract
08:44 < Grum> the real problem is that this code is 'too generic' ... whcih makes it act like bullshit
08:44 <+Amaranth> My whitelist class doesn't have it overloaded
08:44 <+Amaranth> I've got a String b(GameProfile) that returns profile.getId().toString() but that's not overloading
08:45 < Grum> here it does;
08:45 <+Amaranth> Don't tell me this is some decompiler error
08:45 < Grum>    @Override
08:45 < Grum>     protected String getKeyForUser(GameProfile user) {
08:45 < Grum>         return user.getId().toString();
08:45 < Grum>     }
08:45 < Grum> the whole thing is, EVERY usecase of this object uses 'GameProfile' for 'K'
08:45 < Grum> which makes it super fucking pointless to even have K
08:46 <+Amaranth> Right well I didn't need that mouse
08:46 <+Amaranth> Good thing I still have a trackpad
08:46 < Grum> haha what happened? it learned to fly?
08:46 <+Amaranth> Sure did
08:46 < Grum> wired? :D
08:46 < Grum> wireless mouses fly so far ;(
08:46 <+Amaranth> Nah, bluetooth
08:46 < Grum> could be that we have fucked up obfuscation here btw
08:47 <+Amaranth> Wouldn't that mean this really is broken, not just broken for me trying to mod it?
08:47 < Grum> oh, ipbanlist is by string 
08:47 < Grum> i guess ... K is not always gameProfile :/
08:48 < Grum> i have no idea Amaranth
08:48 <+Amaranth> My GameProfileBanList also has String b(GameProfile) that return profile.getId().toString()
08:48 < Grum> ok now it is getting scary
08:48 <+Amaranth> Crap I don't want to learn to read bytecode, let me find Wolvereness
08:49 < Grum> should be 'easy'
08:49 < Grum> hmmz 
08:49 < Grum> which class in the jar?
08:49 < Grum> ah, ol
08:50 < Grum> https://gist.github.com/grum/14ad23e5b8ae48bd9df7 Amaranth
08:51 < Grum> line 61?
08:52 < Grum> is it doing the 'wrong fucking thing'? :/
08:52 <+Amaranth> No idea
08:53 < Grum> it calls 128
08:53 < Grum> which looks like the right method
08:54 <+Amaranth> Grum: Look at oq
08:54 <+Amaranth> That's the whitelist one
08:54 < Grum> yeah it seems ok for oi
08:54 < Grum> protected java.lang.String a(java.lang.Object);
08:54 < Grum>   Code:
08:54 < Grum>    0:	aload_0
08:54 < Grum>    1:	aload_1
08:54 < Grum>    2:	checkcast	#2; //class com/mojang/authlib/GameProfile
08:55 < Grum>    5:	invokevirtual	#16; //Method b:(Lcom/mojang/authlib/GameProfile;)Ljava/lang/String;
08:55 < Grum>    8:	areturn
08:55 < Grum> that is in oq AND oi
08:55 < Grum> so .... it wraps to 'b()' in those classes?
08:55 < Grum> no idea why/how/wtf
08:56 <+Amaranth> It's because the parent takes Object and they take GameProfile?
08:56 < Grum> i guess this is how java handles generics?
08:57 < Grum> by injecting a virtual to make it compatible with 1.3bytecode? >.>
08:57 <+SpaceManiac> Oh, yeah, there's some wacky feature for that
08:58 < Grum> lol without JIT java runs like .......
08:58 <+Amaranth> God damn it this is a jd-gui error
08:58 <+Amaranth> Our real decompiled version has it just fine
08:58 < Grum> i have to say the javap -c output is really nicely readable :)
08:59 <+Amaranth> I've been trying to figure out how to make this work and chasing a decompiler error in a decompiler I don't even actually use (except for browsing)
08:59 < Grum> you suck Amaranth
08:59 < Grum> and you made me angry!
08:59 < Grum> go write a better decompiler
08:59 < Grum> actually, you can use the frontent on procyon
08:59 <+Amaranth> haha, I still haven't found one better than fernflower
08:59 <+Amaranth> procyon fails to decompile
09:00 < Grum> https://github.com/deathmarine/Luyten
09:00 < Grum> yeah but it will do this one right ;)
09:00 < Grum> fernflower gives the 'most usable code' yeah
09:00 < Grum> procyon the best code wherever it gives code ;)
09:00 <+Amaranth> Grum: Did you by any chance upgrade proguard or something?
09:00 < Grum> not recently
09:01 <+Amaranth> Well recently as in between 1.6 and 1.7
09:01 <+Amaranth> 1.7 on seem to have new tricks that are great at breaking decompilers
09:05 < Grum> nah its just that decompilers do not handle abstract enumes
09:05 < Grum> -e;
09:08 < Grum> so tempting to look at the source hehe
12:41 < dividuum> Dinnerbone, MC-52755 is fixed? How?
14:15 < redstonehelper> hmm, player heads still display their skins in multiplayer without the person ever having been on that server, right?
14:16 < redstonehelper> could that be used to make npc players have their skin again?
14:16 < redstonehelper> or do player skulls no longer go by the player skin directly?
14:17 < Thinkofdeath> The skin blob used by skulls is flagged to prevent it from being used for spawning players
14:18 < redstonehelper> so if I play on a server and place a player head of myself, will my skin have to be sent twice?
14:19 < Thinkofdeath> The skin blob will but the client is smart enough to only download it once
14:19 < Thinkofdeath> the blob only contains the url to the skin
14:19 < Thinkofdeath> + some timing/username info
14:20 < redstonehelper> thanks
19:38 < Not-001> [node-minecraft-protocol] roblabla pushed 5 commits to master [+0/-0/±10] http://git.io/9GPA1Q
19:38 < Not-001> [node-minecraft-protocol] roblabla 0a4b365 - Add opportunistic parsing.
19:38 < Not-001> [node-minecraft-protocol] roblabla b7dd1ba - Remove old debug code
19:38 < Not-001> [node-minecraft-protocol] roblabla c1a5b82 - woops, add back the 'packet' event
19:38 < Not-001> [node-minecraft-protocol] ... and 2 more commits.
19:44 < Not-001> [node-minecraft-protocol] roblabla pushed 1 commit to master [+0/-0/±2] http://git.io/ybxXnA
19:44 < Not-001> [node-minecraft-protocol] roblabla 23ecbd3 - Release 0.12.0
20:18 < rom1504> nice :) , I wonder how hard it would be to update mineflayer with that new node minecraft protocol version
20:23 -!- Irssi: #mcdevs: Total of 129 nicks [1 ops, 0 halfops, 10 voices, 118 normal]
23:33 < dividuum> Thanks Grum and Dinnerbone for the changes in 1.7.8!
23:35 < AlphaBlend> fixes you mean?
23:36 <+sadimusi> dividuum: did they remove the timestamp?
23:37 < dividuum> no. the timestamp is still there. but now if the client is presented outdated (but otherwise valid) skin data, it'll go to the session server and renew it from there
23:38 < Grum> which 'fixes' the 'fail clocks' (which will later just be fatal for the user -- when we can properly tell him)
23:39 < Grum> and 'fixes' the users being online for 24hrs .. which is barely ever any usecase anyhow
23:39 <+sadimusi> except for NPCs
23:39 < AlphaBlend> o_O
23:39 < Grum> not sure who cares for npcs and you can cycle those trivially
23:39 < dx> the bug "Staying on a server for more than 24 hours crashes everyone" has 33 votes
23:40 < AlphaBlend> what's about this breaking NPC support?
23:40 < Grum> no idea AlphaBlend
23:40 < AlphaBlend> i've never worked with sole NPCs
23:40 < Grum> we never had npc support
23:40 < AlphaBlend> so i wouldn't know
23:40 < AlphaBlend> well, I guess Amaranth's fake player plugin would break from this
23:41 < Thinkofdeath> Oh the client renews it itself? I thought it just went to steve
23:41 < Grum> we're being nice for now
23:41 < AlphaBlend> Thinkofdeath: What is spigot going to do in 1.8? Create a protocol hack?
23:42 < Thinkofdeath> Well we have a 1.7.8 one so why not? :)
23:42 < AlphaBlend> or have you guys already thought of the impending doom with it
23:42 < Grum> the whole reason this is a problem is because of bullshit like spigot
23:42 < Grum> the 3 day hoop-jumping ... just because people have to fudge the rules
23:42 < Thinkofdeath> Hey I didn't know at first
23:43 < Grum> if we change protocol we do it for a reason
23:43 < Grum> not to 'just do whatever the fuck you want to pretend to be compatible with it'
23:43 < Thinkofdeath> I thought it was intended as a placeholder for 1.8
23:43 < Thinkofdeath> I came and asked when I noticed it was an issue
23:44 < Thinkofdeath> and thats when you found the bug
23:44 < Grum> but its ok, i'm totally happy to randomize the packet-ids for whatever next release we do
23:44 < Thinkofdeath> :P
23:44 < Grum> as for us, it matters nothing
23:44 < dx> grum is such a nice person
23:44 < dx> <3
23:44 < Thinkofdeath> tbh I thought you would do that sooner
23:44 < Thinkofdeath> My server even was ready for it
23:45 < AlphaBlend> Thinkofdeath: Good luck making a protocol hack for 1.8. You're going to end up making hacks so the client can read the data properly, based off what I know already
23:46 < Thinkofdeath> AlphaBlend: I've made one already somewhere
23:46 < AlphaBlend> oh?
23:47 < AlphaBlend> how often does spigot update to snapshot versions?
23:47 < dividuum> <Grum> we're being nice for now <--- should I be scared about the future? :-)
23:47 < Thinkofdeath> Normally closer to release 
23:47 < AlphaBlend> hahaha
23:49 < Thinkofdeath> Also "<Grum> if we change protocol we do it for a reason" wasn't always true in the past
23:50 < AlphaBlend> anytime it changes, there's a reason why
23:55 < AlphaBlend> ha, i laugh at all the fire that is going on in terms of modding and stuff
23:55 < AlphaBlend> "you guys are breaking this and that"
23:56 < AlphaBlend> well, this is what people get for going with 3rd party modifications that introduce communities. When Mojang hasn't had an official API, people get busy. Now they're working hard on one, and people want to complain
--- Day changed sam. avril 12 2014
09:18 <+Amaranth> Thinkofdeath: What makes you think http://wiki.vg/Pre-release_protocol#Teams uses UUIDs now?
09:18 <+Amaranth> Looking at the latest snapshot's /scoreboard command it still seems to use names to add players to teams
10:13 < Thinkofdeath> Amaranth: https://github.com/thinkofdeath/mcprotocol/blob/master/play/PacketClientbound3E.java#L76 String length changed to 40
10:13 < Thinkofdeath> Amaranth: (Note Pre-release is 1.8, The main protocol page is 1.7.6+)
10:14 <+Amaranth> I know, I was looking at 14w11b's scoreboard command
10:14 <+Amaranth> People add arbitrary strings to teams so they can use prefix/suffix to make longer text
10:16 <+Amaranth> I was going through the Bukkit scoreboard API changing it to use Strings and remembered you had said teams were based on UUID now
10:16 <+Amaranth> I think I'll just leave them as "players" in the API and only change the stuff I know can be arbitrary strings
12:35 < Not-001> [node-minecraft-protocol] roblabla tagged a679994 as 0.12.1 http://git.io/41V3Xg
12:35 < Not-001> [node-minecraft-protocol] roblabla pushed 2 commits to master [+0/-0/±3] http://git.io/lDWe5Q
12:35 < Not-001> [node-minecraft-protocol] roblabla d7f3e5c - Support 1.7.6 protocol
12:35 < Not-001> [node-minecraft-protocol] roblabla a679994 - Release 0.12.1
12:35 < Not-001> [node-minecraft-protocol] roblabla tagged 23ecbd3 as 0.12.0 http://git.io/6-EqwA
14:40 < codename_B_> 1.7.8 still has teams as players, yep, I can confirm this
16:46 <+Amaranth> codename_B_: So does the last 1.8 snapshot but apparently they are preparing to change that
16:46 <+Amaranth> Or someone did a search/replace and hit an extra spot
17:30 < codename_B_> shouldn't change the bukkit implementation surely, since OfflinePlayer will have a UUID
17:36 <+Amaranth> codename_B_: I was converting it to not use UUID
17:36 <+Amaranth> Err, to not use OfflinePlayer
17:37 <+Amaranth> I'm going to just leave adding players to teams as OfflinePlayer for now though
17:39 < codename_B_> hmm, yeah fair enough
17:40 < codename_B_> tbh its probably best not to change anything till there's a concrete wtf is going on answer
--- Day changed dim. avril 13 2014
01:43 < libraryaddict> Grum you alive?
01:48 < dx> libraryaddict: don't randomly ping the devs like that
01:48 < libraryaddict> Randomly?
01:48 < dx> well, you're not even asking a real question
01:48 < libraryaddict> Actually I suppose it would be better to ask Dinner
01:49 < libraryaddict> Wanting to inquire if they have any plans to remove the one uuid visible restriction
01:51 < shoghicp> well, just ask here. They might see it later
01:52 < dx> or they might ignore you completely, which is completely acceptable given the amount of noise they deal with, but you might have more luck getting a reply if you send them signal instead of noise
01:53 < libraryaddict> Also another thought I had that was similar to it was that if you are enforcing that. Instead of d/c the client. Why not just remove the old player?
01:53 < libraryaddict> I believe there is bug reports already where you can replicate it in vanilla
01:53 < Thinkofdeath> And they will fix those cases
01:54 < libraryaddict> Yes but I don't honestly think they will give modders any leeway to creating a game experience with fake players
01:59 < mbaxter> What is the limitation on faked players? Do the UUIDs have to be the specific calculated OfflinePlayer:$playername matching the sent $playername ?
02:00 < libraryaddict> From what I understand of it. Currently, if you want a player with a skin you need to send the spawn packet with the skin, correct uuid and correct name. The skin must still be valid or no skin
02:00 < libraryaddict> If you want a player with no skin, you send a player spawn packet without a skin blob and with a non version 4 UUID. Any name works.
02:03 < mbaxter> oh, just non version 4. That's good to know.
02:03 < libraryaddict> Yes but I believe thats removed in 1.8
02:05 < mbaxter> Well, I'm just writing for 1.7.8 at the moment :3
02:07 < dividuum> just a random though: what happens, when mojang has to revoke the signing key for the skins? Every old minecraft version won't display skins any more? Or depending on what happens when they get presented a skin the cannot validate: disconnect
02:09 < libraryaddict> Another thought. What will mojang do when a market starts for buying/selling skin blobs?
02:10 < dividuum> <whisper>add the timestamp back in to prevent that.</whisper>
02:11 < libraryaddict> Yes but if it was large enough, then depending on what names you want and if you really just want random names.
02:12 < Thinkofdeath> They are adding back when they can display a message to the user with the out of date blob/time issues so they can fix that
02:12 < Thinkofdeath> (If I understood correctly)
02:13 < dividuum> Hm? "You're clock is set wrong !" [ I know ] [ Nope ]
10:37 <+SirCmpwn> my friends and I have realized that we are unsatisfied with minecraft as it is today
10:37 <+SirCmpwn> and we're going to start a small beta 1.7.3 server
10:37 <+SirCmpwn> we all agreed that beta 1.7.3 was the peak of Minecraft's life
10:40 < dx> are you sure you're not unsatisfied with minecraft because you're bored of it?
10:40 <+SirCmpwn> maybe, but I've felt like a lot of the changes are superfluous for a while now
10:41 < dx> yeah, probably
10:41 <+SirCmpwn> I was against command blocks, basically the entire redstone update, etc
10:41 < dx> you're against command blocks? but how will sethbling survive? bukkit plugins? oh wait
10:42 <+SirCmpwn> 1.7.3 had all the core stuff to make minecraft minecraft, and none of the extra junk
10:42 <+SirCmpwn> it had a decent amount of redstone, it had the nether (and notably not the end)
10:42 < dx> yeah, i do agree that it seems like a peak
10:42 <+SirCmpwn> no experience or enchanting
10:42 <+SirCmpwn> no hunger
10:42 < prplz> you don't like hunger SirCmpwn ?
10:42 <+SirCmpwn> it's not neccessary to make minecraft minecraft
10:43 <+SirCmpwn> 1.7.3 also had the far lands
10:43 <+SirCmpwn> and the old lighting system <3
10:43 < libraryaddict> I'm only for the latest MC because you can do so much more with modding
10:43 <+SirCmpwn> can you? Old MC was pretty moddable
10:44 < libraryaddict> Well. I'm refering to vanilla clients and modded servers
10:44 < libraryaddict> Namely bukkit
10:44 <+SirCmpwn> ah
10:44 <+SirCmpwn> bukkit was around in 1.7.3 times
10:44 < libraryaddict> Yeah but back then you couldn't name a item or set lore on it. Which removes item menus
10:45 <+SirCmpwn> I might mod in a few changes from beta 1.8 into 1.7.3
10:45 <+SirCmpwn> maybe beef, melons, pressure plates on fences, sprinting
10:46 <+SirCmpwn> hold to charge bows
10:46 < libraryaddict> The 'Ctrl to auto sprint' is basically sprint hacks .-.
10:46 < libraryaddict> You can sprint even as you attack
10:47 <+SirCmpwn> 1.7.3 also had pistons
10:48 <+SirCmpwn> no server list, though, which was annoying
10:49 < prplz> why not the last beta, and remove the unused exp system?
10:49 < prplz> then you get sprinting too
10:50 <+SirCmpwn> don't want 1.8's endermen, or lighting system, or far lands fix, or hunger, or terrain generator
10:50 <+SirCmpwn> most of the features since 1.7.3 weren't terrible alone, but together they all just made minecraft... different
10:52 < dx> i saw all these feature updates as stuff to keep minecraft interesting for a bit longer
10:52 <+SirCmpwn> you're right
10:52 < libraryaddict> I don't see the skin change doing anything positive tho
10:53 < libraryaddict> Everything else they've added. Yes
10:53 <+SirCmpwn> I would pick and choose a small handful of changes since 1.7.3 to add, if I could
10:53 <+SirCmpwn> most of them aren't needed, especially the End
10:53 <+SirCmpwn> and everything related to the end
10:53 < prplz> the end was just to give the game well, an end
10:53 < prplz> :P
10:58 < dx> SirCmpwn: have you tried some of those , btw? playing that seems to be the opposite of what you're saying about
10:58 < dx> ...
10:58 <+SirCmpwn> ?
10:58 < dx> ok seriously how does that happen
10:58 < dx> i dunno, i must have some keybinding for "enter"...
10:58 < dx> ignore that message
10:59 <+SirCmpwn> consider it ignored
10:59 < dx> SirCmpwn: have you tried some of those modpacks like FTB, btw? playing that seems to be the opposite of what you're saying about 1.7.3
10:59 <+SirCmpwn> I have tried a few from time to time
10:59 <+SirCmpwn> I hate them
10:59 <+SirCmpwn> some of my friends tried to get me into Tekkit
10:59 <+SirCmpwn> and I tried FTB myself
10:59 <+SirCmpwn> couldn't stand either
10:59 < dx> heh, the opposite indeed
11:02 <+SirCmpwn> off to sleep
11:02 <+SirCmpwn> perhaps I will make a 1.7.3 branch for Craft.Net
11:03 <+SirCmpwn> I could dig up the old 1.0 code, that's when I started writing minecraft code