10:33 < Grum> And the fun part.  Almost all of them are "this was broken in the past and never noticed" 
10:33 < WizardCM> xD
10:34 < Grum> Well... Fun
10:35 < WizardCM> bugs are always fun
10:35 < WizardCM> until you try to fix them
10:35 < WizardCM> then realise how complicated they are
10:35 < WizardCM> and give up
10:35 < WizardCM> :D
10:36 <+SpaceManiac> classic technique
10:37 < Dinnerbone> md_5: Thanks for the links I will look into it later
10:37 < Grum> Almost all issues are of block polling for meta data properties and assuming it of a certain type and then it actually was the wrong block
10:38 < Grum> So in the past it would have assumed properties of something but that something wasn't actually there
10:38 < Grum> And now. Booooooom
10:39 <+Amaranth> Kind of like that time you added iterators to everything and CMEs like crazy
10:39 < Grum> Yup
10:39 <+Amaranth> Then I added an iterator cache to make it hurt less and then you removed all the iterators :(
10:40 < Grum> Nice indication of fragile code
10:47 < Grum> Yogu: sorry for late answer, phone chat sucks ;)
10:47 < stuntguy3000> Grum: Switching onto the unrelated topic of Account Security, I [and a lot of other people] feel Mojang should implement stronger 2 Factor Authentication. Kudos for having Security questions, but I feel it is not enough. A good popular soultion is a QR Code system, using a system such as Google Authenticator. Is there anything you can do/anyone to speak to, to get this implemented?
10:48 < Grum> Yogu: the 'end goal' will be to know all known blockstates and then assigning the ones with ids as we use them and tell the client (this includes sending the 'mapping over' when we connect)
10:48 < Grum> stuntguy3000: you can poke the webdevs on twitter perhaps
10:48 < stuntguy3000> Who are they?
10:48 < Grum> stuntguy3000: i agree it would be nice but I also know they have other things that are not nice to have on their plates
10:49 < WizardCM> probably a list of them on mojang support
10:49 < Grum> yeah
10:49 < Fenhl> yeah, the security questions are useless
10:49 < stuntguy3000> WizardCM: Yeah found a list 
10:49 < Grum> i'd have to go look it up too ;)
10:49 < Yogu> Grum: That's what I hoped for. But I suppose it will not be in 1.8, right?
10:49 < Grum> What I heard is that we plan to do a; "known location" system -- like Steam
10:49 < Grum> Yogu: nope 1.9 hopefully, requires a worldformat change
10:50 < Yogu> ok, thanks!
10:50 < WizardCM> what will that be, world format change number... 5?
10:51 <+Amaranth> People were getting all excited about the idea of having 32768 ids available (or 65536 if you go unsigned) only to learn you plan to occupy like half of them and if they make their blocks work like yours a single large mod will fill the rest :P
10:51 < Fenhl> counting from when?
10:51 <+Amaranth> WizardCM: Counting from indev?
10:51 <+Amaranth> Since alpha we've been through three formats
10:51 < Grum> WizardCM: will also be the last one
10:51 < Grum> because the idea is that we modularize the format
10:52 < WizardCM> since alpha is what i'd count, yeah
10:52 < Grum> so the chunks themselves will get a 'type id'
10:52 < Yogu> I was wondering about that, too. It wouldn't be a problem if the blockid-blockstate mapping will be per chunk column or so
10:52 < WizardCM> all good things
10:52 < Grum> Yogu: internally the only thing we store are BlockStates
10:52 < Grum> and they are mapped to a short
10:52 <+md_5> https://api.mojang.com/profiles/page/1             <--- why is this throttling by returning 204 no content
10:52 < Fenhl> and type id is versioned?
10:52 < Grum> (well char in java ..mmm unsigned)
10:52 < Grum> Fenhl: no
10:52 < Grum> but our plan is to make a 'reader/writer' system
10:53 < Grum> so we only write in 'certain formats' but we make readers for all old formats
10:53 <+md_5> sending the same request twice within ~5 minutes, gets a 204 as the response
10:53 <+Amaranth> btw looking up the Block/Tile for a given id is one of the most called things in the server
10:53 < Yogu> I was referring to Amaranth's comment about "short" being a bit little to store all blockstates that could occur in one world
10:53 <+Amaranth> Don't remember if it was #1 or not
10:53 < Grum> Amaranth: that is silly cheap now
10:53 < Grum> its an arraylist index -> BlockState
10:53 <+md_5> actually
10:53 < Grum> and the way back is an IdentityHashMap -> id
10:53 <+md_5> the web server is broke
10:54 <+md_5> its returning 204 even when there is content
10:54 <+Amaranth> Grum: if (id > 0 && id < stuff.size()) { return stuff.get(id); } else { return stuff.get(0); } is more of less what it is doing
10:54 < Grum> Yogu: well ideally with the 'typed chunks' we can make a type of chunks that allows for bigger ranges
10:54 <+Amaranth> And iirc it was like 5% of all CPU time, let me pull up the numbers
10:54 <+md_5> and setting Content-Length to 0, even though there is content
10:55 <+Amaranth> Although to be fair I think I figured out like half the time was just method call overhead which I suspect the JIT would normally have optimized out
10:55 < Grum> Amaranth: i think that is different now
10:55 <+SpaceManiac> Typed chunk system sounds intriguing
10:55 < Grum>     @Override
10:55 < Grum>     public V get(final K key) {
10:55 < Grum>         final V v = super.get(key);
10:55 < Grum>         return v == null ? defaultValue : v;
10:55 < Grum>     }
10:55 < Grum> SpaceManiac: this would mean we can do palleted chunks, or a special chunk that is just for 'JUST AIR DAMMIT'
10:55 < Grum> and that chunk would just be 1 varint big
10:55 < WizardCM> ^ i like this idea
10:55 <+Amaranth> Grum: Yeah that's the same I think, I was just collapsing it into a simple thing
10:56 < Grum> Amaranth: that is deep in the guts for the mapping
10:56 < Grum> i dont think we use that anymore
10:56 < Grum> let me see
10:56 < Grum>     public static final IdMapper<BlockState> BLOCK_STATE_REGISTRY = new IdMapper<BlockState>();
10:56 < Grum> nope
10:56 < Grum> that one is just a hash null-returning bastard
10:56 <+Amaranth> Yeah that's what 1.7.10 has too
10:56 < Grum> nope
10:56 < Grum> this is new
10:57 < Grum> in 1.7.10 there is a     public static final DefaultedMappedRegistry<ResourceLocation, Block> REGISTRY = new DefaultedMappedRegistry<ResourceLocation, Block>(DEFAULT_KEY);
10:57 < Grum> in snapshots we have that one *AND* the blockstate mapping
10:57 < Grum> and IdMapper is just:
10:57 < Grum>     private final IdentityHashMap<T, Integer> tToId = new IdentityHashMap<T, Integer>(512);
10:57 < Grum>     private final List<T> idToT = Lists.newArrayList();
10:57 < Grum> and yeah, there is a fuckton of virtual call overhead if you do not have jitted code
10:57 <+Amaranth> I was more or less collapsing the logic into a single bit of code
10:58 < Grum> well, the IdMap has no defaults
10:58 <+Amaranth> Instead of having the if and returning null then checking for null
10:58 <+Amaranth> Which is worse, really
10:58 < Grum> mm how do you mean?
10:58 <+Amaranth> That's an extra branch :P
10:58 < Grum>         final V v = super.get(key);
10:58 < Grum>         return v == null ? defaultValue : v;
10:58 < Grum> i dont see how you can do that more effective
10:59 <+Amaranth> I considered making it just return things.get(id & things.size()); and fuck anyone who goes over :P
10:59 <+Amaranth> But I guess you might have holes too
10:59 < Grum> problem is that in 1.7.x we have gaps
10:59 < Grum> we dont have gaps anymore
11:00 < Grum> and yes, that is what it will be
11:00 < Grum> if you ask for something that has no data ... BOOOOOM
11:00 < Grum> fuck you, dont do that
11:00 < Fenhl> Grum: ah, so you could use chunk type for versioning by simply deprecating one type in favor of another
11:00 < Grum> Fenhl: yup
11:00 < Dinnerbone> md_5: passed it on
11:00 < Grum> and the only 'code' we need to leave around is a way to read a certain 'type of data'
11:00 < mrapple> would you consider making it possible for servers to keep their custom tab lists? or at least some way of displaying the same information
11:01 < Grum> which is most likely going to be a super small conversion class
11:01 < Grum> mrapple: I think the most we should do is allow a 'displayname'
11:01 <+md_5> thanks, I tweeted Xlson too
11:01 <+Amaranth> Like Bukkit! :D
11:01 <+Amaranth> (which btw you also broke with that change)
11:01 < Grum> Amaranth: yes worldfolder names like bukkit hehehe
11:01 < mrapple> Grum: well, that's essentially what we have now. can we control the sorting?
11:01 <+Amaranth> Huh? I meant the player tab list display name
11:01 < Grum> mrapple: no
11:02 < mrapple> can you consider allowing that?
11:02 < Grum> no
11:02 <+Amaranth> Bukkit has API for that which at the moment won't work come 1.8
11:02 < Dinnerbone> You could control the sorting by having lots of teams sent to the client
11:02 < Grum> indeed
11:02 < Dinnerbone> And each player under each team
11:02 <+Amaranth> So I'll be happy with just getting the display name :D
11:02 < mrapple> oh, good point
11:02 < Dinnerbone> And with that, display names are done per team
11:02 < Grum> but the real thing you want is to be able to add extra information
11:02 < Dinnerbone> So you'd have that too
11:02 < Fenhl> but it does more, sounds like a good idea
11:02 < mrapple> does it accept fake UUIDs?
11:02 < Grum> Dinnerbone: can we let them set a 'server message' (like MOTD) that shows up above the tablist? -- multiline textcomponent?
11:02 < mrapple> Grum: we already have the boss bar for that
11:03 <+Amaranth> Grum: There really isn't room for that, is there?
11:03 < mrapple> the point behind the tab list is the toggleable-ness
11:03 <+md_5> Dinnerbone also something about the protocol: Please can you fail gracefully, ie: ignore / log to console. Crashing the client because of an incorrect team name is silly.
11:03 <+Amaranth> Don't break the state machine man
11:04 < Yogu> mrapple: I feel you're talking past each other. As I understand, the 'server message' will be only visible in the tab menu, so it is not the same as the boss bar.
11:04 < mrapple> Yogu: 
11:05 < mrapple> we're more interested in the grid display capabilities of the tab list
11:05 < WizardCM> ^
11:05 < mrapple> (sorry for the early return press)
11:05 < Dinnerbone> md_5: I can't do that sorry. I'm not going to ignore bad data from the server, if the server sends bad data and I ignore it then the behaviour will just be completely unknown
11:05 < Dinnerbone> If the server sent one bad thing, what else would it send wrong?
11:05 <+Amaranth> Grum: btw, guess what the most called method is for a 200 player server (or maybe 500, don't remember). MathHelper.floor(double) :D
11:06 <+md_5> does it still physically crash the client
11:06 < Dinnerbone> And what should the client do when things are then going to reference the now missing thing?
11:06 <+md_5> or just kick the player
11:06 < Dinnerbone> It should kick the player
11:06 < Spurlex> Grum on topic of Account Security. Would you ever implement security questions on launcher if “suspicious” activity detected by an IP (Login from multiple countries etc)
11:06 < mrapple> Dinnerbone: I haven't looked into the new tab behavior, but does it expand to a set size?
11:06 <+Amaranth> Spurlex: He mentioned something like Steam Guard
11:07 < Dinnerbone> It expands or shrinks to accomidate the list of users
11:07 < mrapple> what if there are more than 60 users though
11:07 < Spurlex> Haven’t used Steam so I will go google what that is.
11:07 < Dinnerbone> It sacrifices information on each element, like ping, to show more I think
11:07 <+Amaranth> Spurlex: First login on a "new" system generates an auth token which is emailed to you, if you don't enter it you don't get in
11:07 < Dinnerbone> After some point, based on the user screen size and stuff, it will just now show the remainder
11:08 < mrapple> hm
11:08 < WizardCM> hm indeed
11:08 < Dinnerbone> not show*
11:08 < Spurlex> That is a good idea. I’ve just seen an increase in giant alt lists being shared by tonnes of users. See http://MCAlts.com 10cents an account.
11:08 < mrapple> if it's not consistent will we not be able to reliably show info to users?
11:09 <+Amaranth> On the other hand I get a much nicer looking tab list :D
11:09 <+Amaranth> And maybe it'll even stop generating N^2 packets to update
11:09 < mrapple> sacrificing showing skin/ping is definitely nice
11:09 < mrapple> Amaranth: afaik you can send all tab list updates in one packet
11:09 < Dinnerbone> ^
11:09 <+Amaranth> Finally ;)
11:09 <+SpaceManiac> Is showing skin on tablist a thing?
11:10 < Dinnerbone> Yes
11:10 < mrapple> Dinnerbone: does the skin disappear with the ping info after a certain theshold?
11:10 < Dinnerbone> The reason player list tab is "less customisable" now is because there's a central player info store between server and client. Player add packet is no longer as big as it was, because of this. The tab list uses this store now
11:10 < Dinnerbone> I don't think so
11:10 < Dinnerbone> I'm not sure if the ping even does, I can't remember
11:10 < Dinnerbone> I know scores do
11:10 <+Amaranth> So you no longer have to send the skin blob every time a player comes in to range, just once on join
11:11 < Dinnerbone> Yes
11:11 < Grum> yeah
11:11 < Grum> we're using the serverlist for 'good things now'
11:11 < Grum> so your abuse is no longer 'tolerated' :p
11:11 <+SpaceManiac> Useful
11:11 < Grum> and or easily to hack in
11:11 < Grum> it just made sense
11:11 < Grum> but i fully agree with custom display names 
11:11 < mrapple> I understand your usage of it now, it's definitely good
11:11 < Grum> that is low impact
11:11 < Dinnerbone> http://wiki.vg/Pre-release_protocol#Player_List_Item
11:12 < Dinnerbone> (This page is becoming long. I like that.)
11:12 < mrapple> but allowing custom data display (along with display names) are my only requests
11:12 < mrapple> can we just send a fake UUID/name for fake information?
11:12 < Grum> no
11:12 < Dinnerbone> mrapple: I will let you provide an arbitrary multiline Component to display at the top of the player list
11:12 < mrapple> :O
11:12 < mrapple> that sounds fancy
11:12 <+Amaranth> Formatting is going to be impossible on that of course
11:12 < Fenhl> but does it do tables?
11:13 < mrapple> my only concern would be the loss of grid-like capabilities
11:13 <+Amaranth> Since resource packs can change fonts
11:13 <+md_5> Thinkofdeath you can harass them here now :p
11:13 < mrapple> I would much rather have custom information in the tab list than a multiline text component
11:13 < WizardCM> table is the best part
11:13 < mrapple> indeed
11:13 < dx> minecraft is getting a built-in html renderer?
11:13 <+md_5> Grum we've got that feedback you wanted, some chunk causes the client to crash, but only in latest snapshot
11:13 < Thinkofdeath> md_5: :P
11:13 <+Amaranth> dx: Stabbity stab
11:13 < dx> :D
11:14 < Thinkofdeath> http://i.imgur.com/jIoWUi5.png
11:14 <+SpaceManiac> Imagine how long it would take to decide on an API for *that*!
11:15 < WizardCM> for a html renderer?
11:15 <+Amaranth> Hey it only took 3 years for us to make a damage API that actually lets you modify armor reduction
11:15 < WizardCM> xD
11:16 < WizardCM> Amaranth: i guess 7
11:16 <+Amaranth> (you still can't modify damage boosters like sharpness and strength, go away)
11:18 < Fenhl> dx: btw are you dx on EsperNet too?
11:18 < dx> Fenhl: correct
11:18 < mrapple> Dinnerbone: I know you allowed v2 uuids for NPCs, would you consider allowing v2 uuids on the tab list for information display?
11:19 < mrapple> or v3 I guess
11:19 < Thinkofdeath> v3 == offline mode
11:19 < mrapple> right
11:24 < Fenhl> Thinkofdeath: might be related to https://bugs.mojang.com/browse/MC-61329
11:24 < Fenhl> the index error, that is
11:25 < Thinkofdeath> yeah anything built near the height limit causes it to crash on that chunk being sent
11:25 < Thinkofdeath> just tower up and relog to reproduce
11:26 < Dinnerbone> mrapple: it's all the same datastore
11:26 < Dinnerbone> It matters nothing if you used it to add en entity in the world or not
11:27 < WizardCM> Thinkofdeath: try raising the value by one in server.properties?
11:27 < Grum> md_5: hehe that is without compresssion?
11:28 <+md_5> Grum nope, this is stock standard Vanilla, = 256
11:28 <+md_5> read above for what Thinkofdeath is saying
11:28 <+md_5> just pillar to the top of the map
11:28 < Dinnerbone> mrapple: you will also now get a Component per user for a display name only in the tab list. It will not change sorting. Is this everything you'd need?
11:29 < Thinkofdeath> WizardCM: doesn't help
11:29 <+Amaranth> That means he can only do his custom display if he has 60 players on the server
11:29 <+Amaranth> Since his minigames are more like 20 players I suspect he won't be happy with it
11:31 < Dinnerbone> This is in addition to the custom Component he can display at the top
11:31 <+sadimusi> it won't take long until someone builds a service which distributes valid skin blobs just for this purpose :D
11:33 < WizardCM> lets by 3000 alts in preparation
11:33 < WizardCM> (not really)
11:33 < WizardCM> buy*
11:37 < mrapple> Dinnerbone: have I ever mentioned that I really do love you?
11:38 < mrapple> Dinnerbone: quick question, would you consider not showing skins on the tab list if it's v3 (offline mode)? not sure if that's the behavior already, just an idea
11:38 < eddyb> hey Dinnerbunny :)
11:39 < eddyb> "Chunks now use block states instead of metadata"
11:40  * eddyb hopes that means there's 16bit -> block states mappings now
11:40 < Thinkofdeath> mrapple: looks like that is how it works currently
11:40 < Thinkofdeath> eddyb: Pretty sure there is an array somewhere for that
11:41 < mrapple> ah good to hear. any insight on how the tab list scales with 60+ or 100+ players Thinkofdeath? i was going to work on upgrading the python protocol to investigate
11:41 < Thinkofdeath> mrapple: haven't tested yet
11:41 < mrapple> are you able to spin up a bunch of clients?
11:41 < Thinkofdeath> eddyb: something like array[id << 4 | data] in one of the classes
11:41 < eddyb> Thinkofdeath: well, that's in the code
11:42 < Thinkofdeath> mrapple: not really, could fake them
11:42 < mrapple> we usually use this to test lots of players, I'll probably update it sometime this week I guess https://github.com/YukonAppleGeek/spock
11:43 < WizardCM> ooooh
11:43 < WizardCM> i like this
11:43 < WizardCM> i should use it
11:44 < mrapple> here's a working branch https://github.com/nickelpro/spock/tree/patch-1
11:46 < mrapple> Dinnerbone: one other idea, I know the tab list max used to be configurable in server.properties with max players, any possibility of keeping that around? I'll investigate the scaling with large numbers of players later this week, but having some consistent maximum would be ideal I suppose
11:46 < Thinkofdeath> mrapple: bah can't fake it, having multiple players with the same skin blob hides the extras
11:46 < Thinkofdeath> and the skin icon
11:46 < mrapple> what?
11:46 < mrapple> just make new UUIDs
11:46 < Thinkofdeath> I am
11:46 < Thinkofdeath> its using the ones in the blob
11:47 < mrapple> how does offline mode work then...? am I missing something?
11:47 < Thinkofdeath> hmm maybe I am, this doesn't work in offline mode either.
11:47 < Thinkofdeath> might of messed up somewhere
11:47 < mrapple> how are you faking them? just multiple MC clients?
11:48 < Thinkofdeath> just faking the packets
11:53 < Thinkofdeath> mrapple: I was wrong about offline mode, steve icons
11:53 < Thinkofdeath> mrapple: also 60 players http://i.imgur.com/aEEOlbY.png
11:53 < mrapple> Thinkofdeath: sweet, what happens over 60?
11:54 < Thinkofdeath> mrapple: 200 http://i.imgur.com/SC62pJ8.png
11:54 < Thinkofdeath> so readable
11:54 < mrapple> yikes
11:55 < Dinnerbone> I'll try and cap it
11:55 < mrapple> Dinnerbone: a configurable cap or a hardset of 60 would be nice :P
11:56 < mrapple> Dinnerbone: would it be too much to ask to disable skins for offline mode (and possibly a way to disable ping as well)?
11:57 < Dinnerbone> I will look into it.
11:57 < mrapple> thank you sir, I appreciate it
11:57 <+md_5> hardset 60 imho
11:57 < Thinkofdeath> http://i.imgur.com/7AIPeyZ.png 2000 players, bungee's global ping is no longer an option lol
11:57 < mrapple> lol
11:57 < Spurlex> Looks good to me Thinkofdeath 
11:58 < Thinkofdeath> god the fps drop just from rendering that
11:58 < Dinnerbone> md_5: 60 might be bad in default res but good in higher
11:58 <+md_5> whatever you do, hardcode a cap
11:58 < mrapple> Dinnerbone: I believe this is 60 in default res, eh Thinkofdeath? http://i.imgur.com/aEEOlbY.png
11:59 < Thinkofdeath> yep
11:59 < mrapple> looks pretty good to me
11:59 < WizardCM> bah
11:59 < WizardCM> gg internet
11:59 < WizardCM> what did i miss?
11:59 < Thinkofdeath> although my GUI is different
11:59 < mrapple> dinnerbone being a dinnerbuddy
11:59 < Spurlex> dinnerbae
11:59 < Thinkofdeath> oh large == auto at this res anyway
12:00 < WizardCM> where's the channel logged to?
12:00 < Thinkofdeath> no one knows
12:00 < Fenhl> lol
12:00 < Thinkofdeath> we never found it
12:00 < WizardCM> xD
12:01 < dx> he's serious though
12:01 < Fenhl> well that would explain why it crashes on our Survival but not on Creative
12:01 < Fenhl> Survival has a tower up to y=255 near where I logged out before the update, Creative doesn't
12:02 < Thinkofdeath> Hopefully b has a fix for that
12:05 < Dinnerbone> Yeah it's fixed for b
12:05 < Thinkofdeath> \o/
12:05 < Fenhl> wheee
--- Log closed jeu. juil. 10 12:11:58 2014
--- Log opened jeu. juil. 10 12:18:38 2014
12:18 -!- Irssi: #mcdevs: Total of 136 nicks [1 ops, 0 halfops, 11 voices, 124 normal]
12:20 < WizardCM> yay i killed the channel :(
12:20 < Dinnerbone> Dammit WizardCM.
12:20 -!- Irssi: Join to #mcdevs was synced in 138 secs
12:20 < WizardCM> i do it so frequently
12:20 < WizardCM> it's like a plague
12:21 < Thinkofdeath> I'm sure we can find a new thing to start complaining about
12:21 < Thinkofdeath> oh how about the create inventory action packet, that one kills for creative servers
12:23 < Dinnerbone> The what does the who with the when now?
12:23 < WizardCM> Dinnerbone: all the things with everyone all the time
12:24 < Thinkofdeath> Dinnerbone: letting the client send raw unfiltered nbt :/
12:24 < Thinkofdeath> you said you wanted protocol suggestions :P
12:24 < Dinnerbone> :(
12:25 < Dinnerbone> It's not filtered but it is validated somewhat
12:25 < Dinnerbone> It can't get too big, I think
12:25 <+md_5> Thinkofdeath we can nag dinnerbone re: movement as well
12:25 < eddyb> https://twitter.com/TheMogMiner/status/484698708492451840 does this mean minecraft is not (yet) using VBOs?
12:25 <+md_5> Dinnerbone the limits are WAAAAYA to big
12:25 < Dinnerbone> Haha that's a deep engine change
12:25 < Thinkofdeath> Dinnerbone: Yeah but certain items/attributes crash the client
12:25 <+md_5> s/to/too
12:25 <+md_5> Dinnerbone creative mode should only allow id, amount, data
12:26 <+md_5> not arbitrary NBT
12:26 < yawkat> Aw no bot
12:26 < eddyb> I guess I didn't dive in deep enough into the client-side code
12:26 <+SpaceManiac> Haha, people in bukkitdev have taken to complaining the player speed limits are too small
12:26 < eddyb> s/in//
12:26 < yawkat> SpaceManiac: what
12:26 < yawkat> Why would they be too small? 
12:27 <+SpaceManiac> Some devs just want players to go really, really fast
12:27 < Fenhl> Dinnerbone: might want to mark https://bugs.mojang.com/browse/MC-61128 as resolved then
12:27 < Dinnerbone> Thanks, I didn't see the ticket
12:29 < Thinkofdeath> Dinnerbone: about movement: could you make it not tick the player, that would at least fix the fast health regen/eating issues
12:29 < Dinnerbone> No, sadly
12:29 < Dinnerbone> I've tried that many times
12:29 < Dinnerbone> The ticking stuff is too complex to fix still
12:29 < Thinkofdeath> I've tried too :/
12:31 < Fenhl> Dinnerbone: also the two I reported are probably fixed then, too
12:32 < eddyb> task '<main>' failed at 'called `Option::unwrap()` on a `None` value', /home/rustbuild/src/rust-buildbot/slave/nightly-linux/build/src/libcore/option.rs:265
12:32 < eddyb> okay, so something changed :D
12:40 < eddyb> wait, no, that was me being stupid
12:41 < Spurlex> Dinnerbone: Can you please make security questions on the MC Launcher if you login from multiple IP’s (Or a new computer or something)
12:41 < cindy_k> Thinkofdeath, for creative, do you mean how people some how manage to do this: http://i.imgur.com/8YlWJAk.png
12:41 < Thinkofdeath> yep
12:41 < cindy_k> A player gave me that. :/ luckily it doesn't seem to be crashing the server. 
12:42 < MrARM> Well, give?
12:42 < Thinkofdeath> We cap display names/lore in spigot now
12:42 < eddyb> Thinkofdeath: I don't see block states mappings anywhere
12:42 < Thinkofdeath> bukkit does filtering too
12:43 < eddyb> I have to wonder what they meant by "chunks use block state now" - is it something internal to the code and not affecting region files at all?
12:44 < WizardCM> cindy_k: wat
12:44 < WizardCM> dat potion though
12:44 < WizardCM> what does it give you when you drink it?
12:44 < Thinkofdeath> eddyb: ary.d
12:45 < Thinkofdeath> looks like its filled a runtime
12:47 < WizardCM> ...well THAT shouldn't happen in single player http://wizrd.tk/9povw8
12:48 < WizardCM> thank goodness for F3+A
12:51 < Thinkofdeath> eddyb: http://hastebin.com/kulubelamo.css
12:54 < WizardCM> fail http://minespy.net/api/3dfront/Dinnerbone/reflect
12:54 < eddyb> Thinkofdeath: where is that from O_O?
12:54 < Thinkofdeath> eddyb: I just generated it from that class
12:54 < eddyb> with what?
12:54 < Thinkofdeath> eddyb: http://hastebin.com/jumuhomexo.vala
12:55 < Fenhl> that looks like a potion you shouldn't drink
12:55 < eddyb> Thinkofdeath: haha
12:55 < eddyb> I see
12:56 < Thinkofdeath> just depend on the client/server and run that == full list :)
12:56 < eddyb> Thinkofdeath: thanks a lot :)
12:56 < Thinkofdeath> np :)
12:56 < eddyb> I wonder if I can redistribute *just* that list
12:57 < Dinnerbone> And IDs in memory are just id << 4 | data
12:58 < Thinkofdeath> yeah for some reason that took me longer than a care to admit to work out with the previous snapshot :P
12:58 < Thinkofdeath> *than I
12:59 < eddyb> Dinnerbone: | add << 12. and yeah, I wouldn't store id/data
12:59 < eddyb> I'm so tempted to just hack together some JSON loading code and render all the solid blocks in, uhm, a debug world, I guess
13:00 < Dinnerbone> Why not just, uh, load the debug world?
13:00 < Dinnerbone> It has every possible block in every possible state
13:01 < eddyb> Dinnerbone: this is a toy minecraft client I'm writing :)
13:01 < eddyb> I guess I wouldn't have to load the debug world, but just do the exact same thing
13:01 < eddyb> I have to rewrite the render code anyways, before I can play with it
13:02 < Dinnerbone> The list that Thinkofdeath gave you is not the complete list of every possible block in the game
13:02 < Dinnerbone> It's the ones that we're currently able to store, given legacy compatibility
13:02 < Dinnerbone> At runtime we flesh blocks out to their real state
13:03 < eddyb> I remember sadimusi showing me some other classes, with, like, redstone states
13:03 <+sadimusi> this one? https://gist.github.com/sadimusi/f56d67b385a5ba5b8da6#file-cep-java
13:04 < eddyb> yeah, that
13:04 < eddyb> doesn't look like "block state" to me
13:05 <+sadimusi> those might just be the names of the models or something
13:05 < eddyb> but isn't redstone rendered from one texture? or two, I guess
13:06 < eddyb> Dinnerbone: btw, I see "leaves2". will that turn into leaves with more than just 16 block states?
13:07 <+sadimusi> eddyb: I can't find those two classes in the new snapshot anymore, it might just have been some legacy stuff
13:08 < eddyb> and it would be nice, whenever worlds start storing mappings, to have everything packed together (there's so many blocks that don't use 15 of their "data values")
13:08 < eddyb> sadimusi: oh. what's in that "ary" Thinkofdeath found, btw?
13:08 < Dinnerbone> <eddyb> and it would be nice, whenever worlds start storing mappings, to have everything packed together (there's so many blocks that don't use 15 of their "data values")
13:08 <+sadimusi> ary looks like the block class
13:08 < Dinnerbone> Data doesn't exist anymore.
13:08 < Dinnerbone> Blocks don't waste data values. They don't have data values to waste.
13:09 < eddyb> Dinnerbone: well, I know that
13:09 < Dinnerbone> There's nothing to pack. They're using what they're using.
13:09 < WizardCM> i am SICK of crappy internet :(
13:09 < eddyb> but there is no 0xd1-0xdf, just 0xd0, gravel
13:19 < WizardCM> i'll just leave this to fill the silence http://www.questionablecontent.net/view.php?comic=2743
13:32 < eddyb> well, that was an awkward pause
13:32 < WizardCM> quite
13:32 < WizardCM> my filler didn't help either
13:32 < eddyb> I meant to give a few more examples
13:33 < WizardCM> fascinating
13:33 < WizardCM> tell me more
13:33 < WizardCM> :P
13:34 < WizardCM> ooooh, i just remembered something i'd love changed in minecraft
13:35 < WizardCM> the client-side chat limit
13:35 < eddyb> and mention that I wouldn't really want random mod blocks occupying those unused slots
13:35 < WizardCM> 100 characters isn't enough for a sentence sometimes
13:36 < WizardCM> the debug world, is that fully generated on load, or does it require exploration?
13:37 < Dinnerbone> Neither, technically.
13:37 < Dinnerbone> It doesn't store all the blocks to disk, they're calculated at runtime because it's the debug level type
13:37 < WizardCM> ah
13:37 < WizardCM> so copying to 1.7.10 wouldn't work?
13:37 < Dinnerbone> Most of the blockstates in the world can't be serialized to disk
13:37 < Dinnerbone> No
13:38 < WizardCM> alright
13:38 < WizardCM> makes sense
13:38 < WizardCM> what a large collection of flames
13:38 < Dinnerbone> And each one unique!
13:39 < WizardCM> i was about to say f3 doesn't work on blocks in spectator while using debug, then remembered that fire doesn't have collision so it can't be checked
13:39 < WizardCM> xD
13:39 < eddyb> wait, so it's not even a world
13:39 < eddyb> but a world provider
13:40 < eddyb> Dinnerbone: I'm intrigued. what is such an unserializable block state?
13:40 < Thinkofdeath> fences
13:40 < WizardCM> good to see the wool colour and model names finally match
13:41 < eddyb> Thinkofdeath: as in, the various visual states that depend on neighboring blocks and don't get saved by themselves?
13:41 < Thinkofdeath> yep
13:43 < WizardCM> can't wait to start making a texutre pack in this world
13:43 < WizardCM> it's going to be so much fun
13:44 < eddyb> Thinkofdeath: I'm more interested in whatever is and will be serialized
13:45 < Dinnerbone> Eventually: All the block states.
13:45 < Dinnerbone> Right now: only the ones that are currently serialized.
13:45 < eddyb> I don't expect declarative rules for all those neighbor-dependent block (render?) states
13:45 < Dinnerbone> Well you should
13:46 < Dinnerbone> They are their own block state
13:46 < Dinnerbone> They should be serialized so that things don't have to keep looking around it to figure out what it really is
13:46 < eddyb> Dinnerbone: no, I mean, the update logic
13:47 < eddyb> all the redstone_wire "render" states depend on the placement of adjacent (and diagonal, I guess) redstone_wire blocks
13:47 < Dinnerbone> Yes
13:48 < eddyb> you'd have to JIT the code for handling that if you were to describe the rules in JSON
13:48 < Dinnerbone> I think you've completely missed the idea of block states.
13:48 < Dinnerbone> :D
13:48 < eddyb> I'm talking about rules, not states :P
13:48 < Dinnerbone> You can't describe their rules that easily
13:49 < Dinnerbone> (And they aren't just visual states. There's no difference between a block state for rendering or for existing in a world. It's the same thing.)
13:49 < WizardCM> Dinnerbone: out of curiosity, what would need to be done to allow the ability for players to send messages longer than 100 characters?
13:49 < Dinnerbone> WizardCM: me increasing this constant from 100 to x
13:49 < eddyb> Dinnerbone: contextually unambiguous states :P?
13:49 < WizardCM> funny story, those two messages we both just sent are both 140~ characters
13:50 < WizardCM> we wouldn't be able to send them if this was minecraft
13:50 < WizardCM> xD
13:50 < Dinnerbone> "WizardCM: me increasing this constant from 100 to x".length == 51
13:50 < WizardCM> no i mean right before that :P
13:51 < eddyb> I wonder if we'll be able to keep weird stair configurations in the future
13:51 < Dinnerbone> eddyb: there is never any ambiguity. There is no need for context.
13:51 < eddyb> like, the ones with just a corner showing
13:51 < eddyb> without needing the right configuration around the block in question
13:51 < WizardCM> probably only with set commands and such
13:51 < Dinnerbone> A blockstate is "red glass pane connecting on east and west", not "red glass pane and figure the rest out for yourself"
13:51 < WizardCM> xD
13:52 < Dinnerbone> It will be stored in the world as "a red glass pane connecting on east and west"
13:52 < Dinnerbone> And when things ask it what it is, it'll say "a red glass pane connecting on east and west"
13:52 <+SpaceManiac> Oh no, less convenient "the client figures it out for me"!
13:52 < WizardCM> ^ which is what it does in 1.7 and earlier, right?
13:52 < Dinnerbone> When something changes next to it, it'll re-evaluate that and realize it's now "a red glass pane connecting on east, west and north"
13:53 < eddyb> Dinnerbone: makes complete sense :)
13:53 < eddyb> some things are possible now, at least in the absence of block updates
13:53 < Dinnerbone> Right now every time you ask it for its boundingbox (happens lots and lots), or to render it, it has to check every block around it to find out what it really is
13:53 < eddyb> Dinnerbone: whoa, no caching?
13:53 < Dinnerbone> Cache where?
13:54 < Dinnerbone> What is there to cache to?
13:54 < eddyb> hah, very good point
13:54 < yawkat> well now it caches that state so it doesn't have to reeval, right?
13:54 < Dinnerbone> SpaceManiac: the server still had to figure it out currently
13:54 < Dinnerbone> Not just the client.
13:54 < eddyb> without block states, you would need like an Object per block
13:55 < eddyb> which is a nono
13:55 < WizardCM> so i'm guessing that torch-on-bottom-slab thing was a bug then?
13:57 < eddyb> Dinnerbone: sorry for sounding so obtuse :D. this is all very awesome and I'm glad I can soon be lazy and render most things in a vanilla world without listing blocks or items anywhere :)
13:58 < WizardCM> ^ very much this
13:58 < WizardCM> the questions are because we are excited
13:58 < WizardCM> also that was an unexpected crash http://wizrd.tk/q78pja
13:59 < eddyb> I do wonder if there isn't too much NBT stringly-typed data and how that gets handled at runtime, but it shouldn't show up unless you have several hundreds of entities
14:00 < Thinkofdeath> Dinnerbone: Would you consider the model format stable now? Kinda want to start using it :)
14:00 < WizardCM> Thinkofdeath: wasn't that stated in a tweet recently?
14:01  * Thinkofdeath doesn't follow tweets 
14:01 < Dinnerbone> Yes, as we said on the blog post for the snapshot and that the comment saying "it's unstable" was removed, it's now considered stable
14:01 < Thinkofdeath> \o/
14:01 < Thinkofdeath> I can work on remove things like this then https://github.com/thinkofdeath/ThinkMap/blob/master/core/shared/src/main/java/uk/co/thinkofdeath/thinkcraft/shared/block/blocks/BlockFence.java#L73-L113
14:02 < WizardCM> i don't remember when i remember reading that it's finalised
14:02 < WizardCM> oh right
14:02 < WizardCM> yeah
14:02 < WizardCM> blogpost
14:02 < WizardCM> "The new block model format for resource packs can be considered stable and ready to use."
14:02 < WizardCM> silly Thinkofdeath not reading :P
14:03  * Thinkofdeath may have just decompiled the snapshot and forgot to read the post
14:04 < Dinnerbone> You didn't see that all the model .jsons got modified to remove the "warning this is not final"? :p
14:04 < yawkat> Thinkofdeath is a robot that hasn't mastered reading English yet
14:04 < WizardCM> you could also follow https://twitter.com/WizardCM/lists/minecraft-dev
14:04 < WizardCM> or, 'subscribe' to it
14:04 < WizardCM> i just have a column in tweetdeck for it
14:04 < Thinkofdeath> Dinnerbone: I have actually looked at the models yet either, I just know that they exist :P
14:04 < Thinkofdeath> *haven't
14:05 < Thinkofdeath> tbh I should have looked there first
14:07 < WizardCM> hmm, i need some tea
14:07 < WizardCM> time to also check if there are any cool new irc clients for windows phone
14:16 < WizardCM> answer to that is no, bummer
15:55 < Xor_> hmm, apparently I missed the protocol debacle
15:55 < Xor_> le sad
16:07 < WizardCM> Xor_: you can still contribute
16:07 < WizardCM> just people aren't watching the channel as heavily
16:07 < Xor_> ok
16:07  * Xor_ tries to think of the worst thing in teh protocol
16:07 < Xor_> mostly annoyed because the fun part happened while I slept
16:08 < Dinnerbone> The part where we all joined the circus?
16:08 < Dinnerbone> That was pretty crazy.
16:08 < Dinnerbone> I'm going to treasure that night forever. <3
16:08 < WizardCM> eh, circus never seemed very interesting to me tbh
16:08 < Xor_> oh Dinnerbone
16:09 < Xor_> shame next minecon is in europe, won't get to hug you again for a while
16:09 < WizardCM> next one should be in australia
16:09 < WizardCM> just sayin'
16:09 < Xor_> Dinnerbone actually, why do you drop air-right-clicks if there's no item server side?
16:10 < Xor_> had to make a hack to get around that bit
16:11 < Xor_> TIL `^` is allowed in nicknames
16:15 < Spurlex> #Minecon4AustraliaPlz
16:17 < Xor_> well, if it's un australia, it will force md_5 to turn up, but I live on the other side of the world =(
16:17 < Xor_> florida was so convinient
16:18 <+sadimusi> Is there anything interesting going on there? From what I've seen it looks like a bunch of 13 year old asking stupid questions
16:18 < WizardCM> ...
16:18 < WizardCM> where?
16:18 < Xor_> I don't even
16:19 < WizardCM> oh at minecon
16:19 < WizardCM> yeah panel Q&As tend to be like that
16:19 <+sadimusi> yeah, not australia in general :D
16:19 < WizardCM> sometimes you get some really good questions
16:19 < WizardCM> but usually it's something like "y u so slow dev, i want backpacks and rainbows"
16:20 < stuntguy3000> I want some wam.
16:20 < stuntguy3000> Or however that question went
16:20 < Xor_> minecon was interesting
16:20 < WizardCM> actually, my guess is the next minecon will have a lot of EULA questions
16:20 < stuntguy3000> Minecon is a waste of time.
16:20 < WizardCM> "y u hate community and servers, you're destroying minecraft"
16:20 < Xor_> but I hung out mostly around the spigot booth
16:20 < stuntguy3000> Each panel is 45 of googleable questions
16:20 < stuntguy3000> 45 minutes*
16:20 < WizardCM> mhm
16:20 < Xor_> idc about panels
16:20 < WizardCM> ^
16:21 < Xor_> I just like running into people form IRC
16:21 < WizardCM> you go for the people
16:21 < stuntguy3000> I can't go, so its panels or nothing :P
16:21 < Xor_> and getting to hug db
16:24 < Spurlex> I just want australia so I can go to it
16:24 < Spurlex> I will fly from New Zealand to australia >:)
16:26 < Xor_> and now I realize redstonetorch_off is not displayed by the client.
16:26 < Xor_> oh, protocol
16:26  * Xor_ stabity stab
16:30 < cindy_k> now xor_ you need things to keep you busy
16:31 < cindy_k> if it was easy you would be bored
16:31 < Xor_> cindy_k what are you doing in here?
16:31 < Xor_> dammit she found me
16:31 < cindy_k> sharing pretty pictures
16:31 < Xor_> redstone lamps are not available in on form in inventory? q.q
16:32 < Xor_> but I want to abuse it for an idicator!
16:32 < cindy_k> you can w/e them in that way
16:32 < cindy_k> oh you mean for a menu
16:34 < Xor_> cindy_k q.q
16:34 < Xor_> using glowstone for on-state instead
16:36 < WizardCM> woo 14w28b is out
16:36 < WizardCM> whoa
16:36 < WizardCM> so many bugs fixed
16:36 < Xor_> time to alert Thinkofdeath
16:36 < WizardCM> Dinnerbone, Grum: you work faast sometimes
16:37 < Xor_> WizardCM they're devs, like all of us: black boxes that convert alcohol/coffee/manga into software
16:37 < WizardCM> manga
16:37 < WizardCM> xD
16:37 < WizardCM> very nice
16:37 < Xor_> that's libraryaddict
16:37 < Xor_> he's a special kind of dev
16:37 < WizardCM> ah
16:37 < WizardCM> xD
16:37 < Dinnerbone> None of the above for me :)
16:38 < Xor_> Dinnerbone I do water =p
16:38 < WizardCM> yeah i don't drink either
16:38 < WizardCM> nor do i like coffee or manga
16:38 < Xor_> is <insert fuel here> -> software better?
16:39 < Spurlex> Dinnerbone can I have your furni on Habbo
16:39 < Spurlex> plz
16:39 < WizardCM> xD
16:39 < Xor_> Spurlex well, I knew dinnerphone had phans
16:39 < dx> uh guys please don't ping him for dumb stuff
16:40 < Spurlex> are you saying furni on habbo is dumb
16:40 < Spurlex> :'(
16:40 < dx> yes
16:40 < Xor_> kek
16:40 < Xor_> he spoke, the masses answered
16:40 < Thinkofdeath> ooh new packet
16:40  * Thinkofdeath has no idea what it does
16:40 < Xor_> Thinkofdeath I never like it when you're excited about the protocol
16:40 < dx> shiny new packet
16:40 < WizardCM> Thinkofdeath: how are you so fast
16:40 < shoghicp> this reminds me of mcpe updates :D
16:40 < Xor_> that usually ends with several burning JVMs
16:41 < cindy_k> shh xor_ it keeps him busy
16:41 < shoghicp> at least we have the name there ;)
16:41 < Thinkofdeath> WizardCM: its mostly automated now :)
16:41 < Xor_> WizardCM he has a fancy remapper
16:41 < Dinnerbone> You can have my furni for 5 bobba
16:41 < WizardCM> xD
16:41 < Spurlex> :O
16:41 < Xor_> DInnerbone you moving again, or is this the stuff from your old place?
16:41 < Spurlex> I trade you 2 ducks for HC sofa plz
16:42 < Xor_> or is this something completely unrelated
16:42 < Dinnerbone> Unrelated babble.
16:42  * Xor_ is baffled
16:42 < WizardCM> Xor_: http://habbo.com
16:43 < WizardCM> i find it funny that they have an iPad version
16:43 < Xor_> I see...
16:43  * Xor_ goes back to dicking around with ProtocolLib
16:49 < Thinkofdeath> hmm something changed with the tab list
16:49 < WizardCM> sneaky, sneaky
16:50 < WizardCM> better spacing between face and name, maybe? :P
16:50 < Spurlex> So with the tab list its changing in 1.8 right? So you guys advise that I shouldn’t bother making mine look fancy?
16:50 < Xor_> Thinkofdeath you were here when apple asked grummy for that one thing
16:50 < WizardCM> Spurlex: depends, if you're on bukkit.. could be MANY months before a 1.8 version of that is out
16:50 < WizardCM> and so many changes
16:50 < WizardCM> i'd say it's safe to customise it
16:50 < yawkat> Thinkofdeath: my bet is tab list size pcket
16:50 < yawkat> s/pc/pac
16:53 < Spurlex> Also anyone know why you can see glass above particles such as this http://file.spurlex.com/9UIQ/1YdK0xtZ
16:53 < Xor_> more render shit
16:53 < WizardCM> isn't that how particles have always been?
16:53 < WizardCM> rendered incorrectly?
16:53 < Xor_> I pitty the fools that have to fix that
16:53 < Xor_> minecraft render code is the most horrible thing ever
16:54 < Xor_> I expect that if anyone manages to fix it it'll be ryan
16:54 < Spurlex> Ookay I never noticed. Glad its not just me ;D
16:55 < WizardCM> Thinkofdeath: how goes exploration
16:55 < Thinkofdeath> working on it, hitting an issue atm
16:56 < Thinkofdeath> "fieldSize is too long! Length is 65280, but maximum is 30" no idea where that is from
16:59 < Thinkofdeath> well I know this much, the tab list allows for "display names"
17:09 < WizardCM> maximum is 30.. prefix and name?
17:09 < WizardCM> or is that 30 just for display name
17:34 < WizardCM> Thinkofdeath: any new stuff or still stuck?
17:36 < Thinkofdeath> http://wiki.vg/Pre-release_protocol#14w28b working on the new packet now
17:40 < Thinkofdeath> And the new packet http://i.imgur.com/KtwMMxN.png
17:41 < Thinkofdeath> http://wiki.vg/Pre-release_protocol#Player_List_Header.2FFooter
17:43  * Thinkofdeath hugs Dinnerbone and Grum 
17:43 < Xor_> s/hug/huggle
17:43 < Spurlex> Interesting.
17:46 < Aragas> lol. nice
17:48  * Dinnerbone cuddles Thinkofdeath
17:48 < Thinkofdeath> :D
17:54 < WizardCM> oh that's pretty sweet
17:54 < WizardCM> even for a v1
18:05 < Drainedsoul> looks like a bunch more conditional parses
18:10 < WizardCM> Thinkofdeath: is it width limited?
18:10 < WizardCM> (probably based on player list, right?)
18:11 < Thinkofdeath> Haven't checked, I will once I've fixed something else
18:29 < Thinkofdeath> WizardCM: no limit
18:30 < WizardCM> o.O
18:30 < Thinkofdeath> it wraps onto a second line if needed
18:30 < WizardCM> but the width is based on the actual list of players, yeah?
18:30 < WizardCM> or does it stretch even if there's a max limit of 1 or 2
18:30 < Thinkofdeath> WizardCM: http://i.imgur.com/aG1aKkc.png
18:31 < WizardCM> ahhhh
18:31 < WizardCM> i like
18:31 < WizardCM> very nice
18:31 < HisShadow> Thinkofdeath: so much cake
19:00 < mrapple> Thinkofdeath: how do things look with 60 or 100 players?
19:02 < Thinkofdeath> mrapple: still bad
19:02 < Thinkofdeath> mrapple: http://i.imgur.com/TWxi267.png
19:03 < mrapple> damn :(
19:04 < Thinkofdeath> looks like steve is gone in offline mode
19:04 < Thinkofdeath> *steve icon
19:06 < mrapple> nice
19:06 < mrapple> Dinnerbone: any thoughts about a hardcoded limit of 60 for the next snapshot?
19:07 < Dinnerbone> I will look into it but not this week
19:12 < mrapple> ok, would you like me to file a bug report so it doesn't get lost?
19:27 <+Amaranth> Oh yeah, the limit is client side now so I can't fix it
19:27 <+Amaranth> I think that was my first contribution to CraftBukkit
19:30 < Xor_> wonder whatever happened to shaders
19:30 < Xor_> I was hoping we'd get the ability to create them in resource packs and set them on the clicnet
19:31 <+Amaranth> You can add them in resource packs, can't you?
19:31 < Xor_> don't think so
19:31 < Xor_> ping one of the devs and ask them =p
19:31 <+Amaranth> But you have to client the button to randomly choose one
19:32 <+Amaranth> And it's "just" post processing effects
19:32 < Xor_> Amaranth they said that button was a demo anyways
19:32 <+Amaranth> Sure
19:32 <+Amaranth> Some of the effects actually get used in 1.8 iirc
19:32 < Xor_> and erik said that they would add them to resource packs because "they're a resource"
19:32 <+Amaranth> When you're spectating
19:32 < Xor_> Amaranth yeak, spec mode on mods
19:32 < Xor_> s/mods/mobs
19:33 <+Amaranth> Which means people with old computers have an advantage there
19:33 < Xor_> I also want texture metadata that regexes name/lore
19:33 < tyteen4a03> TkTech, just noticed your tinycrawler; y u no scrapy
19:33 <+Amaranth> Xor_: And I want a pony
19:33 < Xor_> Amaranth we already have ponies
19:33  * Xor_ glares at that one mocreatures guy
19:34 <+Amaranth> You're trying to backdoor custom items at the point they're getting the world and block situation sorted to just support having them
19:34 < Xor_> Amaranth can't wait, lol
19:35 < Xor_> Amaranth I was actually kind of annoyed when a player told me it was a thing and forgot to mention it was an mcpatcher feature
19:35 < Xor_> heads rolled.
19:35 <+Amaranth> That seems like an awful way to pull that off
19:35 <+Amaranth> Which means Mojang will probably do it ;)
19:35 < Xor_> Amaranth it is
19:35 < Xor_> but it's the only way right now
19:36 < Xor_> take COD-type minigames
19:36 <+Amaranth> No
19:36 < Xor_> well, any minigame with custom items
19:36 < Xor_> they can give up vanilla items
19:36 < Xor_> survival people like me can't
19:37 < Xor_> I have a whole mess of custom items based on that
19:37 < Xor_> it's quite the mess
19:40 < Xor_> Amaranth you looked at the model forma yet?
19:40 <+Amaranth> Not really
19:40 < Xor_> meh. I hope they make that obj converter
19:40 < Xor_> might be annoying though, to approximate the cuboids and rotations
21:24 < Aragasas> 123
21:25 < Aragasas> Pardon me. Check.
22:44 < barneygale> "If the varint is 0 then the packet isn't compressed otherwise its the size of the uncompressed packet."
22:44 < barneygale> should this say "compressed"?
22:45 < yawkat> probably not
22:45 < yawkat> compressed packet size is probably the one handled by the splitter
22:45 < barneygale> the splitter?
22:45 < yawkat> that packet header thing
22:46 < yawkat> i think it was added in .6 or .7 or something
22:46 < yawkat> with netty
22:46 < barneygale> Oh, I had assumed the packet header was compressed too
22:46 < barneygale> hm
22:47 < barneygale> to wireshark!
23:19 < Thinkofdeath> barneygale: its Packet length, Data Length, Data.
23:19 < Thinkofdeath> If Data Length is zero then its uncompressed otherwise its the size of the packet once you inflate it
23:20 < Thinkofdeath> I should reword the wiki
23:23 < barneygale> Why is it necessary to send the length of uncompressed data?
23:23 < barneygale> (I assume packet length is sending length of compressed data)
23:23 < mrapple> the packet length includes the data length
23:23 < mrapple> it probably isn't necessary but it standardizes the format
23:24 < Thinkofdeath> It allows minecraft to allocate the byte array for the packet instead of doing chunk by chunk
23:24 < barneygale> You could just put a boolean before the old packet length field for compressed/uncompressed
23:25 < mrapple> no one should be using uncompressed
23:25 < barneygale> the length field gives the length of the raw data that follows
23:25 < mrapple> Thinkofdeath is just documenting the behavior
23:25 < barneygale> which can be compressed or uncompressed
23:25 < barneygale> I know
23:25 < barneygale> I'm not saying he implemented it :P
23:25 < Thinkofdeath> mrapple: uncompressed is still used for small packets
23:25 < barneygale> I'm just wondering about mojang's rationale
23:25 < mrapple> it is?
23:25 < barneygale> Yeah
23:25 < Thinkofdeath> less than 256 bytes by default
23:25 < mrapple> I thought all packets were compressed... nvm then
23:26 < barneygale> that's what the new theshold setting in server.properties is
23:26 < mrapple> is there a threshold and a compression level setting?
23:26 < Thinkofdeath> just threshold
23:26 < mrapple> interesting
23:27 < Thinkofdeath> http://wiki.vg/Pre-release_protocol#Packet_format Cleared up a bit
23:28 < barneygale> Hm, the packet format changes after you get the packet?
23:28 < barneygale> :/
23:28 < Thinkofdeath> yep
23:28 < barneygale> fun
23:29 < Thinkofdeath> http://hastebin.com/vavaveloyo.avrasm <- Why its the size of the uncompressed packet
23:29 < Aragasas> I'm sorry. You're talking about chunk compression?
23:29 < Thinkofdeath> only needs one inflate call
23:29 < Thinkofdeath> Aragasas: nope chunks aren't compressed anymore
23:29 < Thinkofdeath> the whole protocol is
23:29 < Thinkofdeath> (over 256 bytes)
23:30 < Aragasas> lol. New hardcore level.
23:30 < Xor_> compress all the things
23:30 < Dinnerbone> Only the things worth compressing :)
23:30  * Xor_ compresses Thinkofdeath using tar
23:30 < Xor_> Dinnerbone I'll compress your files!
23:30 < Xor_> hmm, doesn't sound as funny in chat
23:31 < Dinnerbone> Thinkofdeath, you should note on the wiki about the disconnects for compressing something that shouldn't be
23:31 < Aragasas> I'm so heavy. Someone, compress me!
23:31 < Thinkofdeath> oh right
23:31 < Xor_> Dinnerbone you guys planning on allowing something other gzip (configurable)?
23:31 < Thinkofdeath> its zlib (I was wrong before)
23:31 < Xor_> if you don't I'll do it anyways, soo.
23:31 < Xor_> Thinkofdeath boo
23:32 < Dinnerbone> Also that compression can be disabled entirely by just not sending the packet (or later sending it again with -1 as a value)
23:32 < Thinkofdeath> oh its -1, I thought it was 0 :P
23:32 < Aragasas> that's good. Debuggin' will be easy
23:32  * Thinkofdeath changes some things
23:33 < Dinnerbone> It's a threshold; 0 would mean "compress everything, it all meets the requirements"
23:33 < Dinnerbone> (Which would be a very stupid thing to do, but I'm sure somebody will.)
23:34 < Aragasas> What about compression level? About 10% win in summary?
23:34 < Dinnerbone> How do you mean?
23:35 < Dinnerbone> Compression level is up to the client/server themselves and is not communicated
23:35 < Aragasas> I mean, how much compressed packets will be lighter in total
23:36 < Dinnerbone> Well that's not an easy question
23:36 < Aragasas> We're doing that for network load reduction , right?
23:36 < Dinnerbone> The bulk chunk packet goes from like 800KB to 30KB on average
23:37 < Dinnerbone> Other misc packets usually 70% compression on average I guess but it's really too hard to say without tuning and proper measuring
23:38 < Xor_> Dinnerbone you can always just unit-test that
23:38 < Xor_> though, some might be mroe annoying to unit-test
23:38 < Xor_> then again, some other profiler might be better
23:39 < Dinnerbone> Hard to simulate something that can change drastically between made-up data and real world data
23:39 < Xor_> Dinnerbone quite.
23:39 < Xor_> unless you guys either have huge test cases or a good random-test-case generator it won't be easy to measure
23:40 < Xor_> could just brute-force it and do an actual stress test on a server, though that's just inefficient
23:41 < Aragasas> Impressive performance then. I question myself if current mobile devices could handle compressing/decompressing without heavy performance loss
23:42 < Xor_> wouldn't be too terrible if the operations were async
23:42 < Xor_> speaking of async, I'm guessing threaded worlds are postponed for 1.9 unless they got slipped in under my nose?
23:42 < Aragasas> Connecting to server from android/ios/wp would be fun
23:43 < Xor_> Aragasas don't give Thinkofdeath ideas
23:43 < Aragasas> Thats my idea!!!!!
23:43 < Xor_> his mapviewer already works at 60 fps on mobile
23:43 < Thinkofdeath> * when on a flat map
23:43 < Thinkofdeath> * on a view distance of 4
23:43 < Xor_> Thinkofdeath oh, now you bring in the fine print
23:44 < Thinkofdeath> pretty sure I said that at the time :P
23:44 < Xor_> did not
23:44 < Xor_> well
23:44 < Xor_> the logs will tell
23:45 < Aragasas> i wanna launch my library on a mobile device. Are there any mobile-minecraft server apps?
23:45 < Xor_> what
23:45 < Aragasas> Hm. Pocket edition works good. It could render vanilla chunks, yes?
23:46 < Xor_> they have disparate block tables
23:47 < Aragasas> I mean, apps that connects to minecraft servers. Better look myself then :D
23:47 < Xor_> I want to say "PE would be more popular if it could connect to the PC server" but then that's a null argument given all the spam support gets on twitter about PE 0.9.0
23:47 < Aragasas> by myself*
23:48 < barneygale> Why is there a set compression packet in both login and play mode?
23:48 < Dinnerbone> So it can be changed at both points
23:48 < barneygale> Why would it change while the server is running?
23:48 < Dinnerbone> In vanilla it does not
23:48 < barneygale> OK
23:48 < barneygale> That figures =D
23:48 < Dinnerbone> Servers might want to lessen load dynamically or something, that's up to them!
23:49 < barneygale> fair enough!
23:49 < Dinnerbone> Makes for good testing between various levels too
23:49 < Xor_> and then there's people that say mojang never adds stuff for us devs. kek.
23:49 < Aragasas> :D
23:49  * Xor_ will not name names
23:49  * Thinkofdeath walks away whistling
23:50 < Xor_> then again, they just did the most painful refactor I can imagine for the new block format
23:50 < Aragasas> Wait what
23:51 < Aragasas> What new block format? Anvil?
23:51 < Aragasas> oh, thats map
23:51 < Dinnerbone> Anvil? Woah. Blast from the past.
23:51 < Dinnerbone> That's a good couple of years old now, isn't it?
23:51 < Xor_> kek, I remember when Anvil came out
23:51 < Aragasas> Whaaat
23:51 < Thinkofdeath> McRegion!
23:51 < Xor_> Thinkofdeath stfu
23:51 <+AndrewPH> anvil is already a few years old?
23:51 < Xor_> we shall not speak of that
23:51 <+AndrewPH> jesus
23:51 <+AndrewPH> time flies
23:51 < Aragasas> Replacement for anvil?
23:51 < Xor_> in 1.9 so they say
23:52 < Dinnerbone> 1.9 will likely be the replacement level format, but that's by the by
23:52 < Aragasas> Any info available? wiki.vg?
23:52 < Xor_> I can bet you can't wait to dig into the voodooland that is entity meshes.
23:53 < Xor_> well, that's more erik and ryan, no?
23:53 < Aragasas> Mmmmm... Hardcore stuff
23:53 < Xor_> last time I saw entity model code was in 1.1
23:53 < Xor_> (which is when I told the client I would never talk to it again)
23:54 < mrapple> Dinnerbone: are you planning on continuing to do behind the scenes work/bug fixes for the rest of the snapshots (until the pre release) or will new features resume in a few snapshots?
23:54 < Xor_> mrapple you read the blog posts no?
23:54 < Xor_> they said they're in the last version cycle
23:54 < Thinkofdeath> No one here does :)
23:54 < Xor_> Thinkofdeath they're sometimes useful
23:55 < Xor_> such as bugfixes
23:55 < Thinkofdeath> Yeah discovered that today 
23:55 < mrapple> ah I see
23:55 < mrapple> so probably no new features
23:55 < mrapple> or at least, not large ones
23:55 < Xor_> mrapple we have enough stuff to break for now, eh?
23:55 < mrapple> :P
23:55  * Dinnerbone is running a count of how many questions in this channel today could've been answered by the blog posts
23:56 < Xor_> Dinnerbone might want to use BigInteger for that
23:56 < Aaron1011> :)
23:56 < yawkat> Xor_: watch for memory
23:56 < Aragasas> Blog? Never heard about that stuff.
23:56 < mrapple> +1 from me, sorry ;(
23:56 < Thinkofdeath> Don't worry, 10 of them are from me :P
23:56 < Xor_> mrapple so tell me, what evil are you going to do with the new bunnies?
23:56 < Xor_> I've already got a few horrid ideas
23:57 < Aaron1011> I'm looking at the pre-release protocol, and it looks like it's using far, normal, short, etc, instead of a number of chunks?
23:57 < Aaron1011> Does the client choose how much to render?
23:57 < Xor_> most of them involve reflection and the pathfinder
23:57 < Aragasas> This new version will be a hell of a ride for minecraft implementations :DDD
23:57 < Xor_> 1.7 was fun, wasn't it?
23:58 < Xor_> I kind of feel bad because us spigot people don't need to do very much
23:58 < Thinkofdeath> This update is easy (protocol wise) compared to 1.7
23:58 < Thinkofdeath> internally, no so much I imagine
23:58 < Xor_> Thinkofdeath yeah, but a ton of APIs will break, namely the ones erik deprecated
23:58 < Xor_> well, some
23:58 < Aragasas> Yep. But worth it. OPTIMIZATION
23:59 < Xor_> so much code calls block.setData(), because bukkit provides no alternative
23:59 < Aragasas> VarInts
23:59 < Xor_> VarBigInteger
23:59 < Thinkofdeath> oh that reminds me, I have things to fix now that display names are in the tab list
23:59 < Aragasas> What what? Lol
23:59 < Thinkofdeath> Xor_: its easy to go id + data to block state
23:59 < Xor_> Thinkofdeath please tell me you're not going to try to make spigot support every single version
--- Day changed ven. juil. 11 2014
00:00 < Thinkofdeath> hell no
00:00 < Xor_> Thinkofdeath it's the refactoring that's annoying
00:00 < yawkat> Xor_: one internet on spigot supporting minecraft classic by tomorrow morning
00:01 < Xor_> yawkat nice knowing you
00:01 < Aragasas> Spigot is interesing. They're makin' a crossgame voxelsandbox, yea?
00:01 < Dinnerbone> Aaron1011, that is a mistake on the wiki, the pre-release protocol sends number of chunks not 0/1/2/3
00:01 < Aaron1011> Ah, I thought so :)
00:01 < Xor_> no we're not
00:02 < Aaron1011> I'll update it
00:02 < Xor_> we're just a fork of craftbukkit that doesn't explode 1t 150 players
00:02 < Thinkofdeath> oh I never updated that packet
00:02 < Thinkofdeath> whoops
00:02 < Aragasas> Thats impressive
00:02 < Xor_> that also does cross-protocol support because Thinkofdeath is a masochist
00:03 < Xor_> I should spam more PRs, someday I'll get write access to master.
00:05 < Xor_> Aragasas oh, and what you're thinking of is spout, whose client sucks.
00:05 < Aaron1011> Spout is still on 1.6, I believe
00:05 < Aragasas> Pardon me. Sounds a bit similar
00:05 < Aaron1011> Spigot and spoutcraft are completely different
00:06 < Xor_> oh, and then there's cauldron, which is spigot and forge. together. in one jar.
00:06 < TkTech> tyteen4a03: I do use scrapy, we have hundreds of scrapy spiders.
00:06 < Aaron1011> Spoutcraft is a server plugin + client side mod
00:06 < Aaron1011> Spigot is a lot of improvements to CraftBukkit, many having to do with performance
00:06 < TkTech> tyteen4a03: tinycrawler is for a different class of crawling and to expirement in replacing the IO core with something more flexible
00:07 < TkTech> Scrapy is also *slow*, it's not suitable to live-price crawling
00:07 < Xor_> Aaron1011 and a few fun APIs I wrote
00:09 < Aaron1011> I've been wondering - how is the laser for the Guardians implemented (protocol-wise)?
00:09 < Xor_> I think that's what that combat packet is for
00:10 < Aragasas> Spigot just make optimizations, right? I always test my library with vanilla and craftbukkit. Should i add Spigot too? Or they handle protocol like cb?
00:10 < Aaron1011> According to the docs, the player ID is just sent for ENTITY_DEAD
00:10 < Aaron1011> How does it target a player?
00:10 < Aaron1011> Or am I just missing something obvious? :)
00:11 < Thinkofdeath> Aragasas: If it works bukkit then it will most likely work on spigot
00:11 < Xor_> ask Thinkofdeath
00:11 < Thinkofdeath> ?
00:11 < Aragasas> Thinkofdeath: okay
00:12 < Xor_> about the prorocol thign with them guardians
00:12 < Xor_> "Increased the max payload size of 'Plugin Message' from 32767 to 1048576 (Broken because of incorrect data type)"
00:12 < Xor_> is that still a thing?
00:12 < Thinkofdeath> yep
00:12 < Thinkofdeath> oh but I never reported
00:13 < Xor_> well, might want to tell someone
00:13 < Xor_> and make a ticket
00:13 < Aaron1011> Xor_: Wait, are you talking about the guardians protocol?
00:13 < Xor_> wat no
00:14 < Xor_> I'm shouting at Thinkofdeath for not reporting at bug
00:14 < Aaron1011> Oh, you said "about the prorocol thign with them guardians"
00:14 < Xor_> ask him about Guardians, he might know
00:14 < Aaron1011> I got confused :)
00:15 < Thinkofdeath> I haven't touched them but I see something about them somewhere in the code when I was looking for something else
00:15 < Aragasas> 1048576? Nice. Now we can make our version of chunks data transmission, with blackjack and lulzes
00:15 < Thinkofdeath> I think the attack is an effect packet
00:15 < Thinkofdeath> Aragasas: The issue being its length prefix is still a short
00:16 < yawkat> unsigned i hope?
00:16 < Thinkofdeath> yeah but still isn't the right size
00:18 < Aragasas> Thinkofdeath: so, make it integer? Or there's some troubles?
00:18 < Xor_> gotta get the mojang people to do that
00:18 < Xor_> which is why Thinkofdeath should be submitting a ticket
00:18 < Aragasas> BigInterger sound perfect, lol
00:18 < Xor_> s/BigInteger/VarInt
00:22 < Aragasas> So, we have now options to encrypt and compress data
00:23 < Thinkofdeath> encryption isn't optional
00:23 < Xor_> but what if I want the NSA to spy on me?
00:24 < yawkat> Xor_: just spam all packets to the client via unencrypted udp manually
00:24 < yawkat> that should do it
00:24 < Aragasas> Thinkofdeath: You mean, it can't be disabled now?
00:24 < Xor_> yawkat, you're a genius
00:24 < Thinkofdeath> Aragasas: its on for online or off or offline mode
00:24 < Xor_> I'll spam the client with so many packets it will explode!
00:25 < Thinkofdeath> you can't change that
00:25 < Thinkofdeath> unlike compression
00:25 < yawkat> Xor_: brilliant idea!
00:25 < Xor_> Thinkofdeath there's probably a way that will probably not make mojang happy.
00:25 < Aragasas> Ah, yeah
00:25 < Xor_> hence, there is effectively no way
00:26 < Aragasas> Hm...What if we compress data, then encrypt and compress secont time? sounds like a good idea