07:33 <+SirCmpwn> I might be, and if they do get done I'll be happy
07:33 < Grum> dx: it is
07:33 <+SirCmpwn> I am happy to hear that you guys are working on this stuff
07:33 < Grum> we just have higher priority things to handle though :(
07:33 < Grum> next high up target is the worldformat
07:33 <+SirCmpwn> remember though that I've been doing this stuff for several years now
07:33 < Grum> SirCmpwn: remember that i've been doing this too
07:33 <+SirCmpwn> my historical view of the cooperation between mojang and our community is not a good one
07:33 <+SirCmpwn> if that's changing, I'm very happy to hear it
07:34 < Grum> SirCmpwn: it's understandably as it is
07:34 <+SirCmpwn> oh, and the debacle around the server list ping that I mentioned in that thread
07:34 < Grum> also the 'development of an api can only happen if you ahve the community do it' .. that obviously is a 100% nogo
07:34 <+SirCmpwn> even the implementation now sucks after you tried to fix it
07:34 <+SirCmpwn> I never said that?
07:34 < Grum> which serverlist ping? :/
07:34 <+SirCmpwn> err, the status thing
07:34 <+SirCmpwn> whatever you call it
07:34 < Grum> "I would want the API to be open source and built by the community, for one."
07:34 < dx> server list ping was awful because of backwards + forwards compatibility, right?
07:34 <+SirCmpwn> ah
07:34 < Grum> dx: indeed
07:35 <+SirCmpwn> "build by the community" was a bad thing to say
07:35 <+SirCmpwn> I meant that it should be built by Mojang and driven by community feedback (and code where relevant)
07:35 <+SirCmpwn> built*
07:35 < Grum> yes, but that assumes a certain kind of api
07:35 < Grum> the kind that Bukkit is and forge 'partially' provides (forge isn't that much of an api, only on the places of many clashes)
07:35 <+SirCmpwn> dx: the part that pisses me off about it is that each time they thought about changing it, we presented them with completely working, elegant solutions that maintained backwards compatability
07:35 <+SirCmpwn> dx: and they went off and did something else every time
07:36 < Grum> SirCmpwn: last time you had one of those solutions holes were trivially made in the solution
07:36 <+SirCmpwn> Grum: right, I desire this kind of API
07:36  * SirCmpwn does not recall holes being pointed out
07:36 <+SirCmpwn> did you see holes and keep quiet about them?
07:36 < Grum> i'd never do that, you know me :P
07:36 < dx> iirc forwards compat was broken in some version
07:36 < Grum> tbh, the 'ugliness' should simply go
07:36 <+SirCmpwn> you and I have spoken serious implementation talk on like two occasions
07:36 < Grum> and we should stop doing backwards compat
07:36 <+SirCmpwn> seperated by like a year at a time
07:37 <+SirCmpwn> well, not being backwards compatabile is fine
07:37 < Grum> SirCmpwn: 'this kind of api' being whaT?
07:37 <+SirCmpwn> but at least being able to properly tell clients that they're outdated or too recent, and how they can fix it, is desirable
07:37 < Grum> at the time of implementation we couldnt afford not to be backwards compatible to the previous version
07:37 <+SirCmpwn> Grum | yes, but that assumes a certain kind of api
07:37 <+SirCmpwn> Grum | the kind that Bukkit is and forge 'partially' provides (forge isn't that much of an api, only on the places of many clashes)
07:37 < Grum> now we could
07:37 <+SirCmpwn> you could, I know it
07:38 < dx> it was <=1.3 that had the shittiest server list ping protocol, right? i barely remember the details of this, but that one is definitely 'old enough' already
07:38 <+SirCmpwn> dx: yeah, it was pretty shitty in the far past
07:38 < Grum> 1.6 still is shitt
07:38 < Grum> and doesnt mesh at all with 1.7 style
07:38 < Grum> and we hacked it sortof working
07:38 <+SirCmpwn> I would have done the 1.7 style differently
07:38 <+SirCmpwn> I would have done a more involved handshake before entering the different protocol modes
07:38 <+SirCmpwn> which could have easily supported legacy pings
07:38 < Grum> yes, but i do not want to support legacy pings or design the new system to take that in account
07:39 < Grum> that hinders the design of the new system
07:39 <+SirCmpwn> so we fundamentally disagree on this point
07:39 <+SirCmpwn> but that's in the past, nothing to be done about it now
07:39 < Grum> which in itself is quite decent
07:39 <+SirCmpwn> also, varints are a joke
07:39 < dx> SirCmpwn: remember we don't get to modify the server to test those theories with the actual server code
07:39 <+SirCmpwn> but I doubt it's even worth having that discussion because I seriously doubt you'll turn back on them now
07:39 < Grum> why are they a joke?
07:40 <+SirCmpwn> you're trading tiny amounts of bandwidth for orders of magnitude more CPU time on the server and client
07:40 <+SirCmpwn> and in places where it really doesn't matter
07:40 <+SirCmpwn> you can't tell me with a straight face that 65535 packets isn't enough
07:40 < Grum> i can tell you with a straight face that coding such an arbitrary limitation into a system is silly?
07:40 < Drainedsoul> And the encoding of negative numbers as a varint (unless that's been changed) is kind of abominable
07:40 < Grum> 256 blocks is also enough right?
07:41 < Grum> 640k ram too
07:41 < Grum> Drainedsoul: it is, we barely do that though
07:41 < dx> he did say "in places where it really doesn't matter"
07:41 <+SirCmpwn> I will bet you $10,000 that you will never use more than 65535 packet IDs
07:41 < dx> sometimes it matters, sometimes it doesn't
07:41 < Grum> Drainedsoul: when we actually start doing it, we'll zip them properly
07:41 < Drainedsoul> -_-
07:41 < Drainedsoul> will you zip them properly
07:41 < Drainedsoul> or just in certain places
07:41 < Drainedsoul> -_-
07:41 < Grum> SirCmpwn: you do realize i can just add the required packets to 'win' that argument right?
07:41 < dx> hahah
07:41 < dx> will it be worth it?
07:42 <+SirCmpwn> 60,000 packets that all do the keep alive is cheating :P
07:42 < Grum> dx: it would come at no cost
07:42 < Grum> SirCmpwn: not really
07:42 < Grum> it would be proving a point and making me lots of monies
07:42 <+SirCmpwn> but you understand my point
07:42 < Grum> yes, but please undersrtand my point too
07:42 <+SirCmpwn> I do understand your point
07:42 < Grum> we initially did that decision based on that we'd not recycle ids
07:42 < Grum> we then later decided we are going to recycle them
07:42 <+SirCmpwn> (fyi, you still have a maximum varint size and thus still have an arbitrary limit)
07:42 < Grum> in theory we could just make them bytes atm
07:43 < Grum> because we simply do not have enough packets
07:43 < dx> hmm
07:43 < dx> varints do help with forwards compat in case that changes...
07:43 < Grum> and on that note, we do not have enough packets to hit the second var-int-byte
07:43 < Grum> so the cost of reading is literally a single bit-check and that's it
07:43 <+SirCmpwn> true, but still higher than just reading the damn thing
07:43 <+SirCmpwn> and for a stupid reason
07:43 < Grum> no, for forward compatibilty reasons
07:44 < Grum> what if some mod decides to add 123123123 packets?
07:44 <+SirCmpwn> that's like buying a billion burgers because you don't want to run out before you die
07:44 <+SirCmpwn> mods use the plugin message
07:44 <+SirCmpwn> this is the blessed means
07:44 < Grum> maybe those should be indexed packets
07:44 < Grum> at least now they *can* be
07:44 <+SirCmpwn> that sounds pretty cool and might justify varints more
07:44 < Grum> at the cost of checking a single bit in a byte
07:45 < Grum> i realize there is a bit of cost
07:45 < Grum> but the increase flexibility justifies it
07:45 < dx> yeah, in performance, string comparison vs a shitton of varint packet ids...
07:45 <+SirCmpwn> I concede on that particular point
07:45 <+SirCmpwn> (packet IDs)
07:45 < Grum> the one big problem we have in our system right now is that we're 100% statically tied down with lots of arbitratily picked limits
07:45 <+SirCmpwn> but I still think you're using a bit too eagerly
07:45 <+SirCmpwn> "you" being mojang
07:45 < Grum> perhaps
07:45 < Grum> especially for locations
07:45 <+SirCmpwn> all the tiny little performance losses add up
07:46 < Grum> yes, but you've never seen a secondary cpu  used by all the reader threads
07:46 < Grum> +fully
07:46 < Grum> so yeah, maybe it moved up from 2 to 2.4%
07:46 < dx> somehow benchmarking the real impact of varints would be cool
07:46  * SirCmpwn shrugs
07:46 <+SirCmpwn> it's a hit
07:46 <+SirCmpwn> and not a justified one in lots of places
07:46 < Grum> its a huuuuge hit
07:46 < Grum> possibly
07:46 <+SirCmpwn> it's not a huge deal, we've spent a lot of time on this one problem
07:46 < dx> it would be huge with thousands of clients
07:46 <+SirCmpwn> let's move on to a bigger one: crypto
07:46 < Grum> dx: which we do not support
07:47 < dx> Grum: :D
07:47 < Grum> maybe we should, but then maybe we should fix other issues first
07:47 < dx> i actually enjoy not giving a fuck about scalability in my personal projects
07:47 <+SirCmpwn> the roll-your-own-crypto in minecraft is stupid for reasons I really shouldn't have to explain
07:47 < Drainedsoul> dx :(
07:47 < Grum> like not spamming a fucking packet 20 times per second just to indicate 'I AM ON TEH FLOOR'
07:47 < dx> Drainedsoul: ?
07:47 < Drainedsoul> idk, I like worrying about that in the same way I like worrying about Unicode :D
07:47 < Grum> SirCmpwn: there is no trivial alternative for a 3-way auth like we're doing
07:48 <+SirCmpwn> I would have just established a TLS connection and did some simple checks afterwards to verify that the user is who they say they are
07:48 < Grum> based on what?
07:48 < Grum> also we basically rolled our own tls setup
07:48 <+SirCmpwn> old auth system would have worked well enough in that situation
07:48 < Grum> 'old auth system' ?
07:48 <+SirCmpwn> like old old
07:49 <+SirCmpwn> not old old old but old odl
07:49 <+SirCmpwn> old old*
07:49 <+SirCmpwn> the one with the shitty HTTP API on mojang.com where you did a super simple juggling things around
07:49 < dx> "rolled our own tls setup" - that's what he's complaining about, mostly because it's not tls
07:49 < Grum> ok, first of all, the encryption is purely against mitm
07:49 <+SirCmpwn> which would have been resolved just by wrapping the old way in TLS
07:49 < Grum> dx: yes because it is not tls, else we would have used it
07:50 < dx> Grum: yes, and that in itself is a problem, unless you manage to get it audited
07:50 < dx> but meh.
07:50 < Grum> SirCmpwn: i have the feeling you are just complaining that you cannot use a library to do something that you think should have been done with a library eventhough a library wouldn't get the job done in this case
07:50 <+SirCmpwn> I have little faith in your TLS implementation, but it's just a game so I guess who cares
07:50 <+SirCmpwn> s/TLS/psuedo-TLS/
07:50 < Grum> it's rather simple tbh :/
07:50 < Drainedsoul> s/psuedo/pseudo/
07:50 <+SirCmpwn> thanks Drainedsoul 
07:51 < dx> fucking up crypto is rather simple
07:51 < Grum> yes it is, but what we have is good :/
07:51 <+SirCmpwn> first rule of crypto is don't roll your own crypto
07:51 <+SirCmpwn> second, third, and fourth rules of crypto are also don't roll your own crypto
07:51 < Grum> SirCmpwn: agreed, until i couldnt find anything that would get the job done
07:51 < dx> first rule of the crypto club
07:51 < Grum> SirCmpwn: i truly didn't like writing it
07:51 <+SirCmpwn> Grum: ...old auth plus TLS would have worked
07:51 < Grum> there is no such thing as 'old auth'
07:51 <+SirCmpwn> meh, whatever
07:51 < Grum> there is no such thing as 3rd-party-tls
07:51 <+SirCmpwn> shall we move on to more greivances
07:52 < dx> fifth rule is just use TLS or PGP or OTR or bcrypt
07:52 < dx> ...that's not a lot of options
07:52 < Grum> and none of them could do what we needed dx
07:52 < dx> i know
07:52 < dx> and ugh i feel so uncomfortable recommending OTR, knowing how awful the implementations are :(
07:52 < dx> but the spec is solid
07:52 < Grum> OTR being?
07:52 < Drainedsoul> given that a lot of crypto is implemented ontop of OpenSSL
07:52 < Drainedsoul> you should feel that way about all crypto
07:52 < Drainedsoul> ;)
07:53 < Grum> yeah ahh imagine us using tls, we'd have heartbleed in all servers too!
07:53 < dx> Grum: completely unrelated, "off-the-record" encryption for instant messaging https://otr.im/
07:53 <+SirCmpwn> who cares about heartbleed in minecraft
07:53 <+SirCmpwn> Craft.Net generates a new keypair every time the server starts up
07:53 <+SirCmpwn> if it's leaked, boo hoo
07:54 < Drainedsoul> if your server instance is long running, it'd matter
07:54 < Grum> dx: that is what we implemented on the wire actually
07:54 <+SirCmpwn> just tell everyone to restart their minecraft servers
07:54 <+SirCmpwn> boom, mitigated
07:54 <+SirCmpwn> anyway
07:54 < Grum> not at all?
07:54 <+SirCmpwn> more problems: client-side prediction
07:54 < Grum> you still have to fix the libraries used
07:54 < dx> Grum: uhhhh no, this is a very specific protocol with a shitty name, it's nothing like your stuff
07:54 < Grum> SirCmpwn: yeah lol sigh, fuck that
07:54 <+SirCmpwn> well, update and then restart your server
07:54 <+SirCmpwn> no need to revoke things
07:54 < Grum> dx: read the high-level overview https://otr.cypherpunks.ca/Protocol-v3-4.0.0.html
07:54 < Grum> that is exactly what we do
07:55 < Grum> except for the 'lets pick how we communicate the key'
07:55 < Grum> that is always RSA1024
07:55 < Grum> and the cypher after is AES128
07:55 < dx> apples to oranges really
07:56 < Grum> also that doesn't involve 'intention checking' through a 3rd party
07:56 < dx> you may be using the primitives but this is for a completely different purpose
07:56 < dx> *the same primitives
07:56 < Grum> setting up a encrypted stream to the server
07:56 <+SirCmpwn> so, err
07:56 <+SirCmpwn> client-side prediction chat, anyone?
07:56 < dx> SirCmpwn: tab completion? or what?
07:57 <+SirCmpwn> no, things like physics being implemented on the client
07:57 <+SirCmpwn> specifically physics is a HUGE pain in my ass
07:57 < dx> oh
07:57 <+SirCmpwn> because I have to reimplement minecraft physics in a black box to the tick or it looks jumpy and shitty
07:57 < dx> ehh
07:57 <+SirCmpwn> also: window management is handled on both and gets shit fucked up
07:57 <+SirCmpwn> the client can craft things and put them in their inventory and PLACE THEM without the server ever agreeing to it
07:57 < Grum> window management being inventories etc?
07:58 <+SirCmpwn> Grum: yeah
07:58 < dx> the physics stuff sucks but that's how networking in games usually works :/
07:58 < Grum> SirCmpwn: yes we planned to fix that with the inventory rewrite which didn't quite finish yet :(
07:58 < dx> implement the same shit in both sides and hope they agree
07:58 <+SirCmpwn> well, glad to hear it's coming, Grum 
07:58 < Grum> dx: the problem is they will never agree because we 'truncate' the values over the wire
07:58 < Grum> so server and client have a different image of reality
07:59 < dx> Grum: like, say, turning from floating point to fixed point?
07:59 < Grum> yes, truncating it into 1/8000 resolution
07:59  * SirCmpwn notices this on wiki.vg
07:59 <+SirCmpwn> >As of 1.7.6 all UUIDs used in the protocol now contain '-'. The session server still returns them without 
08:00 <+SirCmpwn> shall we chat about the session server (and related servers)
08:00 < Drainedsoul> Are UUIDs still sent as strings in the protocol?
08:00 < Grum> ugh i hate the text representation of uuids
08:00 < Grum> it should just be two longs
08:01 <+SirCmpwn> the biggest problem with the new auth system is that I don't know how to pronouce it or spell it
08:01 < Drainedsoul> more like unsigned __int128
08:01 <+SirCmpwn> other than that, it's a huge improvement, thanks
08:01 < Grum> igg-drassil
08:01 < Grum> Drainedsoul: same difference
08:02 < Grum> SirCmpwn: java is a bit whiny on the textual representation of uuids
08:02 <+SirCmpwn> from my perspective, I don't really care
08:03 <+SirCmpwn> the implementation is your problem, I just criticise the spec ;)
08:03 <+SirCmpwn> java was a bad choice for minecraft anyway
08:03 < Drainedsoul> I'd cry if I didn't have unsigned integers myself :(
08:03 < dx> SirCmpwn: spec? what spec?
08:04 <+SirCmpwn> you know what I mean, dx.
08:04 < SinZ> protocol
08:04 < dx> :P
08:04 < dx> implementation is our problem too, unfortunately
08:04 <+SirCmpwn> also, Grum, I don't know much about your upcoming changes to block IDs and such, but I'm a bit worried
08:04 < Grum> most of that is in 1.8
08:04 < Grum> though not reflected in the saved format L(
08:04 <+SirCmpwn> are blocks going to have consistent IDs?
08:04 <+SirCmpwn> into the foreseeable future?
08:04 < Grum> 'consistent ids' ?
08:05 <+SirCmpwn> as in the same series of bits will always refer to the same block type
08:05 <+SirCmpwn> like "stone"
08:05 <+SirCmpwn> I like that you threw out metadata, by the way
08:06 <+SirCmpwn> dislike that I still have to think about it cause it's not in anvil
08:06 < Grum> that will never happen what you just requested
08:06 < Grum> stone can be another id on a different dimension
08:07 <+SirCmpwn> ...
08:07 < Grum> you know why? its a property of the level, not of the stone
08:07 <+SirCmpwn> block IDs are now dictated by the level?
08:07 < Grum> artifact of serialization
08:07 < Grum> not yet
08:07 <+SirCmpwn> let's throw out performance
08:07 <+SirCmpwn> just completely forget about it
08:07 <+SirCmpwn> and code clarity, too, let's forget about that
08:08 < Grum> not sure what you are talking about
08:08 <+SirCmpwn> pairing a block with the logic that defines its behavior will be much worse
08:08 <+SirCmpwn> and reassigning IDs at runtime depending on the level, ewww
08:08 <+SirCmpwn> trying to do a lookup for anything will be dramatically less performant
08:09 < Grum> you do realize there is already a lookup going on right?
08:09 <+SirCmpwn> yeah, an index in an array (at least in Craft.Net)
08:09 <+SirCmpwn> which is super speedy
08:09 <+SirCmpwn> managing several hash tables instead and generating them at runtime sounds like a whole lot of unfun
08:09 < Grum> why would this be different?
08:09 < dx> hmm, the new implementation could also build arrays like that in runtime
08:09 < Grum> yeah, i dont quite see the problem
08:09 < dx> no idea about mod blocks though
08:10 <+SirCmpwn> hmm, I guess you could just build LUTs at runtime
08:10 < dx> harder to assign array indexes to those
08:10 <+SirCmpwn> but again, eww
08:10 < Grum> you ask for 'x,y,z' in world Z and you get back an index, you look up the index in an array ...
08:10 < dexter0> Could you at least assign them at startup time instead of assigning them per-dimenison?
08:10 < Grum> the same as it is now
08:10 < Grum> but the array is dynamically populated
08:10 <+SirCmpwn> and several of them, one per level
08:10 <+SirCmpwn> after thinking about it I can see the value, though
08:10 < Grum> SirCmpwn: why would that matter?
08:10 < dx> dexter0: what's the difference between "startup" and "level load"?
08:10 < Grum> it actually is almost required to do that
08:10 <+SirCmpwn> Grum: it's just uglier
08:10 < Grum> imagine you starting 1 world that is vanilla
08:11 < Grum> and now you load up another world with mods
08:11 <+SirCmpwn> yeah, I know the justification
08:11 < Grum> now you want to have both on the same server
08:11 <+SirCmpwn> it makes sense with more thought
08:11 < Grum> OMG THAT IS IMPOSSIBRU BECAUSE MAPPINGS ARE DIFFERENT
08:11 <+SirCmpwn> are you going to do a level format migration
08:11 < Grum> yes
08:11 < Grum> that is what 1.9 is about
08:11 <+SirCmpwn> can we interject some feedback into it when you settle on a format
08:11 < dexter0> dx: Per level means stone has a different ID in the neather vs Overworld.  Unless I misunderstand.
08:11 <+SirCmpwn> that sentence is weird
08:11 < Grum> we're dropping byte+nib+nib
08:11 <+SirCmpwn> just pretend it was well written
08:11 < Grum> SirCmpwn: give me your thoughts on a sane format
08:12 < Grum> i've not made up my mind
08:12 < Grum> but i want it to be 'pluggable' so we can change it later without problems
08:12 <+SirCmpwn> I haven't even spent 2 minutes thinking about it
08:12 <+SirCmpwn> I'll get back to you
08:12 < Grum> unlike the shit we have now
08:12 < Grum> ok
08:12 < Grum> one goal is that we store only 16^3's
08:12 <+SirCmpwn> first suggestion is a format version early in the file and a series of migration scripts that can take you from any one version to any other
08:12 < Grum> well
08:12 <+SirCmpwn> if you want futureproofing
08:12 <+SirCmpwn> second suggestion is fuck NBT
08:13 < Grum> each chunk (16^3) should be stored with a 'version'
08:13 < Grum> as first varint orso
08:13 < Drainedsoul> What's the source of the antipathy towards NBT
08:13 < Grum> and then whatever the fuck comes after will just get handled by a 'reader' for that 'type'
08:13 < Drainedsoul> not saying NBT is good, jw
08:13 < Grum> nbt sucks
08:13 < Grum> there is 0 reason at all to store this as a nbt struct
08:13 < Grum> just flushed binary data works FAR better
08:14 <+SirCmpwn> agreed 100%
08:14 < dx> SirCmpwn: I'd just like to interject for a moment. What you’re referring to as Block IDs, is in fact, Block UUIDs, or as I’ve recently taken to calling it, UUIDs plus Blocks. Blocks are not integers by themselves, but rather another free component of a fully functioning Mod API system made useful by the Mojang corelibs, forge utilities and vital system components comprising a full voxel game as defined by Notch.
08:14 < Grum> downside? you need a custom reader
08:14 < Grum> but why is that a downside O.o
08:14 < Drainedsoul> so was that "fuck NBT" in general, or just in this specific instance
08:14 <+SirCmpwn> dx: diaf
08:14 < dx> SirCmpwn: ♥
08:14 <+SirCmpwn> Drainedsoul: fuck NBT in generally, but specifically get it out of the level files
08:14 <+SirCmpwn> "in generally"
08:14 < Grum> SirCmpwn: going to happen
08:14 <+SirCmpwn> I am really on top of it tonight
08:14 < Grum> i dislike nbt
08:14 < Grum> its stupid overhead on that level
08:14 < Grum> trivial to corrupt
08:14 < Grum> impossible to recover
08:15 < Drainedsoul> SirCmpwn: Do your reasons mirror those of Grum?
08:15 <+SirCmpwn> more or less
08:15 < Grum> also our code interface into nbt is atrocious
08:15 < Grum> nbt the format itself is not unusasble
08:15 < Grum> but its a non-standard
08:15 <+SirCmpwn> I might suggest having two seperate areas of a chunk in the level format
08:15 < Grum> we might as well pick messagepack
08:15 < dx> or json
08:15 < Grum> anything
08:15 <+SirCmpwn> I would have all the IDs and such
08:15 < Drainedsoul> I was just curious if people hate NBT because reading/writing it is a pain, because interacting with it programmatically is a pain, or because of other, technical reasons
08:15 < Grum> though json sucks at holding binary data
08:16 < dx> ok i need to leave this conversation, i'm not contributing positively anymore
08:16 <+SirCmpwn> and then JSON for the blocks that need it (don't make many blocks need it)
08:16 < Drainedsoul> dude just base64 encode stuff /s
08:16 <+SirCmpwn> and JSON for entities
08:16 <+SirCmpwn> and then gzip the lot
08:16 < Grum> yes, block data and entity data should be stored seperatly
08:16 <+SirCmpwn> I might even go so far as to say seperate files
08:16 < Grum> or 'just world data' vs 'anything that is not world-data'
08:16 < Drainedsoul> is the icon for the multiplayer screen still sent base64 encoded inside JSON?
08:16 <+SirCmpwn> oooh, idea time
08:16 <+SirCmpwn> Grum: you prepared for this?
08:16 < Grum> SirCmpwn: i'd prefer not to do that, in theory we can just have sectors within 1 bigass file
08:17 < Grum> Drainedsoul: ye
08:17 <+SirCmpwn> I have no clue if this is a good idea but it could be cool
08:17 <+SirCmpwn> put extra block data into sqlite
08:17  * SirCmpwn drops the mic
08:17 < Drainedsoul> I was about to ask why they don't use some sort of database for this, but
08:18 <+SirCmpwn> Grum: how radical are you feeling, because I just got more ideas
08:18 < dx> mongodb?
08:18 <+SirCmpwn> hell no
08:18 < Grum> SirCmpwn: i was going to store 'full air chunks' as 2 bytes
08:18 < Grum> or 1
08:18 < Drainedsoul> dx don't make me get my pitchfork out
08:18 <+SirCmpwn> good start but I'm thinking more radical
08:18 <+SirCmpwn> have you ever heard of plan9, Grum 
08:18 < Grum> nope
08:18 <+SirCmpwn> okay
08:18 <+SirCmpwn> so what if the world server was seperate from the game server
08:19 <+SirCmpwn> and you had a dedicated server for serving the world and handling changes to it (which could be bundled into the same thing and run transparently)
08:19 < Grum> http://cm.bell-labs.com/plan9/ ?
08:19 <+SirCmpwn> yeah, that's it, but if you're not already familiar with it it'll take too long to catch up
08:19 <+SirCmpwn> anyway, the pros are that you can easily put the world stuff somewhere else
08:20 < Grum> lets assume that having it run as a seperate server doesn't create too much overhead
08:20 < Grum> but i do want to create an interface clean enough to let you do that
08:20 <+SirCmpwn> it'd be easy to have common state on one server and then have a bunch of seperate game servers that don't need to have much state
08:20 <+SirCmpwn> and load balance them for great things
08:20 <+SirCmpwn> that's starting to get into what I'd like to see in a full-blown ground-up redesign of minecraft, though
08:21 < dx> let's reimplement minecraft in rc shell
08:21 < dx> every block is a file
08:21 <+SirCmpwn> hahaha
08:21 < Grum> sir, not at all
08:21 < Grum> just a very well defined boundary to 'where you interact with leveldata'
08:21 <+SirCmpwn> one thing Mojang is really bad at is supporting bigger servers
08:21 <+SirCmpwn> if you were interested in changing that, I'd have lots of things to talk about
08:23 < Grum> the server hasn't had any love at all
08:23 <+SirCmpwn> unfortunately, the server is what needs the love
08:23 <+SirCmpwn> server modding is way more common than client modding
08:24 <+SirCmpwn> and in general, the servers are what needs the most tuning
08:24 < expir3dcowPC> Is server modding easier than client modding?
08:24 <+SirCmpwn> a poorly optimized client only affects the computer it's running on, a poor server will fall under its own weight much more quickly
08:24 <+SirCmpwn> expir3dcowPC: probably
08:24 < dx> expir3dcowPC: welcome to minecraft
08:24 < expir3dcowPC> dx, Hi
08:24 < dx> well actually, now that bukkit is dead, it's not really easier
08:25  * SirCmpwn does not write mods, he just writes Craft.Net
08:25 < expir3dcowPC> I've played minecraft for a year, just got into programming recently
08:25 <+SirCmpwn> oh man, another one
08:26 < expir3dcowPC> If I wantto send a packet to a server do I do Outputstream.writeBytes(packet)?
08:26 <+SirCmpwn> please learn the basics before you bother us, expir3dcowPC 
08:26 <+SirCmpwn> minecraft is not somethign to cut your teeth on, if you hadn't been paying attention
08:26 <+SirCmpwn> something*
08:26 < Grum> if you name your variables weirdly with an uppercased first letter and it happens to be connected to a socket connected to the server, yes
08:27 < Grum> maybe, if it gets flushed :P
08:27 < expir3dcowPC> Thanks
08:27 < dx> ok ok here's an idea: every block is a filesystem interface, you can launch applications such as rio (the window manager) and choose to display them on any of the faces of a block. you can also mount another instance of the 9craft filesystem *on the block itself* so you can have something like the microblocks mods but with filesystems instead
08:27 < dx> i'm a genius
08:27 <+SirCmpwn> yes, plan9 has a lot of realyl radical ideas and it's a bit much
08:27 <+SirCmpwn> but some of its ideas are pretty cool in isolation
08:28 <+SirCmpwn> really
08:28 <+SirCmpwn> god dammit SirCmpwn 
08:28 < dx> wot
08:28 < dx> oh your typo, i see
08:29 <+SirCmpwn> Grum: so can we talk about command blocks now
08:31 < dx> i'm just waiting for the day in which mojang gives up creating their own programming language in command blocks and switches to lua
08:31 < SinZ> Lua is a pain to type in a one-line textbox though
08:31 < SinZ> <3
08:31 <+SirCmpwn> here's an idea: if you want to mod minecraft, fucking mod it
08:31 < dx> well of course by that point it won't be an one-line textbox
08:32 < dx> i'm half-serious with what i'm saying about lua, btw
08:32 <+SirCmpwn> Grum: I've never felt like you were ruining minecraft, but I feel that way about Dinnerbone
08:32 < dx> it would count as decent client side scripting
08:32 <+SirCmpwn> is he going to be reined in any time soon
08:33 < SinZ> dx: inb4 modding community bitching that Mojang copied computercraft
08:34 < dx> SinZ: hahah i completely forgot about that
08:34 < dx> it's pretty much the same thing i was """suggesting"""
08:34 < dx> absurdly advanced command blocks that can move by themselves
08:35 <+SpaceManiac> command block minecarts
08:35 <+SirCmpwn> do those exist now?
08:35 <+SpaceManiac> that place minecart track as they go in the correct direction
08:35 < SinZ> yes
08:35 <+SirCmpwn> what in god's name could justify those existing
08:35 < SinZ> Welcome back, Sethbling here.
08:35 < SinZ> </3
08:36 < dx> SirCmpwn: minecarts being able to hold anything, i guess
08:36  * SirCmpwn is working on refactoring sethbling's website at the moment
08:36 < dx> sellout!
08:36 < dx> j/k
08:37 <+SirCmpwn> sethbling I don't mind, it's dbone adding a bunch of shit for him and his friends that pisses me off
08:37 < dx> uhm sorry but sethbling is one of those mapmaker friends
08:37 <+SirCmpwn> ?
08:38 < dx> 03:35 <+SirCmpwn> what in god's name could justify those existing
08:38 < dx> 03:35 < SinZ> Welcome back, Sethbling here.
08:38 <+SirCmpwn> ah, I missed the correlation
08:38 <+SirCmpwn> in any case, it feels like minecraft has lots its way and is driving itself steadily into the ground
08:39 <+SirCmpwn> like right now minecraft is about 35 seconds into this video: https://www.youtube.com/watch?v=Zl12dXYcUTo
08:39 < dx> i think the mapmaker related changes are just a bunch of workarounds to keep at least some part of the community entertained with the game
08:39 <+SirCmpwn> I completely agree
08:40 < Grum> SirCmpwn: commandblock sshould die
08:40 < dx> \o/
08:40 < Grum> i dislike them, i tolerate them because they fill a need
08:40 <+SirCmpwn> I agree
08:41 <+SirCmpwn> will it die?
08:41 < dx> Grum: ♥
08:41 < Grum> i only tolerate them because you cannot get them other than through /give
08:41 < Grum> which makes them suited for the purpose they are for, non-standard building
08:41 <+SirCmpwn> I think the very concept they represent is flawed
08:41 <+SirCmpwn> minecraft has limitations
08:41 <+SirCmpwn> this is why minecraft is fun
08:41 <+SirCmpwn> it used to be impressive when you made crazy cool redstone contraptions
08:42 <+SirCmpwn> "you" in the vague sense
08:42 <+SirCmpwn> now minecraft is adding more and more and more and more and appealing to the instant gratification crowd who wants a big flashy game with tons of stuff in it
08:42 < dx> now they are boring, but with better performance!
08:42 < SinZ> thats why commandblocks should've only been done for adventure mode stuff, where you want to do stuff that is impossibru in pure redstone
08:42 <+SirCmpwn> those people grown up and turn into the crowd who makes their own fun from limited tools
08:42 < Grum> commandsblocks are more things like minigames that would require excess abuse of redstone making it slower than needed
08:42 <+SirCmpwn> grow*
08:42 < Grum> but anything a commandblock can do should in the end be a plugin
08:42 < SinZ> not for hurrhurr I have built a full computer in 3 blocks
08:42 < Grum> if at all
08:42 <+SirCmpwn> Grum: the point, I think, is that minigames are not appropriate for minecraft
08:43 < SinZ> Atleast not in pure vannila
08:43 < Grum> SirCmpwn: according to the community they are
08:43 <+SirCmpwn> a rapidly divided community
08:43 <+SirCmpwn> maybe I'm wrong, but I predict that this is the wrong path
08:43 < SinZ> Let that be server plugins like what custom servers have been doing since 2009
08:44 < dx> SirCmpwn: "not appropriate" maybe technically speaking, because they are a bunch of hacks, but whatever, gotta live with it now
08:44 < GauHelldragon> oh ! 
08:44 <+SirCmpwn> not in the technical sense, dx 
08:44 <+SirCmpwn> I mean that it's driving us away from what makes minecraft special to gamers
08:44 < dx> ehhh, meh
08:45 <+SirCmpwn> modders can mod it to do whatever they want, but that's not minecraft any more
08:45 < dx> there are many ways to play minecraft, i find building boring, and pure survival is okay once every 6 months
08:45 <+SirCmpwn> when it actually _is_ minecraft, things change imo
08:45 <+SirCmpwn> this discussion is getting way into the clearly subjective and intangible classification of feedback, though
08:45 < Grum> the power of minecraft is that you can play it in many ways + community + magic
08:45 < dx> ~magic~
08:46 < SinZ> Grum: is that why at one point you guys were trying to break mob farms?
08:46 <+SpaceManiac> *arm flailing*
08:46 <+SirCmpwn> there's a limit on how much I can get upset about
08:46 <+SirCmpwn> the actual vision of the game is clearly in Mojang's hands
08:46 <+SirCmpwn> I can only really justify griping about the technical choices they make to arrive at that vision
08:47 <+SirCmpwn> dx tells me I should be angrier
08:47 < dx> hahahha
08:47 <+SirCmpwn> minecraft fucking sucks and it's all Grum's fault
08:48 < dx> i didn't say anything
08:48  * dx hides
08:48 < Grum> at least we could whore it out
08:48 < Grum> seems to be good at that
08:49 <+SirCmpwn> speaking of being whores
08:49 <+SirCmpwn> I still own minecraft.pe
08:49 <+SirCmpwn> you guys still don't want it?
08:49 < dx> didn't .pe cost like 100 bucks?
08:50 < Spurlex> Psssst. Grum. You should get security questions in the launcher if a user logs in from a suspicious ip.
08:50 <+SirCmpwn> I don't want to sell it for a bunch of money, I just noticed it was available and grabbed it before the real squatters showed up
08:50 <+SirCmpwn> dx: something like that
08:50 < dx> Spurlex: define "suspicious ip"
08:50 < Spurlex> Well if i login from my country New Zealand
08:50 < Spurlex> Then suddenly an ip logs in from United States
08:50 < Spurlex> thats suspicous so ask security question
08:51 < dx> ehhh
08:51 < dx> that's kinda annoying but yeah sure whatever man
08:51 < Spurlex> Because people are buying 100’s of alt accounts for like $5. 
08:51 < Grum> Spurlex: then you assume that ip's actually have fixed countries attached to it
08:51 < Spurlex> and the reason there are so many alts is because those passwords cant be changed by the people who hacked them.
08:51 < Grum> they can be changed?
08:52 < Spurlex> No the alts usually have security questions on them
08:52 <+SirCmpwn> that's not really Grum's responsiblity, is it?
08:52 <+SirCmpwn> send it to tobias
08:52 <+SirCmpwn> (it's tobias, right?)
08:52 < dx> mollstam?
08:52 <+SirCmpwn> yeah
08:52 < Spurlex> Stops the password changing which means they can be shared without people being stubborn and changing the password
08:52 <+SirCmpwn> we should all demand two factor auth
08:52 <+SirCmpwn> no playing minecraft unless you have your RSA key fob handy
08:52 < Grum> agreed
08:52 < SinZ> mollstam was just website, grum and dinnerbro did the client afaik
08:53 < dx> who's launcher?
08:53 <+SirCmpwn> SinZ: right, but I think mollstam owns the auth API
08:53 < Spurlex> I’d love 2step auth but I don’t think this type of community would use it widely
08:53 < Spurlex> because young
08:53 <+SirCmpwn> Spurlex: that was sarcasm
08:53 <+SirCmpwn> apparently I'm bad at it
08:53 < Spurlex> 2step auth would be nice though
08:53 <+SpaceManiac> 2-factor auth that works with mobile app like everyone sane uses would be neat, though probably not widely-used
08:54 < dx> well it's always optional and doesn't hurt to have it
08:54 <+SirCmpwn> god, it was a joke
08:54 <+SirCmpwn> I'm sorry
08:54 < SinZ> or something similar to steamguard, that new computers need a code from email or something
08:54 < dx> SirCmpwn: :D
08:54 < Spurlex> Steamguard would be great
08:54 < dx> SirCmpwn: but really, there are tinfoilhats out there that will appreciate it
08:54 <+SirCmpwn> I wear a tinfoil hat 24/7 but I only care about when it's valuble
08:54 < SinZ> iirc, Steam is actually getting mobile 2-factor auth soon-ish
08:55  * SirCmpwn types from his full disk encryption Linux system
08:55 < Spurlex> Minecraft just needs to pass some sort of un-changeable machine id to server
08:55 < Spurlex> ;)
08:55 < dx> SirCmpwn: some people have valuable minecraft accounts, like, uh, notch
08:55 < SinZ> that is banned on almost all servers?
08:55 <+SirCmpwn> notch can also go kick mollstam's chair and have him fix it
08:55 < dx> SinZ: what a terrible minecraft account
08:56 < dx> but now seriously, i can see a bunch of famous youtubers needing stronger auth
08:57 < Grum> or they just need stronger passwords
08:57 <+SirCmpwn> ^
08:57 < Grum> just pick something 32+ chars long and no-one will ever guess it
08:58 < Grum> *EVER*
08:58 < Grum> well not true
08:58 < Grum> ever is a long time, with enough monkeys it will happen
08:58 <+SirCmpwn> the auth servers should have rate limiting and such
08:58 < SinZ> I still hate that windows has a limit on the length on passwords
08:58 < dx> your bank account password must be between 4 and 6 digits
08:59  * dx has gone full offtopic
08:59 < GauHelldragon> security question: what is your last name
08:59 < SinZ> security question: what is your password
08:59 < Grum> GauHelldragon: you cannot actually answer that question
09:00 < Grum> unless you can see the future
09:00 < dx> mind=blown
09:00 < GauHelldragon> i get it
09:05 <+SirCmpwn> security question: can you see the future
09:06 <+SirCmpwn> Grum: after our discussion I added a note to the post on reddit that started this all
09:06 < Expel> SirCmpwn: No.
09:18 < Grum> Has URL SirCmpwn? 
09:18 <+SirCmpwn> http://redd.it/2fuhj7
09:19 < Grum> Slightly misrepresented but OK :) 
09:19 <+SirCmpwn> oh?
09:19 <+SirCmpwn> meh, whatever
09:20 <+SirCmpwn> thanks for talking about all this seriously, Grum, I'm much more optimistic about the future now
09:20 < Grum> In the process would indicate that my current commits would be touching subjects we talked about. They are not exactly :) 
09:20 < Grum> But as said, fine with me
09:21 <+SirCmpwn> eh, tiny detail
09:21 < dx> well it's mostly long term stuff
09:21 <+SirCmpwn> thinking about how the changes should take shape is a part of making them happen
09:21 < Grum> But I've done that 2 yrs ago
09:22 <+SirCmpwn> yes, but not with me providing input :P
09:22 < Grum> Hehe
09:23 <+SirCmpwn> I think I would be happy if Mojang just brought forth the big changes they had in mind for feedback before finalizing them
09:23 <+SirCmpwn> sounds like you're on the right track with the level format
09:23 < Grum> I don't know a lot of how I want to do it
09:24 <+SirCmpwn> that's fine, too - we can help you get the right ideas
09:24 <+SirCmpwn> working in a black box makes everyone sad
09:24 < Grum> In theory using something like a key value store as leveldb would work quite well too
09:25 <+SirCmpwn> if you'd consider much deeper changes, Minecraft could get really good in terms of technical design
09:25 <+SirCmpwn> if you don't see that as in scope right now, then there are still improvements that can be made
09:25 <+SirCmpwn> but it's a bit too late to start this discussion up again, I need to sleep
09:25 < Grum> Many options,  first step is making it future compatible and adding clear separations between layers
09:25 <+SpaceManiac> G'night, Cmpwn
09:25 <+SirCmpwn> night, SpaceManiac 
09:26 < Grum> Aka adding a y component to region files
09:26 <+SirCmpwn> or knight
09:26  * SirCmpwn cringes
09:26 <+SpaceManiac> :P
09:26 <+SirCmpwn> a long time ago I would say "good knight" on IRC and think I was funny
09:26 < Grum> And introduce readers and writers for chunkformats
09:26  * SirCmpwn isn't even remotely into medival culture
09:27 < Grum> Go play with your sword and nap well
09:29 < dx> ooooh now i get why it's called KnightOS
10:20 < Dinnerbone> Hi SirCmpwn, I don't know why you think that I'm "ruining the game" but I haven't added anything for sethbling in a very, very long time (ever? I can't remember anything. Maybe a bug fix before). If you are referring to lots of mapmaker stuff that was added in the update half dedicated to map maker stuff, that was mostly not me too. Sorry!
10:22 < jast> I think there's some kind of misunderstanding here. the purpose of a scapegoat is to draw flak for everything, whether they were involved or not. ;)
10:25 < SinZ> so we are saying that Jeb is ruining Minecraft now?
10:27 < jast> doesn't matter as long as someone gets blamed for something
10:51 < shoghicp> heh, such a night
10:51 < shoghicp> Most of the conversation overflowed the backlog
10:51 < SinZ> lol
10:56 < shoghicp> about authentication, Minecraft: PE doesn't have it for normal servers
10:57 < shoghicp> only their Realms servers will get nick authentication :(
10:57 < shoghicp> well, that aside
10:58 < shoghicp> I'm sending the Chunk Data
10:58 < shoghicp> but nothing renders on the client side
10:58 < shoghicp> do I have to send an ID mapping? if yes, how?
10:58 < shoghicp> oh, wait
10:59 < shoghicp> little endian...
10:59 < shoghicp> why would you put that there when everything else is big endian...
11:00 < shoghicp> well, at least this is something normal in Minecraft: PE
11:00 < shoghicp> NBT is little endian, but network is big endian
13:16 < UltimateBudgie> http://wiki.vg/Protocol reflects the 1.8 protocol yeah?
13:17 < Thinkofdeath> yep
13:28 < UltimateBudgie> Awesome :D Thanks
13:30 < Grum> md_5: can you prod me when you wake up? cheers!
13:30 <+md_5> Grum hey!
13:30 < Grum> HAI THAR
15:49 <+SirCmpwn> Dinnerbone: I've been led to believe that you're the driving force behind this command nonsense that's getting out of control. If that's not the case, then I apopogise for my comments earlier.
16:15 < TkTech> Note: Opinions are opinions.
16:15 <+SirCmpwn> very insightful
16:53 < bgale_> SirCmpwn, what's the "command nonsense"?
16:53 < jast> opinions are facts!
16:54 < bgale_> opinions are proof of god
16:54 < bgale_> or something
16:59 < Fenhl> opinions are laser beams attached to a shark
17:16 <+SirCmpwn> bgale_: the extentions on commands and command blocks that they have been spending so much of their time on
17:17 < bgale_> didn't you realise command blocks are the mod api?
17:19  * SirCmpwn shudders
17:50 < Grum> SirCmpwn: not much time has been spent on it
17:51 < Grum> just some spare hours here and there by mostly Searge
17:53 < Bibl> back from school
17:53 < Bibl> what have i missed
17:53 <+SirCmpwn> well, this is all super subjective at this point
17:53 < Grum> maybe ~5% of the time is spent on commands from 1 person? if even that
17:54 <+SirCmpwn> that's good to hear
17:54 <+SirCmpwn> my complaint is less about the time and more about just the fact that they exist, but I don't really want to get too into it because it's an opinion I don't feel like I should be pushing for
17:57 < Grum> it's "ok" for now :p
17:57 < Grum> should become a plugin sooner than later
18:13 < Fenhl> is “everything is a plugin” still the ultimate plan?
18:17 <+SirCmpwn> what if the plugin API was a plugin
18:17 <+SirCmpwn> *dramatic music*
18:18 < shoghicp> Bukkit API as a plugin :P
18:18 < Grum> that might work if we were to make a bukkit-like api :p
18:25 < Fenhl> :O
18:25 < Fenhl> the possibilities are endless
18:38 < Aragas> I get disconnected very quickly, i think after SetCompression packet. Do i need to answer to it? or i get disconnected because i don't compress packets?
18:55 <+SpaceManiac> probably because you don't compress packets (you should at least switch to the header form used when compression is enabled)
18:59 < expir3dcowPC> hi
19:01 < Aragas> man, this is soooo lame
19:38 < Aragas> so i need always send packets with headers? Can i fully disable compression? Wanna debug packets, but don't wanna make compression support
19:48 < shoghicp> Aragas: that is decided by the server
19:48 < shoghicp> you can set network-packet-compression to -1
19:48 < shoghicp> that'll disable it
20:00 < Grum> you do not quite want to do that though
20:00 < Grum> chunk packets are HUGE without compression
20:02 <+SpaceManiac> flat world near y=0, 1 view distance? probably still big
20:02 <+SirCmpwn> did you remove the usual compression for chunk packets, Grum?
20:02 < Grum> yes
20:02 <+SirCmpwn> good choice
20:02 < Grum> we do no compression at all anymore in 1.8
20:02 < Grum> except the global one
20:02 <+SirCmpwn> I like that change ^_^
20:02 < Grum> it also means that any packet under the threshold *HAS* to be compressed
20:03 < Grum> and the server will boot you if you ignore this
20:03 <+SirCmpwn> over
20:03 < Grum> err over yes
20:03 < Grum> this might mean the size will end up bigger than before, but that is ok :)
20:03 <+SirCmpwn> what was the justification for that?
20:03 <+SirCmpwn> I imagine a particularly clever implementation would ditch the compression if it proved ineffective at runtime
20:04 < Grum> and spam the server with data?
20:04 <+SirCmpwn> what does the client send that is so heavy to be of consequence
20:04 < Grum> it's just strict, you make an agreement, you keep that agreement :)
20:05 <+SirCmpwn> it'd be trivial to check the size and discard the compressed packet if larger, though, with a net reduction on bandwidth costs
20:05 < shoghicp> Plugin messages, maybe? but they can be compressed by the plugin itself
20:05 < Thinkofdeath> Grum: And if you ever try to change it the client crashes
20:05 < Thinkofdeath> :)
20:05 < Thinkofdeath> s/crashes/disconnects with an error
20:07 < Grum> that would increase the logic significantly, right now it barely has any logic :p
20:07 <+SirCmpwn> would it really increase the logic significantly?
20:07 <+SpaceManiac> the d/c with error is a copy/paste bug afaik
20:07 <+SirCmpwn> if (compresesd.length > original.length) compressed = original;
20:07 < shoghicp> ^
20:07 <+SirCmpwn> I haven't seen the code but just imagining how I would implement it is pretty simple
20:08 < Grum> there was some reason
20:08 < Grum> we had that first
20:09 <+SirCmpwn> is it "we have to somehow tell the other side that it's not compressed"
20:09 <+SirCmpwn> cause that would make sense as a reason not to bother
20:09 <+SirCmpwn> but I imagine no one realized that gzip/zlib has magic numbers
20:11 < Grum> oh! i think it was because we wanted people to experiment with the values
20:12 <+SirCmpwn> to find a good threshold?
20:12 < Grum> and if we start to fudge internally numbers are always messed up
20:13 <+SirCmpwn> does anyone know what vanilla uses as the threshold, by the way
20:13 <+SirCmpwn> I would just try to compress every nontrivial packet (like >100 bytes) and see what sticks
20:13 <+SpaceManiac> I think 256 is the default?
20:13 < shoghicp> yep
20:17 < Grum> the thing is, it's hard to see what sticks and what doesn't
20:17 <+SirCmpwn> it's pretty objectively easy to check the lengths and send whichever is smaller
20:18 < TkTech> SirCmpwn: The couple byte difference would hardly make that additional logic worthwhile.
20:18 <+SirCmpwn> you're probably right, this is mostly speculation
20:19 < Grum> also you just wasted time to compress it
20:19 < Grum> might a well send it :P
20:19 < Grum> and if its bigger, well, you picked a stupid threshold
20:19 <+SirCmpwn> I'm thinking of this mostly for research purposes to determine what kinds of packets and data it's worth compressing
20:19 <+SirCmpwn> rather than putting this into actual production
20:20 < Grum> in that case you want to actually be able to see which data is 'bigger' after compression
20:20 < Grum> and for that you need the compressed data
20:20 <+SirCmpwn> ?
20:21 < Grum> (that is bigger)
20:22 <+SirCmpwn> ??
20:23 < TkTech> SirCmpwn: You need both compressed and uncompressed traffic to compare the two.
20:27 <+SirCmpwn> well, yes
20:27 <+SirCmpwn> that'd be easy to track, though
20:31 < shoghicp> besides that, what would be the max overhead?
20:31 < shoghicp> 11 bytes?
20:32 <+SirCmpwn> there's nontrivial CPU overhead when encoding packets
20:32 < shoghicp> right, but that cannot be guessed before :S
20:33  * SirCmpwn shrugs
20:33 < shoghicp> hmm, this could use the client-server MTU for the threshold
20:34 <+SirCmpwn> that's a good idea
20:34 < shoghicp> Minecraft sets TCP_NODELAY, right?
20:46 < Aragas> shogchip: thanks
20:47 < shoghicp> I should really disconnect the shogchips nick from the bouncer
20:47 <+SirCmpwn> oh, hey shoghicp 
20:47 < shoghicp> o/
20:48 <+SirCmpwn> I was talking about you and how insane you are on reddit the other day
20:48 < Aragas> lol, ops
20:48 < Aragas> sorry
20:48 < shoghicp> oh, where?
20:48 <+SirCmpwn> https://www.reddit.com/r/admincraft/comments/2fuhj7/i_am_a_minecraft_reverse_engineer_here_to_weigh/ckd928d?context=3
20:48 <+SirCmpwn> some idiot wanted to tell me all about why PHP doesn't suck
20:48 < shoghicp> I only read the first comments there :s
20:49 < shoghicp> haha
20:50 < shoghicp> I could quote my works from here, but I'm too lazy to open the logs
20:50 <+AndrewPH> SirCmpwn: I'm confused as per why he thinks doing things that are much more difficult make PHP not shit
20:50 <+SirCmpwn> the sad part is that he names all this stuff that supports his PHP code, none of which is written in PHP
20:50 <+AndrewPH> it's not that you can't scale it, it's that it's a shitty language for that purpose.
20:51 <+SirCmpwn> you have to write similar supporting code if you want to use PHP for a Minecraft server, it's just ridiculous to write that kind of low level stuff in PHP
20:51 <+AndrewPH> also citing facebook as a user of php is about as accurate as calling scala java.
20:52 <+AndrewPH> they have a (public) compiler that compiles the php to like C or something
20:52 <+SirCmpwn> did I mention that facebook hired the best engineers in the world and spend a fuckload of money on making their site work and at scale
20:52 <+AndrewPH> that too
20:52 <+SirCmpwn> and that Minecraft has the intrinsic problem of being very difficult to load balance
20:53 <+SirCmpwn> the typical PHP solution of "just throw more hardware at it" is hard to do in minecraft
20:53 <+AndrewPH> his response basically comes down to "run more servers!!"
20:53 <+SirCmpwn> unfortunately that thread changed the tone of the entire post
20:53 <+SirCmpwn> he's a moderator of that subreddit, he should know better than to derail it like that
20:53 < shoghicp> Thanks for the "It's actually surprisingly polished" part :)
20:54 <+SirCmpwn> shoghicp: sure
20:54 <+AndrewPH> also while I have a general disagreement with many things you say, the downvotes in this case are unwarranted
20:55 < shoghicp> well, I rewrote most of the code, and surprisingly, it has time to spare
20:55 < shoghicp> with all the physics and alike
20:55 <+SirCmpwn> if you rewrote it all, why not take the opportunity to do it in a less obnoxious language
20:56 < shoghicp> hmm, I could say it was for compatibility, but later all the old plugins were outdated
20:57 <+SirCmpwn> you probably just don't have the confidence to build it in something you're less familiar with, right
20:57 < shoghicp> yep, before doing this, I wrote three clients
20:58 < shoghicp> before I could start doing more advanced stuff
20:58 < shoghicp> I started it once in Python
20:58 <+SirCmpwn> python users of #mcdevs, avert your eyes
20:58 <+SirCmpwn> I've been using Python a lot lately and I really like it, but I don't think it's suitable for Minecraft servers
20:59 < shoghicp> I only finished the binary reading part
20:59 < shoghicp> well, there are lots of projects that have started around pocketmine
20:59 < shoghicp> a few done in java
20:59 < shoghicp> that port the code in PHP to Java
20:59  * SirCmpwn sighs
20:59 <+SirCmpwn> why is Java so pervasive in the minecraft community
20:59 <+SirCmpwn> god damn you, notch
21:00 < shoghicp> well, at least I moved world generation to a different thread, same with networking
21:00 < shoghicp> then, I find bugs in the PHP extensions that couldn't be fixed until three days ago >.<
21:01 < shoghicp> oh, and LevelDB
21:01 < shoghicp> Minecraft: PE now uses LevelDB as the format to save worlds
21:02 < shoghicp> and they open-sourced their modifications to that lib to support deflate instead of snappy
21:02 < shoghicp> if anyone is interested, https://github.com/Mojang/leveldb-mcpe
21:03 < shoghicp> SirCmpwn: also, did you hear about BigBrother?
21:03 < shoghicp> more PHP madness, of course.
21:03 <+SirCmpwn> shoghicp: link
21:03 < shoghicp> https://github.com/shoghicp/BigBrother
21:03 <+SirCmpwn> that's pretty cool
21:04 <+SirCmpwn> have you seen RetroCraft?
21:04 < shoghicp> it's a plugin for PocketMine servers, and works as a translator
21:04 <+SirCmpwn> https://github.com/SirCmpwn/RetroCraft
21:04 < shoghicp> I heard about it on the reddit post :)
21:04 <+SirCmpwn> I would show more interest in BigBrother but for obvious reasons I can't seem to muster the enthusiasm
21:05 < shoghicp> I guess
21:06 < shoghicp> but heh, it works better than I expected
21:14 < shoghicp> well, I guess that's enough PHP on #mcdevs for today
21:14 < Aragas> more problems ahead. is there any information about nbt (Slot here) data? In the wiki it isn't full. I mean, i dunno what to read there
21:15 < Aragas> Because old reading don't work, something has changed
21:15 <+SirCmpwn> the NBT data format has changed or the schema of the items it represents has changed
21:15 < shoghicp> in 1.8?
21:15 < Aragas> yep. 1.8
21:16 < Aragas> it doesn't works after Metadata reading
21:17 < Aragas> length reading is now varInt i think, but it doesn't help
21:18 < TkTech> SirCmpwn: NBT on disk or just in a stream?
21:18 < TkTech> Guess I have to update PyNBT if it changed on disk
21:18 <+SirCmpwn> highlighted the wrong person
21:18 <+SirCmpwn> well, I suppose my previous statement could be clearer if I added a ?
21:18 < Aragas> if u mean me, then stream
21:19 < TkTech> SirCmpwn: Turning it into a question does indeed make that clearer.
21:19 <+SpaceManiac> TkTech: as an aside, I think Notifico is kill on Esper again
21:23 < Aragas> Thats not funny, a lot of packets read these nbt
21:24 <+SpaceManiac> Aragas: you said you're trying to read the Slot data type and having troubles in 1.8?
21:25 < Aragas> SpaceManiac: yes
21:26 < Aragas> can't find Slot data structure from 1.8
21:27 <+SpaceManiac> my impl: https://github.com/SpaceManiac/Glowstone/blob/master/src/main/java/net/glowstone/net/GlowBufUtils.java#L141
21:27 <+SpaceManiac> may help you see what's different since 1.7
21:28 <+SirCmpwn> afk for a bit
21:28 < Aragas> totally works for me. Thanks
21:29 < Fenhl> in the slot format, is item ID still the old numeric ID?
21:29 <+SpaceManiac> yes
21:30 < Fenhl> hm. I'm assuming items that have a different block and item ID use the item ID?
21:30 <+SpaceManiac> the id numbers in Slot data are as they were in 1.7
21:31 < Fenhl> okay, that would imply item ID
21:32 < Aragas> SpaceManiac: I think it worked. no more reading errors
21:33 < Aragas> wow. lol. i have no more errors at all
21:34 < TkTech> SpaceManiac: BLARGH
21:34 <+SpaceManiac> TkTech: oh, take me with a grain of salt though.
21:34 <+SpaceManiac> My channels tend to end up on the overflow bot
21:35 < TkTech> SpaceManiac: Bots murdered.
21:35 < TkTech> SpaceManiac: A decent number of contributions, but *no one* wants to touch the IRC protocol bits
21:35 <+SpaceManiac> Hehe.
21:35 < expir3dcowPC> I do
21:35 < TkTech> I'm probably going to say screw performance and switch to Twisted.text
21:36 < TkTech> r, twisted.words
21:36 < TkTech> *Er
22:22 < Not-3b7> [Glowstone] SpaceManiac pushed 2 commits [+0/-0/±4] http://git.io/uUfQfw
22:22 < Not-3b7> [Glowstone] SpaceManiac 3b50ad7 - Load mob spawners and NBT lists more cleanly (fixes #135).
22:22 < Not-3b7> [Glowstone] SpaceManiac 37b40f6 - Recount non-air after chunk generation (fixes #239).
23:12 < Aaron1011> Is there a good way of having the client print out invalid packets it receives?
23:13 < Aaron1011> I'm pretty sure I have everything correct for Block Entity Update, but the client isn't changing the skull
23:14 < dx> Aaron1011: already done the equivalent of this for the client? http://wiki.vg/Debugging
23:15 < Aaron1011> dx: No, I was about to set that up
23:16 < Aaron1011> Do you know of a way I can filter that down?
23:16 < Aaron1011> Actually, I could just do a grep
23:16 < Aaron1011> nvm
23:16 < dx> oh that page also mentions how to do it for the client, neat
23:22 < winny> belak: did you ever figure out what the problem was with mcstat?
23:23 < Aaron1011> dx: Do you how you can actually display the packet data?
23:23 < dx> AndrewPH: like after decryption but before parsing?
23:24 < Aaron1011> dx: Did you mean me?
23:24 < dx> derp
23:24 < dx> yes
23:24 < dx> AndrewPH: sorry
23:24 < Aaron1011> That autocomplete :)
23:24 < Aaron1011> I do it all the time :)
23:24 < Aaron1011> Yeah, that's what I'm looking fo
23:24 < Aaron1011> for^^
23:24 < Aaron1011> Right now, I'm just seeing two-letter sequences
23:25 < Aaron1011> Not even the data displayed on the example page
23:28 < Aaron1011> dx: Do you have any ideas?
23:29 < dx> NOPE
23:29 < Aaron1011> Okay, thanks :)
23:29 < dx> not showing the arguments definitely sounds like something is very wrong, though
23:30 < dx> Aaron1011: what client version, btw?
23:30 < Aaron1011> 1.8
23:30 < Aaron1011> I'm trying to get Update Block Entity working in Glowstone
23:31 < Aaron1011> I've confirmed that the packet is being sent, but that's about it
23:31 < dx> ah, glowstone, neat
23:31 < Aaron1011> Yeah, I'm implementing skulls
23:32 < Aaron1011> I've done enough server-side (I think) to get different types of heads displaying
23:32 < Aaron1011> So far, I've managed to get them to face properly when placed on the sides of blocks
23:32 < Aaron1011> I haven't tackled placing them on the ground yet, as that also depends on this packet
23:32 < Aaron1011> As do player skins on heads
23:35 < dx> i think i found a tiny bug in the launcher
23:36 < dx> separating JVM arguments with two spaces results in "Error: Could not find or load main class"
23:36 < Aaron1011> Yeah, I was getting that too
23:36 < dx> heh
23:36  * dx searches jira
23:40 < dx> no open issues
23:47 < Aaron1011> Has anyone deobfuscated the 1.8 server yet?
23:47 < Aaron1011> I'd seen Searge tweeting about working on MCP
23:49 < dx> deobfuscation is for the weak
23:49 < Aaron1011> dx: :)
23:50 < dx> also you probably should avoid looking at the mojang code while contributing to glowstone
23:51 < Aaron1011> Yeah, I was worried about that
23:51 < Aaron1011> I'm not copying anything though - just checking to make sure wiki.vg is write
23:51 < Aaron1011> right^^
23:51 < Aaron1011> I was worried I had the packet format/NBT data for the packet wrong
23:51 < dx> what part of the wiki? link to section?
23:52 < Aaron1011> http://wiki.vg/Protocol#Update_Block_Entity
23:52 < dx> hm
23:55 < dx> ah, barneygale's "quarry" library can work as a proxy
23:55 < dx> could use that to get some meaningful debug
--- Day changed mer. sept. 10 2014
00:15 < barneygale> yes
00:15 < barneygale> I should add a logging proxy example
00:15 < dx> already doing it
00:15 < barneygale> :D
00:16 < barneygale> dx: this is in there but not used
00:16 < barneygale> https://github.com/barneygale/quarry/blob/master/quarry/util/dispatch.py#L28
00:16 < barneygale> produces `hexdump -C` style output
00:16 < dx> excellent.
00:20 <+Amaranth> http://online.wsj.com/articles/microsoft-near-deal-to-buy-minecraft-1410300213
00:20 < barneygale> paywalled
00:21 < barneygale> $2bn tho wow
00:21 < dx> i bet it's like those "CONFIRMED" reports of google buying twitch
00:21 < dx> s/bet/hope/
00:22 <+Amaranth> Pretty sure Google was actually doing that but backed out due to anti-trust concerns
00:22 < Aaron1011> I thought it was Amazon
00:22 < dx> yes, but journalists still claimed it was a confirmed closed done deal
00:22 < dx> Aaron1011: it turned out to be amazon, but before that google was considering it, or at least that's what those "journalists" claimed
00:25 < dx> >As the Journal notes, the move would be surprising given that Mojang founder Markkus "Notch" Persson, is a darling of the indie gaming scene who has avoided taking venture capital. He has also criticized Microsoft's Windows 8 for potentially being "very, very bad for indie games."
00:25 < dx> evidently notch is dead
00:27 < barneygale> I for one welcome our new corporate overlords etc etc
00:27 < dx> yeah! what could possibly go wrong
00:27 < dx> >Twist: It's actually Amazon, not Microsoft
00:28 < barneygale> controversy over open source? disillusioned early adopters? maligned 3rd party developers? sounds like a perfect microsoft asset
00:29 < dx> nah man, i'm telling you, it's amazon
00:29 < dx> it's always amazon
00:29 < dx> amazon actually owns google *and* microsoft
00:30 < Aaron1011> And apple
00:30 < dx> on a serious note, microsoft may be forced to take the whole legal mess of modding more seriously
00:31 < Aaron1011> dx: I'm personally worried about Mojang dropping support for Linux/OS X
00:31 < dx> Aaron1011: meh, unlikely. skype still supports linux
00:32 < Drainedsoul> HyperV supports linux guests
00:32 < dx> lol
00:32 < Aaron1011> That's true
00:32 < Aaron1011> dx: Bizarrely, Skype for linux got *more* updates after Microsoft bought them
00:33 < dx> also, http://www.microsoft.com/openness
00:33 < dx> and https://github.com/microsoft
00:33 < dx> being open source friendly counts as positive PR nowadays
00:34 < Aaron1011> The cynical part of me notes how few repos are there :) But, I'll admit that Microsoft has become a lot more open
00:34 < dx> they only started the move to github recently
00:34 < Drainedsoul> Roslyn isn't on their github? :(
00:34 < dx> they've had their own shitty open source code hosting site for a long time, https://codeplex.com/
00:47 <+fragmer> I wonder how many Java developers Microsoft has
00:48 < SinZ> Aaron1011: https://github.com/Microsoft/microsoft.github.io/blob/master/README.md
00:48 < jython234[away2]> Lol
00:48 < Aaron1011> I admit my anti-Microsoft bias :)
00:49 < SinZ> running hexchat on linux, I can see that
00:50 < Aaron1011> :)
01:01 < cindy_k> J++
01:01  * cindy_k runs
01:01 < cindy_k> ohh and I forgot J#
03:13 < jython234> a higher form of spam
03:13 < dx> dammit jython234, not again
03:13 < dx> go do something else instead of telling everyone how ajf has a shitty internet connection
03:15 < humerusj> ^
03:17 <+SpaceManiac> fun in join-messages-disabled lancd
03:18 < dx> in other news: "The only way this could get more crazy is if SÄPO (Swedish version of FBI) broke down the Mojang AB offices and arrested them all on suspicion of money laundering and cocaine possession."
03:22 < nickelpro> When did Spawn Player change again? We ditched the weird properties structure already? I thought that was for the new skins system, did we ditch that too?
03:23 < Drainedsoul> Don't you get the "weird properties structure" through the Mojang API?
03:24 < nickelpro> It used to be sent through the Spawn Player packet too
03:24 <+SpaceManiac> those have migrated to the user list
03:25 < nickelpro> SpaceManiac: Lovely, thanks
03:25 < Drainedsoul> that massive abortion of a packet
03:25 < nickelpro> Oh fuck
03:26 < nickelpro> Someone had an accident in the MC protocol
03:26 < nickelpro> Becuase that's a piece of shit
03:26 < Drainedsoul> it'd be much easier to swallow if it was an accident
03:26 < Drainedsoul> but World Border is just as bad/worse
03:26 < Drainedsoul> i.e. it wasn't an accident
03:26 < nickelpro> These are laughably terrible
03:26 < Drainedsoul> yupp
03:27 < nickelpro> The "chat" type is your usual varint prefixed string, except that string is JSON chat, right?
03:27 < Drainedsoul> afaik, yes
03:27 < Drainedsoul> I hope so, that's how I've been implementing it :x
03:27 <+SpaceManiac> correct
03:28 < Drainedsoul> std::optional<JSON::Value> DisplayName;
03:28 < nickelpro> Ha, you and your colons
03:28 < nickelpro> Glorious C doesn't need colons
03:28 < Drainedsoul> "glorious C" doesn't have std::optional either
03:28 < nickelpro> shush
03:29 < Drainedsoul> I implemented that packet using a nice variant type :)
03:29 < Drainedsoul> owait, C doesn't have that either :(
03:29 < nickelpro> int32_t
03:29 < nickelpro> Can barely tell the difference
03:29 < Drainedsoul> wot
03:29 < Drainedsoul> a variant is like, a tagged union
03:30 < nickelpro> Oh I thought you wrote varint