The Communication Protocol for Eulora2, Restated



June 13th, 2023 by Diana Coman


As the game's development is progressing beyond the emergence of time, past the addition of vitals and into enabling player activities, all the design and implementation work done so far feeds as well into clarifying and refining the game-related parts of the communication protocol that were previously tentatively and incompletely stated1. Specifically, it's again that 'Character Actions' section 7 that comes into focus but this time the changes go beyond just simple additions and move things forward from where the previous effort towards specification stopped. Consequently, I have to bring together the different parts and restate the current take on the protocol in full, in here, where it can be further iterated and discussed as needed:


This is the current take on Eulora2's communication protocol, last revised June 13th, 2023.

1. Overall Goals:

  • 1.1. All communications between clients and server to be encrypted.
  • 1.2. Clients to be able to receive from server any data they lack (including maps, skins, sound or video content etcetera), on demand.
  • 1.3. Clients to be able to choose and adjust both the level of security and their volume of communications with the server, as they will ultimately have to pay for the load that they generate.

2. Explicit Dependencies :

3. Data Structures :

    3.0. Basic types :

    • char / uint8 (1 byte) ;
    • uint16 (2 byte) ;
    • uint32 (4 byte) ;
    • uint64 (8 byte) ;
    • float2 (4 byte) ;

    3.1. Special types:

    • hash (128 bits) ;
    • chunk [of file] (bitfield, 11760 bits) ;
    • serpent-packet (1472 bytes) ;
    • rsa-message3 (1872 bits4) ;
    • rsa-packet5 (1470 bytes) ;
    • object (size of 104 bits6: uint327 followed by 3 uint16s representing position8 followed by 3 uint8s representing rotation9 ) ;
    • legacy-text (size of n+n/256+1 bytes ; where the leading byte is the bytecount of the 2nd segment and the 2nd segment is the bytecount of the third segment)10.
    • text (2 byte hearder containing the ~total~ byte length ; up to 1470 bytes of text ).

4. Serpent Packets11 :

    4.1. Serpent Key Set:

    • uint8 (type ID, =100), followed by
    • uint8 (count of keys in this set, n), followed by
    • n*(4*int64 + uint32) (32 bytes each key followed by a 4 byte ID calculated through crc3212 ), followed by
    • an uint8 flag (LSB bit set -- keys to be used to talk to client ; MSB set -- key to be used to talk to server ; client-set MSB is ignored), followed by
    • uint16 (message count13), followed by
    • padding to Serpent-message length.

    4.2. Serpent Keys Lifecycle Management:

    • uint8 (type ID, =102), followed by
    • uint8 (count of server keys requested), followed by
    • uint8 (count of client keys requested), followed by
    • uint8 (id14 of serpent key preferred for further inbound Serpent-messages), followed by
    • uint8 (count of burned keys in this message), followed by
    • n*int8 (id of burned key), followed by
    • uint16 (message count), followed by
    • padding to Serpent-message length.

    4.4.a. File Request, manifest

    • uint8 (type ID, =3), followed by
    • hash (corresponding to the sought file15), followed by
    • uint8 (manifest packets sought count, 0=all), followed by
    • n* uint16 (manifest packet index sought), followed by
    • padding to Serpent-message length.

    4.4.b. File Transfer, manifest (always sent and only sent in response to ID 3)

    • uint8 (type ID, =4), followed by
    • uint16 (count of manifest packets for this file16), followed by
    • uint16 (index of current packet in list above), followed by
    • uint8 (fragment count17), followed by
    • n* uint64 (hash of the nth fragment of manifested file).
    • uint16 (keccak hash of foregoing), followed by
    • padding to Serpent-message length.

    4.4.c. File Request, chunks

    • uint8 (type ID, =5), followed by
    • hash (corresponding to the sought file), followed by
    • uint8 (file chunks sought count), followed by
    • n* uint64 (the hash of fragment sought), followed by
    • padding to Serpent-message length.

    4.4.d. File Transfer, non-last chunk (always sent and only sent in response to ID 5)

    • uint8 (type ID, =6), followed by
    • chunk.

    4.4.f. File Transfer, last chunk (sent at most once per ID 3)

    • uint8 (type ID, =7), followed by
    • uint16 (bytesize of useful part of the chunk following18, followed by
    • chunk19.

    4.5. Client Action20 :

    • uint8 (type ID, =8), followed by
    • text (fully specified action, see section 7), followed by
    • uint16 (message count), followed by
    • padding to Serpent-message length.

    4.6. World Bulletin21:

    • uint8 (type ID, =9), followed by
    • uint32 (id of top level item22), followed by
    • uint8 (count of objects), followed by
    • object list23, followed by
    • uint16 (message count), followed by
    • padding to Serpent-message length.

    4.7. Object Request:

    • uint8 (type ID, =10), followed by
    • uint8 (count of objects), followed by
    • n*int32 (id of object), followed by
    • uint16 (message count), followed by
    • padding to Serpent-message length.

    4.8. Object Info:

    • uint8 (type ID, =11), followed by
    • uint8 (count of objects), followed by
    • n times uint32 (id of object) and text (object properties, as per extant game structures, including art files needed and so on24), followed by
    • uint16 (message count), followed by
    • padding to Serpent-message length.

    4.9 Line Request

    • uint8 (type ID, =12), followed by
    • uint8 (count of lines), followed by
    • n times uint32 (id of chatroom), uint64 (lowest line number requested), uint64 (highest line number requested), followed by
    • padding to Serpent-message length

    4.10 Line Info

    • uint8 (type ID, =13), followed by
    • text (line content), followed by
    • uint32 (chatroom ID), followed by
    • uint32 (speaker ID), followed by
    • uint64 (line number), followed by
    • uint64 (keccak hash of the line's content and number), followed by
    • padding to Serpent-message length

    4.11 Rating Request

    • uint8 (type ID, =14), followed by
    • uint8 (count of requests), followed by
    • n times uint32 (source ID), uint32 (target ID), followed by
    • uint16 (message count), followed by
    • padding to Serpent-message length

    4.12 Rating Info

    • uint8 (type ID, =15), followed by
    • text(rating comment25, followed by
    • uint32 (source ID), followed by
    • uint32 (target ID), followed by
    • uint8 (rating value26 ), followed by
    • uint16 (message count), followed by
    • padding to Serpent-message length

5. RSA Packets27 :

    5.1. RSA key set28.

    • uint8 (equal to 251 to indicate packet contains a new RSA key), followed by
    • uint8 (protocol version), followed by
    • uint16 (subversion), followed by
    • uint32 (IP of server29), followed by
    • uint32 (IP of client30), followed by
    • uint64 (keccak hash of client binary), followed by
    • uint64 (e of RSA key), followed by
    • uint8*490 (N of RSA key), followed by
    • uint64 (preferred padding -- the magic value of 0x13370000 requests random padding ; all other values will be used as such, bitwise, ie like an infinite-length OTP consisting of the value repeated), followed by
    • uint16 (message count), followed by
    • padding to RSA-message length, 1424 (5616-8-8-16-32-64-64-3920-64-16) bits exactly.

    5.2. Serpent key set31:

    • uint8 (equal to 157 to indicate packet contains new Serpent keys), followed by
    • uint8 (count of keys32 in this set, n; n<=1933), followed by
    • n*(4*int64 + uint32) (32 bytes each key followed by a 4 byte ID calculated as crc32 on the key itself), followed by
    • an uint8 flag (LSB bit set -- keys to be used to talk to client ; MSB set -- key to be used to talk to server ; client-set MSB is ignored by server ; server will set LSB on keys requested by client for its own use thus supporting clients with no trustworthy random generators of their own), followed by
    • uint16 (message count), followed by
    • padding to RSA-message length.

    6. Protocol Mechanics :

      6.0. All communications between server and client will consist of messages. These messages may be encrypted either via eucrypt.RSA or eucrypt.Serpent. All RSA-encrypted messages will be exactly 1`470 bytes in length ; all Serpent messages will be exactly 1`472 bytes in length34 . The server will handle Serpent messages in preference of RSA messages (which are processed on an as-available basis). Clients that send garbage will be punished ; the costs involved (encryption/decryption ; generating entropy ; lookups and whatnots) will be pushed onto the client, for which reason writing the clients lightly pays off.

      6.1 The handshake works as follows:

      • New client issues 5.1 packet keyed to the server's public key, including its own RSA key.
      • The client's IP is recorded, and will have to be explicitly changed by the client later if needed. Server replies with 5.1 packet keyed to the client's announced key, including its private RSA key for use by that client ; and with 5.2 packet containing key material for client's use in keying messages to the server. If the client fails to provide its own set of serpent keys, the server will further issue it a set of serpent keys ; thenceforth the server will send more serpent keys mirroring the client's supply, and will similarily mirror key burning and select operations on its own set.
      • Should the client's IP change, it will issue a 5.1 packet keyed to the server public key immediately followed by a 5.1 packet keyed to the server's original private key. The server will then update the client's IP accordingly (this also trashes the extant Serpent keyset and triggers 5.2).
      • The bulk of communication is intended to go through the Serpent system ; outside of identification and bootstrap handshakes RSA isn't used. Should either party believe the Serpent keysets've been FUBAR'd, a 5.2 packet will reset that keyset.

      6.2. The server will issue type 4.6 packets in response to relevant type 4.5 packets received -- these can either signify the acceptance or the rejection of the client action, and the client must adjust its internal state accordingly.

    7. Character Actions :

      7.0. Lock:

      • uint8 (type ID, =0), followed by
      • uint8 (count of objects), followed by
      • n* uint32 (object ids). Defaults to currently targeted item.

      7.1. Make35:

      • uint8 (type ID, =1), followed by
      • uint32 (object id, defaults to current target), followed by
      • uint32 (object id, defaults to currently equipped method), followed by
      • uint32 (object id, defaults to currently equipped tool), followed by
      • uint8 (count of objects). The order for producing count items is entered into the queue (note that crafting only progresses if the player and/or hireling NPCs find themselves in certain situations re equipment, position, etc).

      7.2. Attempt36:

      • uint8 (type ID, =2). This is a request to start the chosen activity (see next field), followed by
      • uint32 (object id, indicating the intended activity, defaults to explore37), followed by
      • uint32 (object id, defaults to current target or location, as relevant), followed by
      • uint32 (object id, defaults to currently equipped method, followed by38)
      • uint32 (object id, defaults to currently equipped tool)

      7.3. Exchange:

      • uint8 (type ID, =3), followed by
      • uint32 (object id, the other party), followed by
      • uint32 (object id, the trade itself39), followed by
      • uint8 (count of objects), followed by
      • n* uint32 (object ids) and uint64 (object count), followed by
      • uint8 (flag, set to 0x10 to lock a trade and to 0x0c to approve a trade previously locked by both players).

      7.4. Attack:

      • uint8 (type ID, =4), followed by
      • uint32 (object id, the other party), followed by
      • uint32 (object id, the battle itself (server set, exactly in the way trade works). This not currently implemented, except player setting itself the bomb results in instadeath.

      7.6. Move:

      • uint8 (type ID, =6), followed by
      • uint32 (destination id, defaults to current target), followed by
      • uint32 (slot id), followed by
      • uint32 (object id, of the item being moved), followed by
      • uint32 (quantity moved).

      7.7. Train:

      • uint8 (type ID, =7), followed by
      • uint32 (object id, the other party), followed by
      • uint32 (object id, the train session itself (server set, exactly in the way trade and battle work). This not currently muchly implemented, except some NPCs train for money -- but will get greatly expanded asap.

      7.8. Relocate:

      • uint8 (type ID, =8), followed by
      • object type, containing new client position.

      7.9 Say40:

      • uint8 (type ID, =9), followed by
      • text (line content), followed by
      • uint32 (room ID), followed by
      • uint64 (keccak hash of the line's content)

      7.10 Rate:

      • uint8 (type ID, =10), followed by
      • text (rating comment), followed by
      • uint32 (target ID)
      • uint8 (rating value)

      Please leave your comments below.


      1. This type of iterative refinement and interaction between design and implementation is neither new nor surprising to me, really. The only difference is that I'm doing now out of necessity all sides of this. Experience helps *a lot*, of course, even or perhaps especially beyond and above what one 'expects' or what one 'wants' or what one 'signed up for' and so on and so forth. It might even help in the sense that it makes it possible at all, at that, even if it doesn't make it easy by any definition of the word. 

      2. Floating point item deliberately not specified 

      3. Each such message is OAEP-padded and then encrypted with a (3920 bit) RSA key. Three such messages are strung together to form a RSA packet. Because of the significant overhead involved (both in terms of space and time), Serpent-encrypted comms are preferred whenever feasible.  

      4. See TMSR-RSA OAEP padding for the principle and this discussion for details. 

      5. This is the total size of a packet containing RSA-encrypted material. The useful size (ie payload) of such a packet is merely 702 bytes. 

      6. We really really want to keep this down. 13 bytes is the lowest I can conceive of, but I would so not mind halving it. 

      7. Representing the identifying hash of the object in question.

        We're using the narrower size to save on network traffic -- all the expenditure of another 32 bits here would buy us is de-ambiguation for cases where the count of objects around makes 1 in 2 billion collisions relevant. It doesn't seem likely a client could support such abundance of objects.

        Note that the hashes used here are client-specific, the server doesn't leak its own internal representation of objects to the clients. 

      8. Coordinates X, Y and Z in that order. Because the map goes from -500 to +500, the relationship between the given figure (GF) and map coordinates (MC) is GF / 65.535 - 500 = MC. 

      9. As a full rotation is 2 pi, the relationship between the given figure (GF) and object rotation (OR) is GF / 128 * pi = OR. 

      10. This arrangement permits the representation of arbitrarily large textfields (2nd segment can represent up to 115`792`089`237`316`195`423`570`985`008`687`907`853`269`984`665`640`564`039`457`584`007`913`129`639`936 bytes, which is more than enough space for all the text ever produced -- or likely to ever be produced -- by humanity) at the modest cost of a fixed 3 byte header.

        Unfortunately, it has no longer any utility for Eulora, since we've moved to fixed packets. I'm preserving it here because I really like it in the abstract and it has no other place to go. 

      11. These packets consist of 92 successive 128 bit chunks, Serpent-enciphered individually. To extract the payload one splits the message into 92 16-byte chunks, deciphers them then collates the output into a final result. To produce the packet one cuts a 11`776 bit payload into 92 128-bit chunks, Serpent-enciphers them, and collates the results into the outbound packet. 

      12. Polynomial generator 0x04c11db7. Keys with null IDs are discarded and regenerated. 

      13. Each client and the server will keep a count of messages they sent each other. This value must be incremented on each subsequent message sent by no less than 1 and no more than 255. 

      14. Keys are maintained by both client and server in an ordered ring buffer 256 elements long. The server will not send more keys than the total count of 0(absent)-keys in the respective buffer, irrespective of request count. If the message contains an unknown ID or otherwise is unprocessable, the issuance of a 5.2 packet is adequate response. 

      15. This is the keccak hash of the actual file contents. By convention this hash rendered as a 32 alphanumeric character string is also used as the filename for the file in question. 

      16. This system allows up to 65`536 manifest packets, adding up to potentially 11`993`088 (65`536 * 183) fragments representing a file of up to about 140 Gb (141`038`726`624 = 11`993`088 * 11`760 + 11`744 bits exactly). This will have to be sufficient. 

      17. From 1 to 146 inclusive. 

      18. This also means the protocol does not allow the transfer of files of certain sizes (within 8 bits of a multiple of 11760), which is fine with me. 

      19. The final fragment of the file will have to be padded to length as per this spec.  

      20. This is never issued by the server. 

      21. This is never issued by the client. 

      22. As discussed in comments, the world is a hierarchical structure of objects within objects. 

      23. This portion will get more clarification later on!  

      24. The complete list of these is currently exposed by the extant client, but in any case we'll publish a complete schematic. The server will set the "target" of the player on the last object in the list. 

      25. Limited to at most 1445 characters, matching the maximum length of a chat line. 

      26. Valid rating values are integers between -10 and 10. For a given rating value here GR, the corresponding actual rating value RV is therefore calculated as: Min(GR, 20) - 10. 

      27. These packets consist of three 490 byte successive chunks RSA-encrypted individually. To extract the payload one splits the message into three 490 byte chunks, RSA-decrypts and de-OAEP-pads each one, the collates the results into a final result. To produce the packet one cuts a 5`616 bit payload into three 1`872 bit chunks, OAEP-pads and encrypts them, and collates the results into the outbound packet. 

      28. This is the manner in which new clients register their RSA key with the server (thereby opening a new game account). Later replacement of a registered key IS NOT POSSIBLE. Keep your client's RSA key safe.

        This is also the manner through which IP changes for an account are registered with the server. See the Protocol Mechanics heading for details. 

      29. This is used by the server when signalling to the client to talk to a different server (which is a thing for scaling, because different sectors will be handled by different servers). 

      30. If the client doesn't know its own IP, it's acceptable for this to be zero.  

      31. This permits either client or server to declare Serpent keys via RSA. It is not mandatory (as there exists a Serpent-encapsulated mechanism for the same end) but entirely legal. The server will always respond with at least one 5.2 packet after an accepted 5.1 packet creates a new player account, consisting of 40 Serpent keys to be used to talk to the server. Should the client respond with any other packet than 5.2 or 4.1, the server will send a 2nd 5.2 packet, containing 40 Serpent keys for the client's use.  

      32. Keys obtained through a 5.2 packet are always indexed in the client's buffer in the order they were found in that packet, starting with the first position. 

      33. A RSA packet has 702 total bytes available, of which 5 are used otherwise and the remainder of 697 are available for packing serpent keys, which take 36 bytes each (crc32 id inclusive). 

      34. Length being actually how they're sorted on the server side. 

      35. This is meant as a production order, essentially. I expect it will get further refined as its implementation gets nearer. 

      36. This is a generic initiation of player activity and as such literally an attempt, with the contents defining the exact activity. It replaces both Explore and Repair from earlier versions of this protocol but it further stands in for *any* game-defined activity even ones that may appear only at a later time. It's as generic as it gets and quite on purpose, since this is the network layer, not the game layer. For more details on the game-defined activities, see the description of the data hierarchy. 

      37. Known activities are obtained like everything else in game as part of the data hierachy representing the client's view of the world at any given time. The relevant grammar for it will be made public as it serves as the in-game protocol specification just as this serves as the network-level client-server protocol specification. 

      38. Known in Eulora 1 as 'recipe' and 'equipped in mind'. At this level, the protocol for Eulora2 is really more generic than that, without any loss otherwise whatsoever. 

      39. This is set by server through a type 6 message for both players involved, the trade is an object like any other that the OP has to request. The server will also expire trades, enforce them etc. 

      40. Note that this is an action and as such just a part of a 4.5 message, Client Action, where the overall message structure is specified. 

      Comments feed: RSS 2.0

8 Responses to “The Communication Protocol for Eulora2, Restated”

  1. Diana Coman says:

    For the record, the main change that triggered this is the switch to a more generic approach for communicating in-game actions, hence the 7.2 message type currently (Attempt), replacing both earlier 7.2 (Explore) and earlier 7.5 (Repair). The previous changes and additions supporting chat and the WoT are brought in as such.

    The reason for the more generic 'Attempt' is that at this level, all 'activities' are really exactly attempts at doing "something" where that something is defined through the intent (aka the chosen 'activity' as such), target (aka the passive part in the resulting inter-action), tool and method. So in this sense and at this level of communicating to the server, there is no need to have different messages and it serves best to have as few and as generic as possible.

    Worth noting that the 'Exchange' and 'Attack' (possibly more accurately described as 'Battle') are different from the above in that they happen in their own micro-environment i.e. the corresponding exchange or battle object. At this stage, 'Train' seems to be of this type as well, hence why I left it in there as it was but I'm not 100% set on this currently and as design advances, I'll see what works best and update it if needed.

  2. Jacob Welsh says:

    Is the "object" special type in 3.1, which seems specialized to physical objects, still relevant at all in light of the client data hierarchy? I'm seeing it referenced only under 7.8 Relocate.

    Following on my last question, are 4.6-4.8 the relevant messages for requesting and sending those client data nodes? Here they seem to contain only an ID and a flat text field for "properties"; has this changed?

    Incidentally, there's something odd with the list item formatting under 6.1 and the footnote numbers: I'm seeing each item number followed by a dot with double right chevron quote character drawn on top of it.

  3. Jacob Welsh says:

    We're using the narrower size to save on network traffic -- all the expenditure of another 32 bits here would buy us is de-ambiguation for cases where the count of objects around makes 1 in 2 billion collisions relevant. It doesn't seem likely a client could support such abundance of objects.

    I don't quite follow the argument here (late as it may be to say so). By the "birthday problem", all that's needed to have significant odds of at least one collision existing by pure chance are on the order of sqrt(N) items, that is, 64k, even if the odds that a particular selected item collides with some other item in that population are only one in 64k. What the implications of such a collision might be to the client I'm unsure, perhaps it's a case like the protocol's encrypted message integrity, where enough internal sanity checks of the tree structure would make it quite likely the situation would be detected, perhaps even resolved.

  4. Diana Coman says:

    Is the "object" special type in 3.1, which seems specialized to physical objects, still relevant at all in light of the client data hierarchy? I'm seeing it referenced only under 7.8 Relocate.

    There is some confusing overlap on this "object" term because the protocol spec uses it quite loosely to mean indeed both physical objects (hence the type in 3.1 with its actual coordinates/location) and generically "data objects" aka the nodes in the data hierarchy (hence the note about the abundance of objects but even that is not quite capturing it all, sigh). While I try to use consistently "nodes" exactly to avoid this sort of confusion, the trouble is likely deeper in that the data hierarchy is simply a view of the world and thus the two are not that easily kept entirely distinct in terms. Add to this that there are in fact physical objects (like items or PCs, which are what the footnote about size of clientside IDs refers to, explicitly) and abstract objects (like that Self for instance, which this spec is not even aware of) and the overlap + trouble are perhaps at least more obvious.

    As for relevancy, as you notice, it's exactly used at 7.8 hence indeed for moving of.... well, physical objects will do in a pinch but it's more PCs and NPCs really, not items (those are moved via 7.6, not relocated because there is a distinction between the two actions). In any case, the special types in the communication protocol here were an early attempt to define in fact what are now the values for leaf types. It took a while to figure out the working structure to support it all and in the beginning the communication protocol aimed to spec it all, in part because at least some parts were already needed and in part because it wasn't yet clear what separation is possible and where exactly it cuts through. I suppose it makes some sense to remove that "object" special type and simply add its components directly under 7.8 where it is used but the footnotes on it make more sense where they are really. So for now and at least until next revision of the article text, I'll leave it as it stands and this discussion in the comments will have to do for clarification on it.

    are 4.6-4.8 the relevant messages for requesting and sending those client data nodes? Here they seem to contain only an ID and a flat text field for "properties"; has this changed?

    Indeed, the 4.6-4.8 are the relevant messages for requesting and sending any and all nodes of the data hierarchy. There is no change really and it works exactly as this spec says. The properties field is indeed text and it quite needs to be so it works for all possible types essentially. This spec here is as detailed as it can get *at this level*, meaning specifically that it is not concerned with nor trying to fix upfront what the properties may be - they are simply text here and that's enough. This is exactly the separation line I was mentioning earlier - from a communications point of view, all one needs to know is how long is that field, to read it accordingly and pass it on to the higher level where it will be made sense of based on the relevant syntax. So this stays exactly as it is and then the data hierarchy spec provides clear information as to what exactly will be in that field for each type of node and in what order.

    Incidentally, there's something odd with the list item formatting under 6.1 and the footnote numbers

    Myeah, there's some css clash on ordered lists beyond some level of indentation, I'm not that keen on diving into that currently. The 6.1 part is more easily fixed simply ditching the numbers as they are not all that mandatory in there. For the footnotes, it'll have to wait until I look at the css again, sorry.

    I don't quite follow the argument here (late as it may be to say so). By the "birthday problem", all that's needed to have significant odds of at least one collision existing by pure chance are on the order of sqrt(N) items, that is, 64k, even if the odds that a particular selected item collides with some other item in that population are only one in 64k.

    The focus of the spec though is not really on the chances of a collision as such but on the potential restrictions imposed by a smaller size. Hence, it's not really about whether a collision might occur - arguably over the whole server's timeline, it *will* definitely occur at some point if the chance is anything other than plain 0. What matters most is whether dealing with such collisions is going to restrict the game significantly/more than it's worth otherwise.

    Worth noting that, from the client point of view, there is no collision possible, though. This is because the server picks those IDs and as such, it's the server's responsibility to deal with any collisions before even assigning the IDs (hence: before the client hears of some ID). So yes, it can be a pita serverside, but it's no trouble clientside anyway because the IDs making it *to* the client are by definition unique/collision-free.

  5. Jacob Welsh says:

    What matters most is whether dealing with such collisions is going to restrict the game significantly/more than it's worth otherwise.

    Makes sense, especially now knowing the server is responsible for preventing collisions. Though, if the server effectively has to keep track of all assigned IDs per client anyway, it occurs to me that fully random IDs might serve just as well as whatever kind of hash.

  6. Diana Coman says:

    The server has to keep track of assigned IDs per client indeed but not at all times. Considering that the game's world itself is infinite, the same principle applies - it has to exist reliably at all times and forever *as a consistent possibility* (hence why not fully random) but it needs to be actually detailed to the level of IDs only while someone is around to see it.

    From a more philosophical perspective, it all gets back if you will to giving a concrete implementation in answer to that old question of whether a falling tree makes a sound if there's nobody around to hear it: the answer as it organically comes up from e2 implementation is that the falling tree certainly does what it does regardless of anyone there or not but any such "doing" will get an actual label retained only for as long as someone is there to hear or at least inquire about it.

  7. [...] ! Television and radio broadcasts were cut off. [^]The platforms really go all the way down : from the cheap money, progressive taxation that unnaturally redistributes away from the worthy [...]

  8. [...] a filename that it doesn't currently have, it will simply request it from the server according to the lower level communication protocol that fully supports such requests and then proceed to use the file as soon as it's fully downloaded [...]

Leave a Reply