<abstract>This specification defines an XMPP protocol extension that enables any two entities to establish a one-to-one bytestream between themselves, where the data is broken down into smaller chunks and transported in-band over XMPP.</abstract>
<remark>Encouraged use of IQ stanzas rather than message stanzas for sending data; clarified bidirectional nature of IBB; more clearly specified data handling, error conditions, XML syntax, and attribute datatypes.</remark>
<p>This document describes In-Band Bytestreams (IBB), a simple XMPP protocol extension that enables two entities to establish a virtual bytestream over which they can exchange Base64-encoded chunks of data over XMPP itself. Because IBB provides a generic bytestream, so its usage is open-ended. To date it has been used as a fallback method for sending files when out-of-band methods such as &xep0065; are not available. However, IBB could also be useful for any kind of relatively low-bandwidth activity, such as games, shell sessions, or encrypted communication.</p>
<p>IBB as specified in this document defines two protocol aspects:</p>
<ol>
<li>How to set up and tear down an IBB session using <open/> and <close/> elements sent within IQ stanzas.</li>
<li>How to send chunks of IBB data using IQ or message stanzas containing <data/> elements.</li>
</ol>
<p>Other methods can be used for setup and teardown, such as &xep0166; as specified in &xep0261;.</p>
<section2topic='Creating a Bytestream'anchor='create'>
<p>In order to set up an in-band bytestream, the initiator sends an IQ stanza of type "set" containing an <open/> element qualified by the 'http://jabber.org/protocol/ibb' namespace. This element possesses two attributes: 'block-size' defines the maximum size in bytes of each data chunk (which SHOULD NOT be greater than 65535) and 'sid' defines a unique session ID for this IBB session (which SHOULD math the NMTOKEN datatype).</p>
<p>If the responder informs the initiator that it wishes to proceed with the session, the initiator can begin to send data over the bytestream (in addition, because the bytestream is bidirectional, the responder can also send data; see the <linkurl='#bidi'>Bidirectionality</link> section of this document for details).</p>
<p>Each chunk of data is contained in a <data/> element qualified by the 'http://jabber.org/protocol/ibb' namespace. The data element SHOULD be sent in an IQ stanza to enable proper tracking and throttling, but MAY be sent in a message stanza. The data to be sent, prior to any wrapping in the <data/> element and IQ or message stanza, MUST NOT be larger than the 'block-size' determined in the bytestream negotiation.</p>
<p>Each chunk of data is included as the XML character data of the <data/> element after being encoded as Base64 as specified in Section 4 of &rfc4648;.</p>
<p>The <data/> element MUST possess a 'seq' attribute; this is a 16-bit unsigned integer that acts as a counter for data chunks sent within this session. The 'seq' value starts at 0 (zero) and MUST be incremented for each packet sent. Thus, the second chunk sent has a 'seq' value of 1, the third chunk has a 'seq' value of 2, and so on. The counter loops at maximum, so that after value 65535 the 'seq' MUST start again at 0.</p>
<p>The <data/> element MUST also possess a 'sid' attribute that ties the data chunk to this particular IBB session.</p>
<p>In the case of IQ stanzas, if the packet can be processed then the recipient MUST reply with an IQ stanza of type "result".</p>
<examplecaption='Acknowledging data using IQ'><![CDATA[
<iqfrom='juliet@capulet.com/balcony'
id='kr91n475'
to='romeo@montague.net/orchard'
type='result'/>
]]></example>
<p>The sender need not wait for these acknowledgements before sending further stanzas. However, it is RECOMMENDED that the sender does wait in order to minimize possible rate-limiting penalties.</p>
<p>It is possible that delivery of the stanza might fail, e.g. because the recipient has gone offline or because a server-to-server link has gone down). In this case the entity that detects the error shall return an appropriate XMPP stanza eror, such as &recipient; or &timeout;. Upon receiving notice that delivery of a data packet has failed, the sender MUST consider the bytestream to be closed and invalid.</p>
<p>It is also possible that the recipient might detect an error with the data packet, e.g. because the session ID is unknown, because the sequence number has already been used, or because the data is not formatted in accordance with Section 4 of <cite>RFC 4648</cite>. In this case the recipient shall return an appropriate XMPP stanza error, such as ¬found;, &unexpected; or &badrequest;. Upon receiving notice that a data packet is cannot be processed by the recipient, the sender SHOULD consider the bytestream to be closed and invalid but MAY attempt to correct the error and re-send the offending data packet using the same sequence number (the recipient MUST NOT consider a sequence number to have been used until the data packet has been successfully processed).</p>
<p>Data packets MUST be processed in the order they are received. If an out-of-sequence packet is received for a particular bytestream (determined by checking the 'seq' attribute), then this indicates that a packet has been lost. The recipient MUST NOT process the data of such an out-of-sequence packet, nor any that follow it within the same bytestream; instead, the recipient MUST consider the bytestream invalid and SHOULD close the bytestream as described in the next section.</p>
<p>It is possible that the recipient of the close notification does not know about the bytestream, in which case it would return an ¬found; error.</p>
<examplecaption='Recipient does not know about the IBB session'><![CDATA[
<p>An in-band bytestream is bidirectional. Therefore, either party to the bytestream is allowed to send data. Each sender MUST initialize the 'seq' attribute to zero and increment the 'seq' value by one with each chunk of data it sends. Each recipient MUST track chunks based on the 'seq' values it receives. The 'seq' values in each direction are independent of the values in the other direction. Thus there are two data sequences for each SessionID. If enabled by the application that uses IBB, the parties MAY negotiate multiple SessionIDs for the same bytestream, however such methods are out of scope for this specification.</p>
<li>For proper tracking of delivery and processing errors related to data packets, the 'id' attribute MUST be used with IQ stanzas and SHOULD be used with message stanzas.</li>
<li>It is RECOMMENDED to use IQ stanzas when sending data packets.</li>
<li>If message stanzas are used when sending data packets, the sender SHOULD also use &xep0079; or some other stanza flow-control method.</li>
<p>The In-Band Bytestreams protocol is as secure as the underlying XMPP transport. The application that uses IBB could have its own security layer, but this is outside of the scope of IBB.</p>
<p>An entity MUST verify any Base64 data received. An implementation MUST reject (not ignore) any characters that are not explicitly allowed by the Base64 alphabet; this helps to guard against creation of a covert channel that could be used to "leak" information. An implementation MUST NOT break on invalid input and MUST reject any sequence of Base64 characters containing the pad ('=') character if that character is included as something other than the last character of the data (e.g., "=AAA" or "BBBB=CCC"); this helps to guard against buffer overflow attacks and other attacks on the implementation. Base encoding visually hides otherwise easily recognized information, such as passwords, but does not provide any computational confidentiality. Base64 encoding MUST follow the definition in Section 4 of RFC 4648.</p>