Internet Engineering Task Force IPTEL WG Internet Draft Lennox/Schulzrinne draft-ietf-iptel-cpl-02.txt Columbia University July 14, 2000 Expires: January, 2001 CPL: A Language for User Control of Internet Telephony Services STATUS OF THIS MEMO This document is an Internet-Draft and is in full conformance with all provisions of Section 10 of RFC2026. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF), its areas, and its working groups. Note that other groups may also distribute working documents as Internet- Drafts. Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress". The list of current Internet-Drafts can be accessed at http://www.ietf.org/ietf/1id-abstracts.txt The list of Internet-Draft Shadow Directories can be accessed at http://www.ietf.org/shadow.html. Abstract The Call Processing Language (CPL) is a language that can be used to describe and control Internet telephony services. It is designed to be implementable on either network servers or user agent servers. It is meant to be simple, extensible, easily edited by graphical clients, and independent of operating system or signalling protocol. It is suitable for running on a server where users may not be allowed to execute arbitrary programs, as it has no variables, loops, or ability to run external programs. This document is a product of the IP Telephony (IPTEL) working group of the Internet Engineering Task Force. Comments are solicited and should be addressed to the working group's mailing list at iptel@lists.research.bell-labs.com and/or the authors. 1 Introduction The Call Processing Language (CPL) is a language that can be used to Lennox/Schulzrinne [Page 1] Internet Draft CPL July 14, 2000 describe and control Internet telephony services. It is not tied to any particular signalling architecture or protocol; it is anticipated that it will be used with both SIP [1] and H.323 [2]. The CPL is powerful enough to describe a large number of services and features, but it is limited in power so that it can run safely in Internet telephony servers. The intention is to make it impossible for users to do anything more complex (and dangerous) than describing Internet telephony services. The language is not Turing-complete, and provides no way to write loops or recursion. The CPL is also designed to be easily created and edited by graphical tools. It is based on XML [3], so parsing it is easy and many parsers for it are publicly available. The structure of the language maps closely to its behavior, so an editor can understand any valid script, even ones written by hand. The language is also designed so that a server can easily confirm scripts' validity at the time they are delivered to it, rather that discovering them while a call is being processed. Implementations of the CPL are expected to take place both in Internet telephony servers and in advanced clients; both can usefully process and direct users' calls. This document primarily addresses the usage in servers. A mechanism will be needed to transport scripts between clients and servers; this document does not describe such a mechanism, but related documents will. The framework and requirements for the CPL architecture are described in RFC 2824, "Call Processing Language Framework and Requirements." [4]. 1.1 Conventions of this document In this document, the key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" are to be interpreted as described in RFC 2119 [5] and indicate requirement levels for compliant CPL implementations. In examples, non-XML strings such as -action1, -action2, and so forth, are sometimes used. These represent further parts of the script which are not relevant to the example in question. Some paragraphs are indented, like this; they give motivations of design choices, or questions for future discussion in the development of the CPL, and are not essential to the specification of the language. Lennox/Schulzrinne [Page 2] Internet Draft CPL July 14, 2000 2 Structure of CPL scripts 2.1 High-level structure A CPL script consists of two types of information: ancillary information about the script, and call processing actions. A call processing action is a structured tree that describes the decisions and actions a telephony signalling server performs on a call set-up event. There are two types of call processing actions: top-level actions are actions that are triggered by signalling events that arrive at the server. Two top-level action names are defined: incoming, the action performed when a call arrives whose destination is the owner of the script; and outgoing, the action performed when a call arrives whose originator is the owner of the script. Sub-actions are actions which can be called from other actions. The CPL forbids sub-actions from being called recursively: see section 9. Note: The names "action," "sub-action," and "top-level action" are probably not ideal. Suggestions for better names for these concepts are welcomed. Ancillary information is information which is necessary for a server to correctly process a script, but which does not directly describe any actions. Currently, no ancillary information is defined, but the section is reserved for future extensions. 2.2 Abstract structure of a call processing action Abstractly, a call processing action is described by a collection of nodes, which describe actions that can be performed or choices which can be made. A node may have several parameters, which specify the precise behavior of the node; they usually also have outputs, which depend on the result of the condition or action. For a graphical representation of a CPL action, see figure 1. Nodes and outputs can be thought of informally as boxes and arrows; the CPL is designed so that actions can be conveniently edited graphically using this representation. Nodes are arranged in a tree, starting at a single root node; outputs of nodes are connected to additional nodes. When an action is run, the action or condition described by the top-level node is performed; based on the result of that node, the server follows one of the node's outputs, and that action or condition is performed; this process continues until a node with no specified outputs is reached. Because the graph is acyclic, this will occur after a bounded and predictable number of nodes are visited. Lennox/Schulzrinne [Page 3] Internet Draft CPL July 14, 2000 If an output to a node is not specified, it indicates that the CPL server should perform a node- or protocol-specific action. Some nodes have specific default actions associated with them; for others, the default action is implicit in the underlying signalling protocol, or can be configured by the administrator of the server. For further details on this, see section 11. _________________ ___________________ ________ busy | Address-switch | | location | | proxy |--------\ Call --->| field: origin | ->| url: sip:jones@ |--->|timeout:| timeout| | subfield: host | / | example.com | | 10s |--------| |-----------------|/ |___________________| | | failure| | subdomain-of: | |________|--------| | example.com | | |-----------------| _____________________________________________/ | otherwise | /.......................................... | |\|. Voicemail . |_________________| \. ____________________ . ->| location | __________ . . | url: sip:jones@ | | redirect | . . | voicemail. |--->| | . . | example.com | |__________| . . |____________________| . .......................................... Figure 1: Sample CPL Action: Graphical Version 2.3 Location model For flexibility, one piece of information necessary for the function of a CPL is not given as node parameters: the set of locations to which a call is to be directed. Instead, this set of locations is stored as an implicit global variable throughout the execution of a processing action (and its sub-actions). This allows locations to be retrieved from external sources, filtered, and so forth, without requiring general language support for such actions (which could harm the simplicity and tractability of understanding the language). The specific actions which add, retrieve, or filter location sets are given in section 6. Lennox/Schulzrinne [Page 4] Internet Draft CPL July 14, 2000 For the incoming top-level processing action, the location set is initialized to the empty set. For the outgoing action, it is initialized to the destination address of the call. 2.4 XML structure Syntactically, CPL scripts are represented by XML documents. XML is thoroughly specified by [3], and implementors of this specification should be familiar with that document, but as a brief overview, XML consists of a hierarchical structure of tags; each tag can have a number of attributes. It is visually and structurally very similar to HTML [6], as both languages are simplifications of the earlier and larger standard SGML [7]. See figure 2 for the XML document corresponding to the graphical representation of the CPL script in figure 1. Both nodes and outputs in the CPL are represented by XML tags; parameters are represented by XML tag attributes. Typically, node tags contain output tags, and vice-versa (with one exception; see section 2.3). The connection between the output of a node and another node is represented by enclosing the tag representing the pointed-to node inside the tag for the outer node's output. Convergence (several outputs pointing to a single node) is represented by sub-actions, discussed further in section 9. The higher-level structure of a CPL script is represented by tags corresponding to each piece of meta-information, sub-actions, and top-level actions, in order. This higher-level information is all enclosed in a special tag cpl, the outermost tag of the XML document. A complete Document Type Declaration for the CPL is provided in Appendix A. The remainder of the main sections of this document describe the semantics of the CPL, while giving its syntax informally. For the formal syntax, please see the appendix. 3 Document information This section gives meta-information about CPL scripts. 3.1 CPL Document Identifiers for XML A CPL script list which appears as a top-level XML document is identified with the formal public identifier "-//IETF//DTD RFCxxxx CPL 1.0//EN". If this document is published as an RFC, "xxxx" will be replaced by the RFC number. Lennox/Schulzrinne [Page 5] Internet Draft CPL July 14, 2000
Figure 2: Sample CPL Script: XML Version An CPL embedded as a fragment within another XML document is identified with the XML namespace identifier "http://www.ietf.org/internet-drafts/draft-ietf-iptel-cpl-02.txt". If this document is published as an RFC, the namespace identifier will be "http://www.rfc-editor.org/rfc/rfcxxxx.txt", where xxxx is the RFC number. Note that the URIs specifying XML namespaces are only globally unique names; they do not have to reference any particular actual object. The URI of a canonical source of this specification meets the requirement of being globally unique, and is also useful to document the format. 3.2 MIME Registration Lennox/Schulzrinne [Page 6] Internet Draft CPL July 14, 2000 As an XML type, CPL's MIME registration conforms with "XML Media Types" [8] as well as RFC 2048 [9]. MIME media type name: application MIME subtype name: cpl+xml Mandatory parameters: none Optional parameters: charset As for application/xml in "XML Media Types." Encoding considerations: As for application/xml in "XML Media Types." Security considerations: See section 13, and section 10 of "XML Media Types." Interoperability considerations: Different CPL servers may use incompatible address types. However, all potential interoperability issues should be resolvable at the time a script is uploaded; there should be no interoperability issues which cannot be detetected until runtime. Published specification: This document. Applications which use this media type: None publically released at this time, as far as the authors are aware. Additional information: Magic number: None File extension: .cpl or .xml Macintosh file type code: "TEXT" Person and e-mail address for further information: Jonathan Lennox Henning Schulzrinne Intended usage: COMMON Author/Change Controller: The IETF. 4 Script structure: overview As mentioned, a CPL script consists of ancillary information, Lennox/Schulzrinne [Page 7] Internet Draft CPL July 14, 2000 subactions, and top-level actions. The full syntax of the cpl node is given in figure 3. Tag: cpl Parameters: none Sub-tags: ancillary See section 10 subaction See section 9 outgoing Top-level actions to take on this user's outgoing calls incoming Top-level actions to take on this user's incoming calls Output: outgoing Parameters: none Output: incoming Parameters: none Figure 3: Syntax of the top-level cpl tag Call processing actions, both top-level actions and sub-actions, consist of nodes and outputs. Nodes and outputs are both described by XML tags. There are four categories of CPL nodes: switches , which represent choices a CPL script can make; location modifiers , which add or remove locations from the location set; signalling actions , which cause signalling events in the underlying protocol; and non- signalling actions, which take an action but do not effect the underlying protocol. 5 Switches Switches represent choices a CPL script can make, based on either attributes of the original call request or items independent of the call. All switches are arranged as a list of conditions that can match a variable. Each condition corresponds to a node output; the output points to the next node to execute if the condition was true. The conditions are tried in the order they are presented in the script; the output corresponding to the first node to match is taken. There are two special switch outputs that apply to every switch type. The output not-present, which MAY occur anywhere in the list of outputs, is true if the variable the switch was to match was not Lennox/Schulzrinne [Page 8] Internet Draft CPL July 14, 2000 present in the original call setup request. The output otherwise, which MUST be the last output specified if it is present, matches if no other condition matched. If no condition matches and no otherwise output was present in the script, the default script action is taken. See section 11 for more information on this. 5.1 Address switches Address switches allow a CPL script to make decisions based on one of the addresses present in the original call request. They are summarized in figure 4. Node: address-switch Outputs: address Specific addresses to match Parameters: field origin, destination, or original-destination subfield address-type, user, host, port, tel, display, password, or alias-type Output: address Parameters: is exact match contains substring match (for display only) subdomain-of sub-domain match (for host, tel only) Figure 4: Syntax of the address-switch node Address switches have two node parameters: field, and subfield. The mandatory field parameter allows the script to specify which address is to be considered for the switch: either the call's origin address (field "origin"), its current destination address (field "destination"), or its original destination (field "original- destination"), the destination the call had before any earlier forwarding was invoked. Servers MAY define additional field values. The optional subfield specifies what part of the address is to be considered. The possible subfield values are: address-type, user, host, port, tel, and display. Additional subfield values MAY be defined: two additional ones, password and asn1 are defined specifically for SIP and H.323 respectively, in sections 5.1.1 and 5.1.2 below. If no subfield is specified, the "entire" address is matched; the precise meaning of this is defined for each underlying signalling protocol. Servers MAY define additional subfield values. Lennox/Schulzrinne [Page 9] Internet Draft CPL July 14, 2000 The subfields are defined as follows: address-type This indicates the type of the underlying address; i.e., the URI scheme, if the address can be represented by a URI. The types specifically discussed by this document are sip, tel, and h323. The address type is not case- sensitive. It has a value for all defined address types. user This subfield of the address indicates, for e-mail style addresses, the user part of the address. For telephone number style address, it includes the subscriber number. This subfield is case-sensitive; it may be not present. host This subfield of the address indicates the Internet host name or IP address corresponding to the address, in host name, IPv4, or IPv6 format. For host names only, subdomain matching is supported with the subdomain-of match operator. It is not case sensitive, and may be not present. port This subfield indicates the TCP or UDP port number of the address, numerically in decimal format. It is not case sensitive, as it MUST only contain decimal digits. It may be not present; however, for address types with default ports, an absent port matches the default port number. tel This subfield indicates a telephone subscriber number, if the address contains such a number. It is not case sensitive (the telephone numbers may contain the symbols `A' `B' `C' and `D'), and might not be present. It may be matched using the subdomain-of match operator. Punctuation and separator characters in telephone numbers are discarded. display This subfield indicates a "display name" or user-visible name corresponding to an address. It is a Unicode string, and is matched using the case-insensitive algorithm described in section 5.2. The contains operator may be applied to it. It may be not present. For any completely unknown subfield, the server MAY reject the script at the time it is submitted with an indication of the problem; if a script with an unknown subfield is executed, the server MUST consider the not-present output to be the valid one. The address output tag may take exactly one of three possible parameters, indicating the kind of matching allowed. is An output with this match operator is followed if the Lennox/Schulzrinne [Page 10] Internet Draft CPL July 14, 2000 subfield being matched in the address-switch exactly matches the argument of the operator. It may be used for any subfield, or for the entire address if no subfield was specified. subdomain-of This match operator applies only for the subfields host and tel. In the former case, it matches if the hostname being matched is a subdomain of the domain given in the argument of the match operator; thus, subdomain- of="example.com" would match the hostnames "example.com", "research.example.com", and "zaphod.sales.internal.example.com". IP addresses may be given as arguments to this operator; however, they only match exactly. In the case of the tel subfield, the output matches if the telephone number being matched has a prefix that matches the argument of the match operator; subdomain-of="1212555" would match the telephone number "1 212 555 1212." contains This match operator applies only for the subfield display. The output matches if the display name being matched contains the argument of the match as a substring. 5.1.1 Address switch mapping for SIP For SIP, the origin address corresponds to the address in the From header; destination corresponds to the Request-URI; and original- destination corresponds to the To header. The display subfield of an address is the display-name part of the address, if it is present. Because of SIP's syntax, the destination address field will never have a display subfield. The address-type subfield of an address is the URI scheme of that address. Other address fields depend on that address-type. For sip URLs, the user, host, and port subfields correspond to the "user," "host," and "port" elements of the URI syntax. The tel subfield is defined to be the "user" part of the URI if and only if the "user=phone" parameter is given to the URI. An additional subfield, password is defined to correspond to the "password" element of the SIP URI; however, use of this field is NOT RECOMMENDED for general security reasons. For tel URLs, the tel and user subfields are the subscriber name; in the former case, visual separators are stripped. The host and port subfields are both not present. Lennox/Schulzrinne [Page 11] Internet Draft CPL July 14, 2000 For h323 URLs, the subfields are set as in section 5.1.2 below. For other URI schemes, only the address-type subfield is defined by this specification; servers MAY set other pre-defined subfields, or MAY support additional subfields. If no subfield is specified for addresses in SIP messages, the string matched is the URI part of the address. For "sip" URLs, all parameters are stripped; for other URLs, the URL is used verbatim. 5.1.2 Address switch mapping for H.323 For H.323, the origin address corresponds to the primary alias address in the sourceAddress field of the Setup-UUIE user-user information element, and to the Q.931 information element callingPartyNumber. If both fields are present, which one has priority is a matter of local server policy; the server SHOULD use the same resolution as it would use for routing decisions in this case. Similarly, the destination address corresponds to the primary alias address of the destinationAddress field, and to the Q.931 information element calledPartyNumber. This discussion is based on H.323 version 4 [10], which is expected to be approved in November 2000. The original-destination address corresponds to the redirectedNumber Q.931 information element, if it is present; otherwise it is the same as the destination address. The mapping of H.323 addresses into subfields depends on the type of the alias address. An additional subfield type, alias-type, is defined for H.323 servers, corresponding to the type of the address. Possible values are dialedDigits, h323-ID, url-ID, transportID, email-ID, partyNumber, mobileUIM, and Q.931IE. If future versions of the H.323 specification define additional types of alias addresses, those names MAY also be used. In versions of H.323 prior to version 4, dialedDigits was known as e164. The new name should be used. The value of the address-type subfield for H.323 messages is "h323" unless the alias type is url-ID and the URL scheme is something other than h323; in this case the address-type is the URL scheme, as specified above for SIP. If an alias address of type h323-ID is present anywhere among the sequence of aliases, the first such h323-ID alias address is used for the display subfield of the address. The values of all other Lennox/Schulzrinne [Page 12] Internet Draft CPL July 14, 2000 subfields depend only on the first alias address in the sequence. The following mappings are used for H.323 alias types: dialedDigits, partyNumber, mobileUIM, and Q.931IE: the tel and user subfields are the string of digits, as is the "entire-address" form. The host and port subfields are not present. url-ID with a "h323" URI: the user, host, and port subfields are set to the corresponding parts of the H.323 URL. The tel subfield is not present. The "entire-address" form corresponds to the entire URI. url-ID with other URI schemes: the same mapping is used as for SIP, above. email-ID: the user and host subfields are set to the corresponding parts of the e-mail address. The port and tel subfields are not present. The "entire-address" form corresponds to the entire e-mail address. transportID: if the TransportAddress is of type "ipAddress," "ipSourceRoute," or "ip6Address," the host subfield is set to the "ip" element of the sequence, translated into the standard IPv4 or IPv6 textual representation, and the port subfield is set to the "port" element of the sequence represented in decimal. The tel and user fields are not present. The "entire-address" form is not defined. The representation and mapping of transport addresses is not defined for non-IP addresses. 5.2 String switches String switches allow a CPL script to make decisions based on free- form Unicode strings present in a call request. They are summarized in figure 5. String switches have one node parameter: field. The mandatory field parameter specifies which string is to be matched. Currently five fields are defined. Three fields are currently applicable only to SIP, one is currently applicable only to H.323, and one is applicable to both. The three fields which are applicable only to SIP are: subject, indicating the subject of the call; organization, indicating the Lennox/Schulzrinne [Page 13] Internet Draft CPL July 14, 2000 Node: string-switch Outputs: string Specific string to match Parameters: field subject, organization, user-agent, language, or display Output: string Parameters: is exact match contains substring match Figure 5: Syntax of the string-switch node originator's organization; and user-agent, indicating the program or device with which the call request was made. All these fields correspond to the contents of the SIP header fields with the same names. The field applicable only to H.323 is display, which corresponds to the Q.931 information element of the same name. This is conventionally used for Caller-ID purposes, so arguably it should be mapped to the display subfield of an address-match with the field originator. However, since a) it is a message-level information element, not an address- level one, and b) the Q.931 specification [11] says only that "[t]he purpose of the Display information element is to supply display information that may be displayed by the user," it seems to be more appropriate to match it as a string instead. The field appropriate both to SIP and H.323 is language. This field contains a list of RFC 1766 [12] language tags, separated by commas, corresponding to the SIP Accept-Language header and the H.323 language UUIE. Note that matching based on contains is likely to be much more useful than matching based on is, for this field. Strings are matched as case-insensitive Unicode strings, in the following manner. First, strings are canonicalized to the "Compatibility Composition" (KC) form, as specified in Unicode Technical Report 15 [13]. Then, strings are compared using locale- insensitive caseless mapping, as specified in Unicode Technical Report 21 [14]. Lennox/Schulzrinne [Page 14] Internet Draft CPL July 14, 2000 Code to perform the first step, in Java and Perl, is available; see the links from Annex E of UTR 15 [13]. The case-insensitive string comparison in the Java standard class libraries already performs the second step; other Unicode-aware libraries should be similar. The output tags of string matching are named string, and have a mandatory argument, one of is or contains, indicating whole-string match or substring match, respectively. 5.3 Time switches Time switches allow a CPL script to make decisions based the time and/or date the script is being executed. They are summarized in figure 6. Node: time-switch Outputs: time Specific time to match Parameters: tzid RFC 2445 Time Zone Identifier tzurl RFC 2445 Time Zone URL Output: time Parameters: dtstart Start of interval (RFC 2445 DATE-TIME) dtend End of interval (RFC 2445 DATE-TIME) duration Length of interval (RFC 2445 DURATION) freq Frequency of recurrence (one of "secondly", "minutely", "hourly", "daily", "weekly", "monthly", or "yearly") interval How often the recurrence repeats until Bound of recurrence (RFC 2445 DATE-TIME) count Number of occurences of recurrence bysecond List of seconds within a minute byminute List of minutes within an hour byhour List of hours of the day byday List of days of the week bymonthday List of days of the month byyearday List of days of the year byweekno List of weeks of the year bymonth List of months of the year wkst First day of week bysetpos List of values within set of events specified Figure 6: Syntax of the time-switch node Lennox/Schulzrinne [Page 15] Internet Draft CPL July 14, 2000 Time switches are based closely on the specification of recurring intervals of time from the Internet Calendaring and Scheduling Core Object Specification (iCal COS), RFC 2445 [15]. This allows CPLs to be generated automatically from calendar books. It also allows us to re-use the extensive existing work specifying time intervals. The time-switch tag takes two optional parameters, tzid and tzurl, both of which are defined in RFC 2445 (sections 4.8.3.1 and 4.8.3.5 respectively). The TZID is the identifying label by which a time zone definition is referenced. If it begins with a forward slash (solidus), it references a to-be-defined global time zone registry; otherwise it is locally-defined at the server. The TZURL gives a network location from which an up-to-date VTIMEZONE definition for the timezone can be retrieved. If a script is uploaded with a tzid and tzurl which the CPL server does not recognize or cannot resolve, it SHOULD diagnose and reject this at script upload time. If neither tzid nor tzurl are present, all non-UTC times within this time switch should be interpreted as being "floating" times, i.e. that they are specified in the local timezone of the CPL server. Because of daylight-savings-time changes over the course of a year, it is necessary to specify time switches in a given timezone. UTC offsets are not sufficient, or a time-of-day routing rule which held between 9 am and 5 pm in the eastern United States would start holding between 8 am and 4 pm at the end of October. Authors of CPL servers should be careful to handle correctly the intervals when local time is discontinuous, at the beginning or end of daylight-savings time. Time nodes specify a list of periods during which their output should be taken. They have two required parameters: dtstart, which specifies the beginning of the first period of the list, and exactly one of dtend or duration, which specify the ending time or the duration of the period, respectively. The dtstart and dtend parameters are formatted as iCal COS DATE-TIME values, as specified in section 4.3.5 of RFC 2445. The duration parameter is given as an iCal COS DURATION parameter, as specified in section 4.3.6 of RFC 2445. If no other parameters are specified, a time node indicates only a single period of time. More complicated sets periods intervals are Lennox/Schulzrinne [Page 16] Internet Draft CPL July 14, 2000 constructed as recurrences. A recurrence is specified by including the freq parameter, which indicates the type of recurrence rule. No parameters other than dtstart, dtend, and duration SHOULD be specified unless freq is present. The freq parameter takes one of the following values: secondly, to specify repeating periods based on an interval of a second or more; minutely, to specify repeating periods based on an interval of a minute or more; hourly, to specify repeating periods based on an interval of an hour or more; daily, to specify repeating periods based on an interval of a day or more; weekly, to specify repeating periods based on an interval of a week or more; monthly, to specify repeating periods based on an interval of a month or more; and yearly, to specify repeating periods based on an interval of a year or more. These values are not case-sensitive. The interval parameter contains a positive integer representing how often the recurrence rule repeats. The default value is "1", meaning every second for a secondly rule, or every minute for a minutely rule, every hour for an hourly rule, every day for a daily rule, every week for a weekly rule, every month for a monthly rule and every year for a yearly rule. The until parameter defines an iCal COS DATE or DATE-TIME value which bounds the recurrence rule in an inclusive manner. If the value specified by until is synchronized with the specified recurrence, this date or date-time becomes the last instance of the recurrence. If specified as a date-time value, then it MUST be specified in an UTC time format. If not present, and the count parameter is also not present, the recurrence is considered to repeat forever. The count parameter defines the number of occurrences at which to range-bound the recurrence. The dtstart parameter counts as the first occurrence. The until and count parameters MUST NOT occur in the same time output. The bysecond parameter specifies a comma-separated list of seconds within a minute. Valid values are 0 to 59. The byminute parameter specifies a comma-separated list of minutes within an hour. Valid values are 0 to 59. The byhour parameter specifies a comma-separated list of hours of the day. Valid values are 0 to 23. The byday parameter specifies a comma-separated list of days of the week. MO indicates Monday; TU indicates Tuesday; WE indicates Wednesday; TH indicates Thursday; FR indicates Friday; SA indicates Saturday; SU indicates Sunday. These values are not case-sensitive. Each byday value can also be preceded by a positive (+n) or negative Lennox/Schulzrinne [Page 17] Internet Draft CPL July 14, 2000 (-n) integer. If present, this indicates the nth occurrence of the specific day within the monthly or yearly recurrence. For example, within a monthly rule, +1MO (or simply 1MO) represents the first Monday within the month, whereas -1MO represents the last Monday of the month. If an integer modifier is not present, it means all days of this type within the specified frequency. For example, within a monthly rule, MO represents all Mondays within the month. The bymonthday parameter specifies a comma-separated list of days of the month. Valid values are 1 to 31 or -31 to -1. For example, -10 represents the tenth to the last day of the month. The byyearday parameter specifies a comma-separated list of days of the year. Valid values are 1 to 366 or -366 to -1. For example, -1 represents the last day of the year (December 31st) and -306 represents the 306th to the last day of the year (March 1st). The byweekno parameter specifies a comma-separated list of ordinals specifying weeks of the year. Valid values are 1 to 53 or -53 to -1. This corresponds to weeks according to week numbering as defined in [ISO 8601]. A week is defined as a seven day period, starting on the day of the week defined to be the week start (see wkst). Week number one of the calendar year is the first week which contains at least four (4) days in that calendar year. This parameter is only valid for yearly rules. For example, 3 represents the third week of the year. Note: Assuming a Monday week start, week 53 can only occur when Thursday is January 1 or if it is a leap year and Wednesday is January 1. The bymonth parameter specifies a comma-separated list of months of the year. Valid values are 1 to 12. The wkst parameter specifies the day on which the workweek starts. Valid values are MO, TU, WE, TH, FR, SA and SU. This is significant when a weekly recurrence has an interval greater than 1, and a byday parameter is specified. This is also significant in a yearly recurrence when a byweekno parameter is specified. The default value is MO. The bysetpos parameter specifies a comma-separated list of values which corresponds to the nth occurrence within the set of events specified by the rule. Valid values are 1 to 366 or -366 to -1. It MUST only be used in conjunction with another byxxx parameter. For example "the last work day of the month" could be represented as: Lennox/Schulzrinne [Page 18] Internet Draft CPL July 14, 2000