Element Documentation

how to write Click element documentation

Documentation for a Click element is automatically generated from formatted comments in its header files. This manual page describes how to write one of these formatted comments.

Click element documentation syntax is based on Perl-style PODs; see perlpod(1) for more information. However, not all POD constructs are usable in element documentation at the moment; and Click element documentation is line-oriented, not paragraph-oriented.

These are the sectioning commands and the sections they correspond to, listed in the conventional order.

=c

SYNOPSIS

=s category

NAME (one-line summary)

=io

INPUTS AND OUTPUTS

=processing

PROCESSING TYPE

=d

DESCRIPTION

=n

NOTES

=e

EXAMPLES

=h name type

ELEMENT HANDLERS

=a

SEE ALSO

=head1 text

other heading

These are the other commands used in element documentation.

=head2 text

subheading

=over amount

begin item list

=item text

add item

=back

end item list

=for format

selective formatting

=begin format

selective formatting

=end format

selective formatting

=deprecated new-element

element is deprecated

These are the formatting commands, used inside ordinary text.

B<text>

boldface

I<text>

italics

U<text>

underlined (if possible)

P<text>

plain text (i.e., no automatic links)

C<text>

code (fixed-width)

F<text>

filename (italics)

L<text|link>

hyperlink

N<>

line break

E<name>

entity

V<text>

hide text

And these are the category keywords, used in the summary section to categorize elements.

basicsources

Basic Sources and Sinks

classification

Basic Classification and Selection

basictransfer

Basic Packet Transfer

counters

Counters

timestamps

Timestamps

basicmod

Basic Packet Modification

storage

Packet Storage

aqm

Active Queue Management

scheduling

Packet Scheduling

shaping

Traffic Shaping

information

Information Elements

netdevices

Network Devices

comm

Host and Socket Communication

ethernet

Ethernet

arp

ARP

ip

IPv4

iproute

IPv4 Routing

icmp

ICMP

nat

Network Address Translation

tcp

TCP

udp

UDP

app

Applications

traces

Trace Manipulation

ipmeasure

TCP/IP Measurement

aggregates

Aggregates

ip6

IPv6

ipsec

IPsec

crc

CRCs

paint

Paint Annotations

annotations

Annotations

debugging

Debugging

control

Control

smpclick

Multithreaded Click

test

Regression Tests

Each piece of documentation is stored in a single block comment '/*...*/'. Two kinds of block comment are recognized:

    /*
    =c
    ElementName(...)
    ... and so on ...
    */

    /*
     * =c
     * ElementName(...)
     * ... and so on ...
     */

In the first form, commands and normal text MUST begin in the first column of each line. In the second form, commands and normal text MUST begin in the fourth column of each line, immediately following the initial star and spaces ' * '. Indented lines are treated as verbatim text, as in POD.

Commands are lines that begin with '=' and a lower-case letter. There are two kinds of commands, those that start new sections and those that occur within sections. There is also a set of formatting commands -- for turning text bold, for example -- that occurs inside normal text; they are described in the next section.

=s [category]

Begin the summary section. This should contain a very short, one-line summary of the element's function. For example, for a Queue(n) element: =s storage stores packets in a FIFO queue The summary text should generally be a verb phrase. The optional category specifies one or more element categories into which this element fits, separated by commas. Specifying meaningful categories helps a lot; documentation tools use categories to divide elements into manageable groups. Use existing categories, defined by the list of category keywords above in the Command Quick Reference, or create your own.

=c

Begin the synopsis section. This section is mandatory. The =c section gives the element's name and any of its configuration arguments. For example: =c IPEncap(PROTOCOL, SADDR, DADDR) Configuration arguments should be specified as all-upper-case words. The description section will use those upper-case words to talk about the arguments. Use brackets to show that an argument is optional: =c UDPIPEncap(SADDR, SPORT, DADDR, DPORT [, CHECKSUM?]) Do not use anything more complicated than brackets. If an element has complex syntax, either use upper-case words and talk about the syntax more in the description section, or give multiple lines: =c ControlSocket(tcp, PORTNUMBER [, READONLY?]) ControlSocket(unix, PORTNUMBER [, READONLY?]) ('tcp' and 'unix' are lowercase because they should be typed verbatim.)

=io

Begin the inputs and outputs section. This section mentions how many inputs and outputs the element has. It is usually quite short; for example: =io None This section is optional, and most elements don't bother to have one; they mention inputs and outputs in the description section.

=processing

Begin the processing type section. This section mentions the processing types of the element's input and output ports. It is usually quite short; for example: =processing Push inputs, pull outputs This section is optional. Documentation processing tools will generate a =processing section from the element's processing() method, if possible.

=d

Begin the description section. This section tells how the element should be used. It is usually the longest section. When mentioning configuration arguments, use the upper-case words given in the =c section.

=n

Begin the notes section.

=e

Begin the examples section.

=h handlername type

Begin a handler description. Use this section to describe any special handlers that the element installs. Handlername should be the name of the handler, and type its type (either 'read-only', 'write-only', or 'read/write'). The following text should describe that handler. For example: =h capacity read/write Returns or sets the queue's capacity.

=a [text]

Begin the "see also" section. Use this section to mention other relevant elements and programs, when appropriate. The more references, the better. For example: =a RED, FrontDropQueue The optional text is just part of the body of the section. The references in this section should be either manual page references, like 'tcpdump(1)', or text references, like 'RFC 959: File Transfer Protocol'. However, the first paragraph in the section is special; there, you can just give element names without '(n)' suffixes. If one of these references occurs in some other section, it will be formatted like a link. For example, in =d This element is like Queue. =a Queue the mention of 'Queue' in the description section will be formatted like a link.

=head1 sectionname

Begin a section other than those listed. Sectionname is the name of the section.

=head2 text

Produce a subheading with text as the text.

=over amount

Begin a list of items that is indented by amount characters. (Some translators may ignore amount.)

=item text

Add an item to the latest list opened by =over. It is illegal to use =item outside of any =over list. The text of the item is text. If you are creating a bulleted list, use '*' as the text; if you are creating a numbered list, use '1.', '2.', and so forth.

=back

Close the latest list opened by =over.

=for format

Output the next paragraph only when generating documentation in format. Valid formats include 'html', 'man', and 'roff'. The paragraph ends at the next command or blank line, and it consists of text in the given format, not element documentation. For example, this code includes a picture in HTML mode: =for html <p>Here is a pretty picture: <img src="pretty.gif"></p> Back to B<normal text> here.

=begin format ... =end format

This is like =for, but can encompass multiple paragraphs. It outputs text in between the =begin command and the =end command only when generating documentation in format.

=deprecated new-element

This command notes that the element has been deprecated in favor of new-element. It does not result in any output.

Each line that doesn't begin with '=' and a lower-case letter is treated as text. (Unless it starts with a space or tab; see verbatim text, below.) This text is formatted nicely, and perhaps even justified. You can use several formatting commands inside normal text; they consist of an uppercase letter, followed by '<', some text, and '>'. The commands are:

B<text>

Print text in boldface.

I<text>

Print text in italic.

R<text>

Print text in roman. Useful inside B<...> and so forth, or to prevent words from being highlighted.

C<text>

Print text like source code, in a constant-width font.

F<text>

Print text like a filename. By default, filenames appear in italics.

L<text|link>

Print text as a hyperlink with destination link. This usually just comes out as text.

N<>

Put a line break here.

E<name>

Print the HTML-style entity named name. There are six entities: E<lt> is '<', E<gt> is '>', E<amp> is '&', E<solid> is '/', E<verbar> is '|', E<star> is '*', and E<eq> is '='. This is useful for typing one of these characters in a context that would seem like a command or formatting command.

V<text>

Do not print text.

Lines that start with a space or tab character are printed out verbatim -- that is, without any changes, and with the line breaks and indentation you specified. You can't use formatting commands in verbatim text. Verbatim text is useful for showing example code; for example:

  This code
     q :: Queue;
     ... -> RED(5, 50, 0.02) -> q -> ...
  adds RED dropping to q.

/* =c
 * Align(MODULUS, OFFSET)
 * =s modification
 * aligns packet data
 * =d
 * Aligns packet data. Each input packet is aligned so that
 * its first byte is OFFSET bytes off from a MODULUS-byte
 * boundary. This may involve a packet copy.
 *
 * MODULUS I<must> be 2, 4, or 8.
 * =n
 * The click-align(1) tool will insert this element 
 * automatically wherever it is required.
 * =e
 *   ... -> Align(4, 0) -> ...
 * =a AlignmentInfo, click-align(1) */

/* =c
 * Counter([TYPE])
 * =s measurement
 * measures packet count and rate
 * =d
 * Passes packets unchanged from its input to its output,
 * maintaining statistics information about packet count and
 * rate if TYPE is "packets", or byte count and byte rate if
 * TYPE is "bytes". The default TYPE is "packets".
 * =h count read-only
 * Returns the number of packets/bytes that have passed through.
 * =h rate read-only
 * Returns the recent arrival rate (measured by exponential
 * weighted moving average) in packets/bytes per second.
 * =h reset write-only
 * Resets the count and rate to zero.
 */

perlpod(1), click(1), click(5)

Eddie Kohler, kohler@cs.ucla.edu
Robert Morris, rtm@lcs.mit.edu
http://www.pdos.lcs.mit.edu/click/