% \iffalse meta-comment
%
% Copyright (C) 2011 by Scott Pakin <scott+bf@pakin.org>
% ------------------------------------------------------
%
% This file may be distributed and/or modified under the
% conditions of the LaTeX Project Public License, either version 1.3a
% of this license or (at your option) any later version.
% The latest version of this license is in:
%
%    http://www.latex-project.org/lppl.txt
%
% and version 1.3c or later is part of all distributions of LaTeX
% version 2006/05/20 or later.
%
% \fi
%
% \iffalse
%<*driver>
\ProvidesFile{bytefield.dtx}
%</driver>
%<package>\NeedsTeXFormat{LaTeX2e}[1999/12/01]
%<package>\ProvidesPackage{bytefield}
%<*package>
    [2011/06/19 v2.1 Network protocol diagrams]
%</package>
%
%<*driver>
\documentclass{ltxdoc}
\usepackage{textcomp}
\usepackage{bytefield}
\usepackage{color}
\usepackage{rotating}
\usepackage{multirow}
\usepackage{calc}
\usepackage{array}
\usepackage{wasysym}
\IfFileExists{hypdoc.sty}{\usepackage{hypdoc}}{}
\usepackage{hyperref}
\IfFileExists{hypcap.sty}{\usepackage[figure]{hypcap}}{}

\hypersetup{%
  pdftitle={The bytefield package},
  pdfauthor={Scott Pakin <scott+bf@pakin.org>},
  pdfsubject={Protocol diagrams for LaTeX},
  pdfkeywords={bits, bytes, bit fields, communication, network protocol
    diagrams, LaTeX2e, memory maps},
  colorlinks=false,
  bookmarksopen=true}
\EnableCrossrefs
\CodelineIndex
\RecordChanges
\setcounter{IndexColumns}{2}
\begin{document}
  \DocInput{bytefield.dtx}
  \PrintChanges
  \PrintIndex
\end{document}
%</driver>
% \fi
%
% \CheckSum{917}
%
% \CharacterTable
%  {Upper-case    \A\B\C\D\E\F\G\H\I\J\K\L\M\N\O\P\Q\R\S\T\U\V\W\X\Y\Z
%   Lower-case    \a\b\c\d\e\f\g\h\i\j\k\l\m\n\o\p\q\r\s\t\u\v\w\x\y\z
%   Digits        \0\1\2\3\4\5\6\7\8\9
%   Exclamation   \!     Double quote  \"     Hash (number) \#
%   Dollar        \$     Percent       \%     Ampersand     \&
%   Acute accent  \'     Left paren    \(     Right paren   \)
%   Asterisk      \*     Plus          \+     Comma         \,
%   Minus         \-     Point         \.     Solidus       \/
%   Colon         \:     Semicolon     \;     Less than     \<
%   Equals        \=     Greater than  \>     Question mark \?
%   Commercial at \@     Left bracket  \[     Backslash     \\
%   Right bracket \]     Circumflex    \^     Underscore    \_
%   Grave accent  \`     Left brace    \{     Vertical bar  \|
%   Right brace   \}     Tilde         \~}
%
%
% \changes{v1.0}{2000/07/02}{Initial version}
% \changes{v1.1}{2002/09/14}{Restructured the \texttt{.dtx} file}
% \changes{v1.4}{2011/01/09}{Made assignments to \cs{counting@words}
%   global to prevent vertical-spacing problems with back-to-back word
%   groups (bug fix due to Steven R. King)}
% \changes{v1.4}{2011/01/16}{Split \cs{curlyspace}, \cs{labelspace}, and
%   \cs{curlyshrinkage} into \texttt{left} and \texttt{right} versions}
% \changes{v2.0}{2011/01/18}{Made a number of non-backwards-compatible
%   changes, including replacing \cs{wordgroupr} and \cs{endwordgroupr}
%   with a \texttt{rightwordgroup} environment and \cs{wordgroupl} and
%   \cs{endwordgroupl} with a \texttt{leftwordgroup} environment and
%   also replacing a slew of user-visible lengths and macros with a
%   single \cs{bytefieldsetup} macro}
%
% \GetFileInfo{bytefield.dtx}
%
% \DoNotIndex{\&,\{,\}}
% \DoNotIndex{\addtocounter,\advance}
% \DoNotIndex{\begin,\begingroup,\bgroup}
% \DoNotIndex{\catcode,\ch@ck,\count,\cr,\csname}
% \DoNotIndex{\DeclareRobustCommand,\def,\dimen,\dimendef}
% \DoNotIndex{\edef,\egroup,\else,\end,\endcsname,\endgroup,\expandafter}
% \DoNotIndex{\fi}
% \DoNotIndex{\gdef,\global}
% \DoNotIndex{\hspace}
% \DoNotIndex{\if,\ifnum,\ifx,\ignorespaces,\insc@unt}
% \DoNotIndex{\left,\let,\loop}
% \DoNotIndex{\MessageBreak}
% \DoNotIndex{\newbox,\newcommand,\newcounter,\newenvironment,\newif,\next}
% \DoNotIndex{\newlength,\newsavebox,\noexpand}
% \DoNotIndex{\raisebox,\ratio,\real,\relax,\renewcommand,\right}
% \DoNotIndex{\par,\protect}
% \DoNotIndex{\renewenvironment,\repeat}
% \DoNotIndex{\sbox,\setcounter,\setlength,\settowidth,\space,\string}
% \DoNotIndex{\the}
% \DoNotIndex{\unskip,\usebox,\usepackage}
% \DoNotIndex{\value,\vbox,\vspace}
% \DoNotIndex{\wlog}
% \DoNotIndex{\xdef}
%
%
% \title{The \textsf{bytefield} package\thanks{This document
%   corresponds to \textsf{bytefield}~\fileversion, dated \filedate.}}
% \author{Scott Pakin \\ \texttt{scott+bf@pakin.org}}
%
% \sloppy
% \maketitle
%
% ^^A  Define an environment for command/environment delcarations.
% \newsavebox{\declbox}
% \newenvironment{decl}{^^A
%   \begin{lrbox}{\declbox}\begin{tabular}{l}}{^^A
%   \end{tabular}\end{lrbox}^^A
%   \vspace{3ex}\noindent\hspace*{-3em}\fbox{\usebox{\declbox}}\vspace*{2ex}}
%
% ^^A  Define an ellipsis that permits a subsequent line break.
% \newcommand{\ellipsis}{$\ldots$\linebreak[0]}
%
% ^^A  Define an environment for showing sample bytefield diagrams.
% \newenvironment{bffigure}{%
%   \bigskip
%   \noindent\hspace*{3em}%
% }{%
%   \bigskip
% }
%
% ^^A  Define something like \cs to use within \StopEventually.
% \DeclareRobustCommand{\cseq}[1]{\texttt{\string#1}}
%
% ^^A  Index a package.  This code was adapted from \SpecialEnvIndex.
% \makeatletter
% \def\SpecialPkgIndex#1{^^A
%   \@bsphack
%     \index{#1\actualchar{\protect\sffamily#1} (package)\encapchar usage}^^A
%     \index{packages:\levelchar#1\actualchar{\protect\sffamily#1}^^A
%       \encapchar usage}^^A
%   \@esphack
% }
%
% ^^A  The following was adapted from hypdoc's redefinition of \SpecialEnvIndex.
% \expandafter\ifx\csname c@HD@hypercount\endcsname\relax
% \else
%   \let\HDorg@SpecialPkgIndex\SpecialPkgIndex
%   \renewcommand*\SpecialPkgIndex[1]{^^A
%     \@bsphack
%     \begingroup
%       \HD@target
%       \let\HDorg@encapchar\encapchar
%       \edef\encapchar usage{^^A
%         \HDorg@encapchar hdpindex{usage}^^A
%       }^^A
%       \HDorg@SpecialPkgIndex{#1}^^A
%     \endgroup
%     \@esphack
%   }
% \fi
% \makeatother
%
% ^^A  Define a logical style for package names.
% \DeclareRobustCommand{\pkgname}[1]{^^A
%   \textsf{#1}^^A
%   \SpecialPkgIndex{#1}^^A
% }
%
%
% ^^A  Index an option.  This code was adapted from \SpecialEnvIndex.
% \makeatletter
% \def\SpecialOptionIndex#1{^^A
%   \@bsphack
%     \index{#1\actualchar{\protect\sffamily#1} (option)\encapchar usage}^^A
%     \index{options:\levelchar#1\actualchar{\protect\sffamily#1}^^A
%       \encapchar usage}^^A
%   \@esphack
% }
%
% ^^A  The following was adapted from hypdoc's redefinition of \SpecialEnvIndex.
% \expandafter\ifx\csname c@HD@hypercount\endcsname\relax
% \else
%   \let\HDorg@SpecialOptionIndex\SpecialOptionIndex
%   \renewcommand*\SpecialOptionIndex[1]{^^A
%     \@bsphack
%     \begingroup
%       \HD@target
%       \let\HDorg@encapchar\encapchar
%       \edef\encapchar usage{^^A
%         \HDorg@encapchar hdpindex{usage}^^A
%       }^^A
%       \HDorg@SpecialOptionIndex{#1}^^A
%     \endgroup
%     \@esphack
%   }
% \fi
% \makeatother
%
% ^^A  Define a logical style for package-option names.
% \DeclareRobustCommand{\optname}[1]{^^A
%   \texttt{#1}^^A
%   \SpecialOptionIndex{#1}^^A
% }
%
% ^^A  We use this box to help center protocol diagrams.
% \newsavebox{\protocoldiagram}
%
%
% \begin{abstract}
% The \pkgname{bytefield} package helps the user create illustrations for
% network protocol specifications and anything else that utilizes fields of
% data.  These illustrations show how the bits and bytes are laid out in a
% packet or in memory.
% \end{abstract}
%
%
% \vspace{\baselineskip}
% \begin{center}
% \fbox{%
%   \begin{minipage}{0.65\linewidth}
%     \textsc{Warning}: \pkgname{bytefield} version~2.\textit{x} breaks
%     compatibility with older versions of the package.  See
%     Section~\ref{sec:upgrading} for help porting documents to the new
%     interface.
%   \end{minipage}%
% }
% \end{center}
% \vspace{\baselineskip}
%
%
% \section{Introduction}
%
% Network protocols are usually specified in terms of a sequence of bits
% and bytes arranged in a field.  This is portrayed graphically as a
% grid of boxes.  Each row in the grid represents one word (frequently,
% 8, 16, or 32~bits), and each column represents a bit within a word.
% The \pkgname{bytefield} package makes it easy to typeset these sorts
% of figures.  \pkgname{bytefield} facilitates drawing protocol diagrams
% that contain
%
% \begin{itemize}
%   \item words of any arbitrary number of bits,
%   \item column headers showing bit positions,
%   \item multiword fields---even non-word-aligned and even if the total
%         number of bits is not a multiple of the word length,
%   \item word labels on either the left or right of the figure, and
%   \item ``skipped words'' within fields.
% \end{itemize}
%
% Because \pkgname{bytefield} draws its figures using only
% the \LaTeX{} \texttt{picture} environment, these figures are not
% specific to any particular backend, do not require PostScript support,
% and do not need support from external programs.  Furthermore, unlike an
% imported graphic, \pkgname{bytefield} pictures can include arbitrary
% \LaTeX{} constructs, such as mathematical equations, |\ref|s and
% |\cite|s to the surrounding document, and macro calls.
%
%
% \section{Usage}
% \label{sec:usage}
%
% \subsection{A first example}
% \label{sec:first-example}
%
% The Internet Engineering Task Force's Request for Comments (RFC)
% number~3016 includes the following ASCII-graphics illustration of
% the RTP packetization of an \mbox{MPEG-4} Visual bitstream:
%
% \begin{verbatim}
%       0                   1                   2                   3
%       0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
%       +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
%       |V=2|P|X|  CC   |M|     PT      |       sequence number         | RTP
%       +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
%       |                           timestamp                           | Header
%       +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
%       |           synchronization source (SSRC) identifier            |
%       +=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+
%       |            contributing source (CSRC) identifiers             |
%       |                             ....                              |
%       +=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+
%       |                                                               | RTP
%       |       MPEG-4 Visual stream (byte aligned)                     | Pay-
%       |                                                               | load
%       |                               +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
%       |                               :...OPTIONAL RTP padding        |
%       +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
% \end{verbatim}
%
% \noindent
% The following \LaTeX\ code shows how straightforward it is to typeset
% that illustration using the \pkgname{bytefield} package:
%
% \begin{verbatim}
%       \begin{bytefield}[bitwidth=1.1em]{32}
%         \bitheader{0-31} \\
%         \begin{rightwordgroup}{RTP \\ Header}
%           \bitbox{2}{V=2} & \bitbox{1}{P} & \bitbox{1}{X}
%           & \bitbox{4}{CC} & \bitbox{1}{M} & \bitbox{7}{PT}
%           & \bitbox{16}{sequence number} \\
%           \bitbox{32}{timestamp}
%         \end{rightwordgroup} \\
%         \bitbox{32}{synchronization source (SSRC) identifier} \\
%         \wordbox[tlr]{1}{contributing source (CSRC) identifiers} \\
%         \wordbox[blr]{1}{$\cdots$} \\
%         \begin{rightwordgroup}{RTP \\ Payload}
%           \wordbox[tlr]{3}{MPEG-4 Visual stream (byte aligned)} \\
%           \bitbox[blr]{16}{}
%           & \bitbox{16}{\dots\emph{optional} RTP padding}
%         \end{rightwordgroup}
%       \end{bytefield}
% \end{verbatim}
%
% Figure~\ref{fig:rtp-mpeg4} presents the typeset output of the
% preceding code.  Sections~\ref{sec:basic-cmds} and~\ref{sec:options}
% explain each of the environments, macros, and arguments that were
% utilized plus many additional features of the \pkgname{bytefield}
% package.
%
% \begin{figure}[hbp]
%     \centering
%     \begin{lrbox}{\protocoldiagram}
%       \begin{bytefield}[bitwidth=1.1em]{32}
%         \bitheader{0-31} \\
%         \begin{rightwordgroup}{RTP \\ Header}
%           \bitbox{2}{V=2} & \bitbox{1}{P} & \bitbox{1}{X}
%           & \bitbox{4}{CC} & \bitbox{1}{M} & \bitbox{7}{PT}
%           & \bitbox{16}{sequence number} \\
%           \bitbox{32}{timestamp}
%         \end{rightwordgroup} \\
%         \bitbox{32}{synchronization source (SSRC) identifier} \\
%         \wordbox[tlr]{1}{contributing source (CSRC) identifiers} \\
%         \wordbox[blr]{1}{$\cdots$} \\
%         \begin{rightwordgroup}{RTP \\ Payload}
%           \wordbox[tlr]{3}{MPEG-4 Visual stream (byte aligned)} \\
%           \bitbox[blr]{16}{}
%           & \bitbox{16}{\dots\emph{optional} RTP padding}
%         \end{rightwordgroup}
%       \end{bytefield}
%     \end{lrbox}
%     \makebox[0pt][c]{\usebox{\protocoldiagram}}
%     \caption{Sample \pkgname{bytefield} output}
%    \label{fig:rtp-mpeg4}
% \end{figure}
%
%
% \subsection{Basic commands}
% \label{sec:basic-cmds}
%
% This section explains how to use the \pkgname{bytefield} package.  It
% lists all the exported macros and environments in approximately
% decreasing order of usefulness.
%
% \begin{decl}
%   \SpecialEnvIndex{bytefield}
%   |\begin{bytefield}| \oarg{parameters} \marg{bit-width} \\
%   \meta{fields} \\
%   |\end{bytefield}|
% \end{decl}
%
% The \pkgname{bytefield} package's top-level environment is called, not
% surprisingly, ``|bytefield|''.  It takes one mandatory argument, which
% is the number of bits in each word, and one optional argument, which
% is a set of parameters, described in Section~\ref{sec:options}, for
% formatting the bit-field's layout.  One can think of a |bytefield| as
% being analogous to a |tabular|: words are separated by~``|\\|'', and
% fields within a word are separated by~``|&|''.  As in a |tabular|,
% ``|\\|'' accepts a \meta{length} as an optional argument, and this
% specifies the amount of additional vertical whitespace to include
% after the current word is typeset.
%
% \begin{decl}
%   \SpecialUsageIndex{\bitbox}
%   |\bitbox| \oarg{sides} \marg{width} \marg{text} \\
%   \SpecialUsageIndex{\wordbox}
%   |\wordbox| \oarg{sides} \marg{height} \marg{text}
% \end{decl}
%
% The two main commands one uses within a \pkgname{bytefield}
% environment are |\bitbox| and |\wordbox|.  The former typesets a field
% that is one or more bits wide and a single word tall.  The latter
% typesets a field that is an entire word wide and one or more words
% tall.
%
% The optional argument, \meta{sides}, is a list of letters specifying
% which sides of the field box to draw---[|l|]eft, [|r|]ight, [|t|]op,
% and/or [|b|]ottom.  The default is ``|lrtb|'' (i.e.,~all sides are
% drawn).  \meta{text} is the text to include within the |\bitbox| or
% |\wordbox|.  It is typeset horizontally centered within a vertically
% centered |\parbox|.  Hence, words will wrap, and ``|\\|'' can be used
% to break lines manually.
%
% The following example shows how to produce a simple 16-bit-wide field:
%
% \begin{verbatim}
%       \begin{bytefield}{16}
%         \wordbox{1}{A 16-bit field} \\
%         \bitbox{8}{8 bits} & \bitbox{8}{8 more bits} \\
%         \wordbox{2}{A 32-bit field.  Note that text wraps within the box.}
%       \end{bytefield}
% \end{verbatim}
%
% The resulting bit field looks like this:
%
% \begin{bffigure}
%       \begin{bytefield}{16}
%         \wordbox{1}{A 16-bit field} \\
%         \bitbox{8}{8 bits} & \bitbox{8}{8 more bits} \\
%         \wordbox{2}{A 32-bit field.  Note that text wraps within the box.}
%       \end{bytefield}
% \end{bffigure}
%
% It is the user's responsibility to ensure that the total number of
% bits in each row adds up to the number of bits in a single word (the
% mandatory argument to the |bytefield| environment);
% \pkgname{bytefield} does not currently check for under- or overruns.
%
% Within a |\bitbox| or |\wordbox|, the \pkgname{bytefield} package
% defines |\height|, |\depth|, |\totalheight|, and |\width| to the
% corresponding dimensions of the box.  Section~\ref{sec:tricks} gives an
% example of how these lengths may be utilized.
%
% \begin{decl}
%   \SpecialUsageIndex{\bitheader}
%   |\bitheader| \oarg{parameters} \marg{bit-positions}
% \end{decl}
%
% To make the bit field more readable, it helps to label bit positions
% across the top.  The |\bitheader| command provides a flexible way to
% do that.  The optional argument is a set of parameters from the set
% described in Section~\ref{sec:options}.  In practice, the only
% parameters that are meaningful in the context of |\bitheader| are
% |bitformatting|, |endianness|, and |lsb|.  See
% Section~\ref{sec:options} for descriptions and examples of those
% parameters.
%
% |\bitheader|'s mandatory argument, \meta{bit-positions}, is a
% comma-separated list of bit positions to label.  For example,
% ``|0,2,4,6,8,10,12,14|'' means to label those bit positions.  The
% numbers must be listed in increasing order.  (Use the
% \optname{endianness} parameter to display the header in reverse
% order.)  Hyphen-separated ranges are also valid.  For example,
% ``|0-15|'' means to label all bits from~0 to~15, inclusive.  Ranges
% and single numbers can even be intermixed, as in ``|0-3,8,12-15|''.
%
% The following example shows how |\bitheader| may be used:
%
% \begin{verbatim}
%       \begin{bytefield}{32}
%         \bitheader{0-31} \\
%         \bitbox{4}{Four} & \bitbox{8}{Eight} &
%           \bitbox{16}{Sixteen} & \bitbox{4}{Four}
%       \end{bytefield}
% \end{verbatim}
%
% The resulting bit field looks like this:
%
% \begin{bffigure}
%       \begin{bytefield}{32}
%         \bitheader{0-31} \\
%         \bitbox{4}{Four} & \bitbox{8}{Eight} &
%           \bitbox{16}{Sixteen} & \bitbox{4}{Four}
%       \end{bytefield}
% \end{bffigure}
%
%
% \begin{decl}
%   \SpecialEnvIndex{rightwordgroup}
%   |\begin{rightwordgroup}| \marg{text} \\
%   \meta{rows of bit boxes and word boxes} \\
%   |\end{rightwordgroup}| \\
%   \\
%   \SpecialEnvIndex{leftwordgroup}
%   |\begin{leftwordgroup}| \marg{text} \\
%   \meta{rows of bit boxes and word boxes} \\
%   |\end{leftwordgroup}|
% \end{decl}
%
% When a set of words functions as a single, logical unit, it helps to
% group these words together visually.  All words defined between
% |\begin{rightwordgroup}| and |\end{rightwordgroup}| will be labeled on
% the right with \meta{text}.  Similarly, all words defined between
% |\begin{leftwordgroup}| and |\end{leftwordgroup}| will be labeled on
% the left with \meta{text}.  |\begin{|\meta{side}|wordgroup}| must lie
% at the beginning of a row (i.e.,~right after a ``|\\|''), and
% |\end{|\meta{side}|wordgroup}| must lie right \emph{before} the end of
% the row (i.e.,~right before a ``|\\|'').
%
% Unlike other \LaTeX\ environments, |rightwordgroup| and
% |leftwordgroup| do not have to nest properly with each other.
% However, they cannot overlap themselves.  In other words,
% |\begin{rightwordgroup}|\ellipsis |\begin{leftwordgroup}|\ellipsis
% |\end{rightwordgroup}|\ellipsis |\end{leftwordgroup}| is a valid
% sequence, but |\begin{rightwordgroup}|\ellipsis
% |\begin{rightwordgroup}|\ellipsis |\end{rightwordgroup}|\ellipsis
% |\end{rightwordgroup}| is not.
%
% The following example presents the basic usage of
% |\begin{rightwordgroup}| and |\end{rightwordgroup}|:
%
% \begin{verbatim}
%       \begin{bytefield}{16}
%         \bitheader{0,7,8,15} \\
%         \begin{rightwordgroup}{Header}
%           \bitbox{4}{Tag} & \bitbox{12}{Mask} \\
%           \bitbox{8}{Source} & \bitbox{8}{Destination}
%         \end{rightwordgroup} \\
%         \wordbox{3}{Data}
%       \end{bytefield}
% \end{verbatim}
%
% \noindent
% Note the juxtaposition of ``|\\|'' to the |\begin{rightwordgroup}| and
% the |\end{rightwordgroup}| in the above.  The resulting bit field
% looks like this:
%
% \begin{bffigure}
%       \begin{bytefield}{16}
%         \bitheader{0,7,8,15} \\
%         \begin{rightwordgroup}{Header}
%           \bitbox{4}{Tag} & \bitbox{12}{Mask} \\
%           \bitbox{8}{Source} & \bitbox{8}{Destination}
%         \end{rightwordgroup} \\
%         \wordbox{3}{Data}
%       \end{bytefield}
% \end{bffigure}
%
% \noindent
% As a more complex example, the following nests left and right labels:
%
% \begin{verbatim}
%       \begin{bytefield}{16}
%         \bitheader{0,7,8,15} \\
%         \begin{rightwordgroup}{Header}
%           \bitbox{4}{Tag} & \bitbox{12}{Mask} \\
%           \begin{leftwordgroup}{Node IDs}
%             \bitbox{8}{Source} & \bitbox{8}{Destination}
%           \end{leftwordgroup}
%         \end{rightwordgroup} \\
%         \wordbox{3}{Data}
%       \end{bytefield}
% \end{verbatim}
%
% \begin{bffigure}
%       \begin{bytefield}{16}
%         \bitheader{0,7,8,15} \\
%         \begin{rightwordgroup}{Header}
%           \bitbox{4}{Tag} & \bitbox{12}{Mask} \\
%           \begin{leftwordgroup}{Node IDs}
%             \bitbox{8}{Source} & \bitbox{8}{Destination}
%           \end{leftwordgroup}
%         \end{rightwordgroup} \\
%         \wordbox{3}{Data}
%       \end{bytefield}
% \end{bffigure}
%
% \noindent
% Because |rightwordgroup| and |leftwordgroup| are not required to nest
% properly, the resulting bit field would look the same if the
% |\end{leftwordgroup}| and |\end{rightwordgroup}| were swapped.  Again,
% note the justaposition of ``|\\|'' to the various word-grouping
% commands in the above.
%
% \begin{decl}
%   \SpecialUsageIndex{\skippedwords}
%   |\skippedwords|
% \end{decl}
%
% Draw a graphic representing a number of words that are not shown.
% |\skippedwords| is intended to work with the \meta{sides} argument
% to |\wordbox|, as in the following example:
%
% \begin{verbatim}
%       \begin{bytefield}{16}
%         \wordbox{1}{Some data} \\
%         \wordbox[lrt]{1}{Lots of data} \\
%         \skippedwords \\
%         \wordbox[lrb]{1}{} \\
%         \wordbox{2}{More data}
%       \end{bytefield}
% \end{verbatim}
%
% \begin{bffigure}
%       \begin{bytefield}{16}
%         \wordbox{1}{Some data} \\
%         \wordbox[lrt]{1}{Lots of data} \\
%         \skippedwords \\
%         \wordbox[lrb]{1}{} \\
%         \wordbox{2}{More data}
%       \end{bytefield}
% \end{bffigure}
%
% \begin{decl}
%   \SpecialUsageIndex{\bytefieldsetup}
%   |\bytefieldsetup| \marg{key=value list}
% \end{decl}
%
% Alter the formatting of all subsequent bit fields.
% Section~\ref{sec:options} describes the possible values for each
% \meta{key}=\meta{value} item in the (comma-separated) list that
% |\bytefieldsetup| accepts as its argument.  Note that changes made
% with |\bytefieldsetup| are local to their current scope.  Hence, if
% used within an environment (e.g.,~|figure|), |\bytefieldsetup| does
% not impact bit fields drawn outside that environment.
%
%
% \subsection{Formatting options}
% \label{sec:options}
%
% A document author can customize many of the \pkgname{bytefield}
% package's figure-formatting parameters, either globally or on a
% per-figure basis.  The parameters described below can be specified in
% four locations:
%
% \begin{itemize}
%   \item as package options (i.e.,~in the
%     |\usepackage|\oarg{options}|{bytefield}| line), which affects all
%     |bytefield| environments in the entire document,
%
%   \item anywhere in the document using the |\bytefieldsetup| command,
%     which affects all subsequent |bytefield| environments in the
%     current scope,
%
%   \item as the optional argument to a |\begin{bytefield}|, which
%     affects only that single bit-field figure, or
%
%   \item as the optional argument to a |\bitheader|, which affects only
%     that particular header.  (Only a few parameters are meaningful in
%     this context.)
% \end{itemize}
%
% Unfortunately, \LaTeX\ tends to abort with a ``\texttt{TeX capacity
%   exceeded}'' or ``\texttt{Missing \cs{endcsname} inserted}'' error
% when a control sequence (i.e.,~|\|\meta{name}) or |\|\meta{symbol}) is
% encountered within the optional argument to |\usepackage|.  Hence,
% parameters that typically expect a control sequence in their
% argument---in particular, \optname{bitformatting},
% \optname{boxformatting}, \optname{leftcurly}, and
% \optname{rightcurly}---should best be avoided within the
% |\usepackage|\oarg{options}|{bytefield}| line.
%
% \begin{decl}
%   \optname{bitwidth} = \meta{length} \\
%   \optname{bitheight} = \meta{length}
% \end{decl}
%
% The above parameters represent the width and height of each bit in a
% bit field.  The default value of \optname{bitwidth} is the width of
% ``|{\tiny 99i}|'', i.e.,~the width of a two-digit number plus a small
% amount of extra space.  This enables |\bitheader| to show two-digit
% numbers without overlap.  The default value of \optname{bitheight}
% is~|2ex|, which should allow a normal piece of text to appear within a
% |\bitbox| or |\wordbox| without abutting the box's top or bottom edge.
%
% As a special case, if \optname{bitwidth} is set to the word
% ``|auto|'', it will be set to the width of ``|99i|'' in the current
% bit-number formatting (cf.~\optname{bitformatting} below).  This
% feature provides a convenient way to adjust the bit width after a
% formatting change.
%
% \begin{decl}
%   \optname{endianness} = |little| \emph{or} |big|
% \end{decl}
%
% Specify either little-endian (left-to-right) or big-endian
% (right-to-left) ordering of the bit numbers.  The default is
% little-endian numbering.  Contrast the following two examples.  The
% first formats a bit field in little-endian ordering using an explicit
% |endianness=little|, and the second formats the same bit field in
% big-endian ordering using |endianness=big|.
%
% \begin{verbatim}
%       \begin{bytefield}[endianness=little,bitwidth=0.11111\linewidth]{8}
%         \bitheader{0-7} \\
%         \bitbox{1}{Res} & \bitbox{1}{BE} & \bitbox{1}{CF}
%         & \bitbox{3}{$\mbox{Name\_Len}-1$} & \bitbox{2}{Len\_Len} \\
%       \end{bytefield}
% \end{verbatim}
%
% \begin{bffigure}
%       \begin{bytefield}[endianness=little,bitwidth=0.11111\linewidth]{8}
%         \bitheader{0-7} \\
%         \bitbox{1}{Res} & \bitbox{1}{BE} & \bitbox{1}{CF}
%         & \bitbox{3}{$\mbox{Name\_Len}-1$} & \bitbox{2}{Len\_Len} \\
%       \end{bytefield}
% \end{bffigure}
%
% \begin{verbatim}
%       \begin{bytefield}[endianness=big,bitwidth=0.11111\linewidth]{8}
%         \bitheader{0-7} \\
%         \bitbox{2}{Len\_Len} & \bitbox{3}{$\mbox{Name\_Len}-1$}
%         & \bitbox{1}{CF} & \bitbox{1}{BE} & \bitbox{1}{Res} \\
%       \end{bytefield}
% \end{verbatim}
%
% \begin{bffigure}
%       \begin{bytefield}[endianness=big,bitwidth=0.11111\linewidth]{8}
%         \bitheader{0-7} \\
%         \bitbox{2}{Len\_Len} & \bitbox{3}{$\mbox{Name\_Len}-1$}
%         & \bitbox{1}{CF} & \bitbox{1}{BE} & \bitbox{1}{Res} \\
%       \end{bytefield}
% \end{bffigure}
%
% \begin{decl}
%   \optname{bitformatting} = \meta{command} \emph{or} \marg{commands}
% \end{decl}
%
% The numbers that appear in a bit header are typeset in the
% \optname{bitformatting} style, which defaults to |\tiny|.  To alter
% the style of bit numbers in the bit header, set
% \optname{bitformatting} to a macro that takes a single argument (like
% |\textbf|) or no arguments (like |\small|).  Groups of commands
% (e.g.,~|{\large\itshape}|) are also acceptable.
%
% When \optname{bitformatting} is set, \optname{bitwidth} usually needs
% to be recalculated as well to ensure that a correct amount of spacing
% surrounds each number in the bit header.  As described above, setting
% |bitwidth=auto| is a convenient shortcut for recalculating the
% bit-width in the common case of bit fields containing no more than 99
% bits per line and no particularly wide labels in bit boxes that
% contain only a few bits.
%
% The following example shows how to use \optname{bitformatting} and
% \optname{bitwidth} to format a bit header with small, boldface text:
%
% \begin{verbatim}
%       \begin{bytefield}[bitformatting={\small\bfseries},
%                         bitwidth=auto,
%                         endianness=big]{20}
%         \bitheader{0-19} \\
%         \bitbox{1}{\tiny F/E} & \bitbox{1}{\tiny T0} & \bitbox{1}{\tiny T1}
%         & \bitbox{1}{\tiny Fwd} & \bitbox{16}{Data value} \\
%       \end{bytefield}
% \end{verbatim}
%
% \noindent
% The resulting bit field looks like this:
%
% \begin{bffigure}
%       \begin{bytefield}[bitformatting={\small\bfseries},
%                         bitwidth=auto,
%                         endianness=big]{20}
%         \bitheader{0-19} \\
%         \bitbox{1}{\tiny F/E} & \bitbox{1}{\tiny T0} & \bitbox{1}{\tiny T1}
%         & \bitbox{1}{\tiny Fwd} & \bitbox{16}{Data value} \\
%       \end{bytefield}
% \end{bffigure}
%
% \begin{decl}
%   \optname{boxformatting} = \meta{command} \emph{or} \marg{commands}
% \end{decl}
%
% The text that appears in a |\bitbox| or |\wordbox| is formatted in the
% \optname{boxformatting} style, which defaults to |\centering|.  To
% alter the style of bit numbers in the bit header, set
% \optname{boxformatting} to a macro that takes a single argument (like
% |\textbf| but not |\textbf|---see below) or no arguments (like
% |\small|).  Groups of commands (e.g.,~|{\large\itshape}|) are also
% acceptable.
%
% If \optname{boxformatting} is set to a macro that takes an argument,
% the macro must be defined as a ``long'' macro, which means it can
% accept more than one paragraph as an argument.  Commands defined with
% |\newcommand| are automatically made long, but commands defined with
% |\newcommand*| are not.  \LaTeX's |\text|\dots\ formatting commands
% (e.g.,~|\textbf|) are not long and therefore cannot be used directly
% in \optname{boxformatting}; use the zero-argument versions
% (e.g.,~|\bfseries|) instead.
%
% The following example shows how to use \optname{boxformatting} to
% format the text within each box horizontally centered and italicized:
%
% \begin{verbatim}
%       \begin{bytefield}[boxformatting={\centering\itshape},
%                         bitwidth=1.5em,
%                         endianness=big]{20}
%         \bitheader{0-19} \\
%         \bitbox{1}{\tiny F/E} & \bitbox{1}{\tiny T0} & \bitbox{1}{\tiny T1}
%         & \bitbox{1}{\tiny Fwd} & \bitbox{16}{Data value} \\
%       \end{bytefield}
% \end{verbatim}
%
% \noindent
% The resulting bit field looks like this:
%
% \begin{bffigure}
%       \begin{bytefield}[boxformatting={\centering\itshape},
%                         bitwidth=1.5em,
%                         endianness=big]{20}
%         \bitheader{0-19} \\
%         \bitbox{1}{\tiny F/E} & \bitbox{1}{\tiny T0} & \bitbox{1}{\tiny T1}
%         & \bitbox{1}{\tiny Fwd} & \bitbox{16}{Data value} \\
%       \end{bytefield}
% \end{bffigure}
%
% \begin{decl}
%   \optname{leftcurly} = \meta{delimiter} \\
%   \optname{rightcurly} = \meta{delimiter}
% \end{decl}
%
% Word groups are normally indicated by a curly brace spanning all of
% its rows.  However, the curly brace can be replaced by any other
% extensible math delimiter (i.e.,~a symbol that can meaningfully follow
% |\left| or |\right| in math mode) via a suitable redefinition of
% \optname{leftcurly} or \optname{rightcurly}.  As in math mode, ``|.|''
% means ``no symbol'', as in the following example (courtesy of
% Steven~R. King):
%
% \begin{verbatim}
%       \begin{bytefield}[rightcurly=., rightcurlyspace=0pt]{32}
%         \bitheader[endianness=big]{0,7,8,15,16,23,24,31} \\
%         \begin{rightwordgroup}{0Ch}
%           \bitbox{8}{Byte 15 \\ \tiny (highest address)}
%           & \bitbox{8}{Byte 14}
%           & \bitbox{8}{Byte 13}
%           & \bitbox{8}{Byte 12}
%         \end{rightwordgroup} \\
%         \begin{rightwordgroup}{08h}
%           \bitbox{32}{Long 0}
%         \end{rightwordgroup} \\
%         \begin{rightwordgroup}{04h}
%           \bitbox{16}{Word 1} & \bitbox{16}{Word 0}
%         \end{rightwordgroup} \\
%         \begin{rightwordgroup}{00h}
%           \bitbox{8}{Byte 3}
%           & \bitbox{8}{Byte 2}
%           & \bitbox{8}{Byte 1}
%           & \bitbox{8}{Byte 0 \\ \tiny (lowest address)}
%         \end{rightwordgroup}
%       \end{bytefield}
% \end{verbatim}
%
% \begin{bffigure}
%       \begin{bytefield}[rightcurly=., rightcurlyspace=0pt]{32}
%         \bitheader[endianness=big]{0,7,8,15,16,23,24,31} \\
%         \begin{rightwordgroup}{0Ch}
%           \bitbox{8}{Byte 15 \\ \tiny (highest address)}
%           & \bitbox{8}{Byte 14}
%           & \bitbox{8}{Byte 13}
%           & \bitbox{8}{Byte 12}
%         \end{rightwordgroup} \\
%         \begin{rightwordgroup}{08h}
%           \bitbox{32}{Long 0}
%         \end{rightwordgroup} \\
%         \begin{rightwordgroup}{04h}
%           \bitbox{16}{Word 1} & \bitbox{16}{Word 0}
%         \end{rightwordgroup} \\
%         \begin{rightwordgroup}{00h}
%           \bitbox{8}{Byte 3}
%           & \bitbox{8}{Byte 2}
%           & \bitbox{8}{Byte 1}
%           & \bitbox{8}{Byte 0 \\ \tiny (lowest address)}
%         \end{rightwordgroup}
%       \end{bytefield}
% \end{bffigure}
%
% \begin{decl}
%   \optname{leftcurlyspace} = \meta{length} \\
%   \optname{rightcurlyspace} = \meta{length} \\
%   \optname{curlyspace} = \meta{length}
% \end{decl}
%
% \optname{leftcurlyspace} and \optname{rightcurlyspace} specify the
% space to insert between the bit field and the curly brace in a left or
% right word group (default:~|1ex|).  Setting \optname{curlyspace} is a
% shortcut for setting both \optname{leftcurlyspace} and
% \optname{rightcurlyspace} to the same value.
%
% \begin{decl}
%   |leftlabelspace| = \meta{length} \\
%   |rightlabelspace| = \meta{length} \\
%   |labelspace| = \meta{length}
% \end{decl}
%
% |leftlabelspace| and |rightlabelspace| specify the space to insert
% between the curly brace and the text label in a left or right word
% group (default:~|0.5ex|).  Setting |labelspace| is a shortcut for
% setting both |leftlabelbrace| and |rightlabelspace| to the same value.
%
% \bigskip
%
% Figure~\ref{fig:rightspace} illustrates the juxtaposition of
% \optname{rightcurlyspace} and |rightlabelspace| to a word group and
% its label.  The \optname{leftcurlyspace} and |leftlabelspace|
% parameters are symmetric.
%
% \begin{figure}[htbp]
%   \[
%   \begin{array}{*5{@{}c}@{}}
%     \cline{1-1}
%     \multicolumn{1}{c|}{\cdots\quad\meta{bit-field rows}}
%     &
%     & \left.\rule{0pt}{3ex}\right\}
%     &
%     & \meta{label} \\[2ex] \cline{1-1}
%     & \underbrace{\hspace*{6em}}_{\mathtt{rightcurlyspace}}
%     &
%     & \underbrace{\hspace*{6em}}_{\mathtt{rightlabelspace}} \\
%   \end{array}
%   \]
%   \caption{Role of \texttt{rightcurlyspace} and \texttt{rightlabelspace}}
%   \label{fig:rightspace}
% \end{figure}
%
% \begin{decl}
%   \optname{leftcurlyshrinkage} = \meta{length} \\
%   \optname{rightcurlyshrinkage} = \meta{length} \\
%   \optname{curlyshrinkage} = \meta{length}
% \end{decl}
%
% In \TeX/\LaTeX, the height of a curly brace does not include the tips.
% Hence, in a word group label, the tips of the curly brace will extend
% beyond the height of the word group.
% \optname{leftcurlyshrinkage}\slash \optname{rightcurlyshrinkage} is an
% amount by which to reduce the height of the curly brace in a
% left\slash right word group's label.  Setting \optname{curlyshrinkage}
% is a shortcut for setting both \optname{leftcurlyshrinkage} and
% \optname{rightcurlyshrinkage} to the same value.  Shrinkages default
% to~|5pt|, and it is extremely unlikely that one would ever need to
% change them.  Nevertheless, these parameters are included here in case
% a document is typeset with a math font containing radically different
% curly braces from the ones that come with \TeX/\LaTeX\ or that
% replaces the curly braces (using \optname{leftcurly}\slash
% \optname{rightcurly}, described above) with symbols of substantially
% different heights.
%
% \begin{decl}
%   \optname{lsb} = \meta{integer}
% \end{decl}
%
% Designate the least significant bit (LSB) in the bit header.  By
% default, the LSB is zero, which means that the first bit position in
% the header corresponds to bit~0.  Specifying a different LSB shifts
% the bit header such that the first bit position instead corresponds to
% \meta{integer}.  Note that the \optname{lsb} option affects bit
% \emph{positions} regardless of whether these positions are labeled, as
% demonstrated by the following two examples:
%
% \begin{verbatim}
%       \begin{bytefield}{32}
%         \bitheader[lsb=0]{4,12,20,28} \\
%         \bitbox{16}{ar\$hrd} & \bitbox{16}{ar\$pro} \\
%         \bitbox{8}{ar\$hln} & \bitbox{8}{ar\$pln} & \bitbox{16}{ar\$op} \\
%       \end{bytefield}
% \end{verbatim}
%
% \begin{bffigure}
%       \begin{bytefield}{32}
%         \bitheader[lsb=0]{4,12,20,28} \\
%         \bitbox{16}{ar\$hrd} & \bitbox{16}{ar\$pro} \\
%         \bitbox{8}{ar\$hln} & \bitbox{8}{ar\$pln} & \bitbox{16}{ar\$op} \\
%       \end{bytefield}
% \end{bffigure}
%
% \begin{verbatim}
%       \begin{bytefield}{32}
%         \bitheader[lsb=4]{4,12,20,28} \\
%         \bitbox{16}{ar\$hrd} & \bitbox{16}{ar\$pro} \\
%         \bitbox{8}{ar\$hln} & \bitbox{8}{ar\$pln} & \bitbox{16}{ar\$op} \\
%       \end{bytefield}
% \end{verbatim}
%
% \begin{bffigure}
%       \begin{bytefield}{32}
%         \bitheader[lsb=4]{4,12,20,28} \\
%         \bitbox{16}{ar\$hrd} & \bitbox{16}{ar\$pro} \\
%         \bitbox{8}{ar\$hln} & \bitbox{8}{ar\$pln} & \bitbox{16}{ar\$op} \\
%       \end{bytefield}
% \end{bffigure}
%
%
% \subsection{Common tricks}
% \label{sec:tricks}
%
% This section shows some clever ways to use \pkgname{bytefield}'s commands
% to produce some useful effects.
%
% \paragraph{Odd-sized fields}
% To produce a field that is, say, 1\textonehalf{} words long, use a
% |\bitbox| for the fractional part and specify appropriate values
% for the various \meta{sides} parameters.  For instance:
%
% \begin{verbatim}
%       \begin{bytefield}{16}
%         \bitheader{0,7,8,15} \\
%         \bitbox{8}{8-bit field} & \bitbox[lrt]{8}{} \\
%         \wordbox[lrb]{1}{24-bit field}
%       \end{bytefield}
% \end{verbatim}
%
% \begin{bffigure}
%       \begin{bytefield}{16}
%         \bitheader{0,7,8,15} \\
%         \bitbox{8}{8-bit field} & \bitbox[lrt]{8}{} \\
%         \wordbox[lrb]{1}{24-bit field}
%       \end{bytefield}
% \end{bffigure}
%
% \paragraph{Ellipses}
% To skip words that appear the middle of enumerated data, put some
% |\vdots| in a |\wordbox| with empty \meta{sides}:
%
% \begin{verbatim}
%       \begin{bytefield}{16}
%         \bitbox{8}{Type} & \bitbox{8}{\# of nodes} \\
%         \wordbox{1}{Node~1} \\
%         \wordbox{1}{Node~2} \\
%         \wordbox[]{1}{$\vdots$} \\[1ex]
%         \wordbox{1}{Node~$N$}
%       \end{bytefield}
% \end{verbatim}
%
% \begin{bffigure}
%       \begin{bytefield}{16}
%         \bitbox{8}{Type} & \bitbox{8}{\# of nodes} \\
%         \wordbox{1}{Node~1} \\
%         \wordbox{1}{Node~2} \\
%         \wordbox[]{1}{$\vdots$} \\[1ex]
%         \wordbox{1}{Node~$N$}
%       \end{bytefield}
% \end{bffigure}
%
% \noindent
% The extra |1ex| of vertical space helps vertically center the |\vdots|
% a bit better.
%
% \paragraph{Narrow fields}
% There are a number of options for labeling a narrow field (e.g.,~one
% occupying a single bit):
%
% \newsavebox{\defaultOK}
% \begin{lrbox}{\defaultOK}
%   \begin{bytefield}{8}
%     \bitbox{1}{OK} & \bitbox{7}{Data} \\
%   \end{bytefield}
% \end{lrbox}
%
% \newsavebox{\bitwidthOK}
% \begin{lrbox}{\bitwidthOK}
%   \bytefieldsetup{bitwidth=\widthof{OK~}}\relax
%   \begin{bytefield}{8}
%     \bitbox{1}{OK} & \bitbox{7}{Data} \\
%   \end{bytefield}
% \end{lrbox}
%
% \newsavebox{\tinyOK}
% \begin{lrbox}{\tinyOK}
%   \begin{bytefield}{8}
%     \bitbox{1}{\tiny OK} & \bitbox{7}{Data} \\
%   \end{bytefield}
% \end{lrbox}
%
% \newsavebox{\verticalOK}
% \begin{lrbox}{\verticalOK}
%   \begin{bytefield}{8}
%     \bitbox{1}{\tiny O \\ K} & \bitbox{7}{Data} \\
%   \end{bytefield}
% \end{lrbox}
%
% \newsavebox{\rotateOK}
% \begin{lrbox}{\rotateOK}
%   \begin{bytefield}{8}
%     \bitbox{1}{\rotatebox{90}{\small OK}} & \bitbox{7}{Data} \\
%   \end{bytefield}
% \end{lrbox}
%
% \newsavebox{\resizeOK}
% \begin{lrbox}{\resizeOK}
%   \begin{bytefield}{8}
%     \bitbox{1}{\let\bw=\width\resizebox{\bw}{!}{~OK~}} & \bitbox{7}{Data} \\
%   \end{bytefield}
% \end{lrbox}
%
% \bigskip
%
% \begin{tabular}{ll}
%   \emph{Default}: & \raisebox{-4ex}{\usebox{\defaultOK}} \\
%   |\bytefieldsetup{%|
%   & \multirow{2}*{\raisebox{-4ex}{\usebox{\bitwidthOK}}} \\
%   |  bitwidth=\widthof{OK~}}|: & \\[2ex]
%   |\tiny OK|: & \raisebox{-4ex}{\usebox{\tinyOK}} \\
%   |\tiny O \\ K|: & \raisebox{-4ex}{\usebox{\verticalOK}} \\
%   |\rotatebox{90}{\small OK}|: & \raisebox{-4ex}{\usebox{\rotateOK}} \\
%   |\let\bw=\width| & \multirow{2}*{\raisebox{-4ex}{\usebox{\resizeOK}}} \\
%   |\resizebox{\bw}{!}{~OK~}|: & \\[2ex]
% \end{tabular}
%
% \paragraph{Multi-line bit fields}
% Presentations of wide registers are often easier to read when split
% across multiple lines.  (This capability was originally requested by
% Chris L'Esperance and is currently implemented in \pkgname{bytefield}
% based on code provided by Renaud Pacalet.)  The trick behind the
% typesetting of multi-line bit fields is to pass the \optname{lsb}
% option to |\bitheader| to change the starting bit number used in each
% bit header:
%
% \begin{verbatim}
%       \begin{bytefield}[endianness=big,bitwidth=2em]{16}
%         \bitheader[lsb=16]{16-31} \\
%         \bitbox{1}{\tiny Enable} & \bitbox{7}{Reserved}
%         & \bitbox{8}{Bus} \\[3ex]
%         \bitheader{0-15} \\
%         \bitbox{5}{Device} & \bitbox{3}{Function} & \bitbox{6}{Register}
%         & \bitbox{2}{00}
%       \end{bytefield}
% \end{verbatim}
%
% \begin{bffigure}
%       \begin{bytefield}[endianness=big,bitwidth=2em]{16}
%         \bitheader[lsb=16]{16-31} \\
%         \bitbox{1}{\tiny Enable} & \bitbox{7}{Reserved}
%         & \bitbox{8}{Bus} \\[3ex]
%         \bitheader{0-15} \\
%         \bitbox{5}{Device} & \bitbox{3}{Function} & \bitbox{6}{Register}
%         & \bitbox{2}{00}
%       \end{bytefield}
% \end{bffigure}
%
% Note the use of the optional argument to |\\| to introduce three
% \mbox{x-heights} of additional whitespace between the two rows of
% bits.
%
% \paragraph{Rotated bit labels}
% A problem with using very large bit numbers is that the labels run
% into each other, as in the following example:
%
% \begin{verbatim}
%       \begin{bytefield}[endianness=big]{8}
%         \bitheader[lsb=995]{995-1002} \\
%         \bitbox{4}{A} & \bitbox{4}{B}
%       \end{bytefield}
% \end{verbatim}
%
% \begin{bffigure}
%       \begin{bytefield}[endianness=big]{8}
%         \bitheader[lsb=995]{995-1002} \\
%         \bitbox{4}{A} & \bitbox{4}{B}
%       \end{bytefield}
% \end{bffigure}
%
% One solution is to use the \optname{bitformatting} option and the
% \pkgname{graphicx} package's |\rotatebox| command to rotate each bit
% label by 90\textdegree.  Unfortunately, the naive use of
% \optname{bitformatting} and |\rotatebox| does not typeset nicely:
%
% \begin{verbatim}
%       \begin{bytefield}[endianness=big]{8}
%         \bitheader[lsb=995,
%                    bitformatting={\tiny\rotatebox[origin=B]{90}}]{995-1002} \\
%         \bitbox{4}{A} & \bitbox{4}{B}
%       \end{bytefield}
% \end{verbatim}
%
% \begin{bffigure}
%       \begin{bytefield}[endianness=big]{8}
%         \bitheader[lsb=995,
%                    bitformatting={\tiny\rotatebox[origin=B]{90}}]{995-1002} \\
%         \bitbox{4}{A} & \bitbox{4}{B}
%       \end{bytefield}
% \end{bffigure}
%
% The two problems are that (1)~the numbers are left-justified, and
% (2)~the numbers touch the top margin of the word box.  To address these
% problems we use |\makebox| to construct a right-justified region that
% is sufficiently wide to hold our largest number plus some additional
% space to shift the rotated numbers upwards:
%
% \begin{verbatim}
%       \newlength{\bitlabelwidth}
%       \newcommand{\rotbitheader}[1]{%
%         \tiny
%         \settowidth{\bitlabelwidth}{\quad 9999}%
%         \rotatebox[origin=B]{90}{\makebox[\bitlabelwidth][r]{#1}}%
%       }
%
%       \begin{bytefield}[endianness=big]{8}
%         \bitheader[lsb=995,bitformatting=\rotbitheader]{995-1002} \\
%         \bitbox{4}{A} & \bitbox{4}{B}
%       \end{bytefield}
% \end{verbatim}
%
% \begin{bffigure}
%       \newlength{\bitlabelwidth}
%       \newcommand{\rotbitheader}[1]{^^A
%         \tiny
%         \settowidth{\bitlabelwidth}{\quad 9999}%
%         \rotatebox[origin=B]{90}{\makebox[\bitlabelwidth][r]{#1}}^^A
%       }
%
%       \begin{bytefield}[endianness=big]{8}
%         \bitheader[lsb=995,bitformatting=\rotbitheader]{995-1002} \\
%         \bitbox{4}{A} & \bitbox{4}{B}
%       \end{bytefield}
% \end{bffigure}
%
% \paragraph{Unused bits}
% Because |\width| and |\height| are defined within bit boxes (also word
% boxes), we can represent unused bits by filling a |\bitbox| with a
% rule of size |\width|~$\times$ |\height|:
%
% \begin{verbatim}
%       \begin{bytefield}{32}
%         \bitheader{0,4,8,12,16,20,24,28} \\
%         \bitbox{8}{Tag} & \bitbox{8}{Value} &
%         \bitbox{4}{\rule{\width}{\height}} &
%         \bitbox{12}{Mask} \\
%         \wordbox{1}{Key}
%       \end{bytefield}
% \end{verbatim}
%
% \begin{bffigure}
%       \begin{bytefield}{32}
%         \bitheader{0,4,8,12,16,20,24,28} \\
%         \bitbox{8}{Tag} & \bitbox{8}{Value} &
%         \bitbox{4}{\rule{\width}{\height}} &
%         \bitbox{12}{Mask} \\
%         \wordbox{1}{Key}
%       \end{bytefield}
% \end{bffigure}
%
% \noindent
% The effect is much better when the \pkgname{color} package is used to
% draw the unused bits in color.  (Light gray looks nice.)
%
% \begin{verbatim}
%       \definecolor{lightgray}{gray}{0.8}
%       \begin{bytefield}{32}
%         \bitheader{0,4,8,12,16,20,24,28} \\
%         \bitbox{8}{Tag} & \bitbox{8}{Value} &
%         \bitbox{4}{\color{lightgray}\rule{\width}{\height}} &
%         \bitbox{12}{Mask} \\
%         \wordbox{1}{Key}
%       \end{bytefield}
% \end{verbatim}
%
% \definecolor{lightgray}{gray}{0.8}
% \begin{bffigure}
%       \begin{bytefield}{32}
%         \bitheader{0,4,8,12,16,20,24,28} \\
%         \bitbox{8}{Tag} & \bitbox{8}{Value} &
%         \bitbox{4}{\color{lightgray}\rule{\width}{\height}} &
%         \bitbox{12}{Mask} \\
%         \wordbox{1}{Key}
%       \end{bytefield}
% \end{bffigure}
%
% \paragraph{Aligning text on the baseline}
% Because \pkgname{bytefield} internally uses \LaTeX's |picture|
% environment and that environment's |\makebox| command to draw bit
% boxes and word boxes, the text within a box is centered vertically
% with no attention paid to the text's baseline.  As a result, some
% bit-field labels appear somewhat askew:
%
% \begin{verbatim}
%       \begin{bytefield}[bitwidth=1.5em]{2}
%         \bitbox{1}{M} & \bitbox{1}{y}
%       \end{bytefield}
% \end{verbatim}
%
% \begin{bffigure}
%       \begin{bytefield}[bitwidth=1.5em]{2}
%         \bitbox{1}{M} & \bitbox{1}{y}
%       \end{bytefield}
% \end{bffigure}
%
% A solution is to use the \optname{boxformatting} option to trick
% |\makebox| into thinking that all text has the same height and depth.
% Here we use |\raisebox| to indicate that all text is as tall as a
% ``|W|'' and does not descend at all below the baseline:
%
% \begin{verbatim}
%       \newlength{\maxheight}
%       \setlength{\maxheight}{\heightof{W}}
%
%       \newcommand{\baselinealign}[1]{%
%         \centering
%         \raisebox{0pt}[\maxheight][0pt]{#1}%
%       }
%
%       \begin{bytefield}[boxformatting=\baselinealign,
%                         bitwidth=1.5em]{2}
%         \bitbox{1}{M} & \bitbox{1}{y}
%       \end{bytefield}
% \end{verbatim}
%
% \begin{bffigure}
%       \newlength{\maxheight}
%       \setlength{\maxheight}{\heightof{W}}
%       \newcommand{\baselinealign}[1]{^^A
%         \centering
%         \raisebox{0pt}[\maxheight][0pt]{#1}^^A
%       }
%       \begin{bytefield}[boxformatting=\baselinealign,
%                         bitwidth=1.5em]{2}
%         \bitbox{1}{M} & \bitbox{1}{y}
%       \end{bytefield}
% \end{bffigure}
%
% \paragraph{Register contents}
% Sometimes, rather than listing the \emph{meaning} of each bit field
% within each |\bitbox| or |\wordbox|, it may be desirable to list the
% \emph{contents}, with the meaning described in an additional label
% above each bit number in the bit header.  Although the
% \pkgname{register} package is more suited to this form of layout,
% \pkgname{bytefield} can serve in a pinch with the help of the
% |\turnbox| macro from the \pkgname{rotating} package:
%
% \begin{verbatim}
%      \newcommand{\bitlabel}[2]{%
%        \bitbox[]{#1}{%
%          \raisebox{0pt}[4ex][0pt]{%
%            \turnbox{45}{\fontsize{7}{7}\selectfont#2}%
%          }%
%        }%
%      }
%
%      \begin{bytefield}[bitwidth=1em]{16}
%         \bitlabel{1}{Carry} & \bitlabel{1}{Reserved} &
%         \bitlabel{1}{Parity} & \bitlabel{1}{Reserved} &
%         \bitlabel{1}{Adjust} & \bitlabel{1}{Reserved} &
%         \bitlabel{1}{Zero} & \bitlabel{1}{Sign} &
%         \bitlabel{1}{Trap} & \bitlabel{1}{Interrupt enable} &
%         \bitlabel{1}{Direction} & \bitlabel{1}{Overflow} &
%         \bitlabel{2}{I/O privilege level (12--13)} &
%         \bitlabel{1}{Nested task} & \bitlabel{1}{Reserved} \\
%         \bitheader{0-15} \\
%         \bitbox{1}{0} & \bitbox{1}{1} & \bitbox{1}{0} & \bitbox{1}{0} &
%         \bitbox{1}{0} & \bitbox{1}{0} & \bitbox{1}{0} & \bitbox{1}{1} &
%         \bitbox{1}{0} & \bitbox{1}{1} & \bitbox{1}{0} & \bitbox{1}{0} &
%         \bitbox{1}{0} & \bitbox{1}{0} & \bitbox{1}{0} & \bitbox{1}{0}
%       \end{bytefield}
% \end{verbatim}
%
% \vspace{3\baselineskip}   ^^A  Allow extra room for the rotated headers.
% \begin{bffigure}
%      \newcommand{\bitlabel}[2]{%
%        \bitbox[]{#1}{%
%          \raisebox{0pt}[4ex][0pt]{%
%            \turnbox{45}{\fontsize{7}{7}\selectfont#2}%
%          }%
%        }%
%      }
%      \begin{bytefield}[bitwidth=1em]{16}
%         \bitlabel{1}{Carry} & \bitlabel{1}{Reserved} &
%         \bitlabel{1}{Parity} & \bitlabel{1}{Reserved} &
%         \bitlabel{1}{Adjust} & \bitlabel{1}{Reserved} &
%         \bitlabel{1}{Zero} & \bitlabel{1}{Sign} &
%         \bitlabel{1}{Trap} & \bitlabel{1}{Interrupt enable} &
%         \bitlabel{1}{Direction} & \bitlabel{1}{Overflow} &
%         \bitlabel{2}{I/O privilege level (12--13)} &
%         \bitlabel{1}{Nested task} & \bitlabel{1}{Reserved} \\
%         \bitheader{0-15} \\
%         \bitbox{1}{0} & \bitbox{1}{1} & \bitbox{1}{0} & \bitbox{1}{0} &
%         \bitbox{1}{0} & \bitbox{1}{0} & \bitbox{1}{0} & \bitbox{1}{1} &
%         \bitbox{1}{0} & \bitbox{1}{1} & \bitbox{1}{0} & \bitbox{1}{0} &
%         \bitbox{1}{0} & \bitbox{1}{0} & \bitbox{1}{0} & \bitbox{1}{0}
%       \end{bytefield}
% \end{bffigure}
%
%
% \subsection{Not-so-common tricks}
%
% \paragraph{Colored fields}
% A similar approach to that utilized to indicate unused bits can be
% applied to coloring an individual bit field.  The trick is to use the
% \TeX\ |\rlap| primitive to draw a colored box that overlaps whatever
% follows it to the right:
%
% \begin{verbatim}
%       \newcommand{\colorbitbox}[3]{%
%         \rlap{\bitbox{#2}{\color{#1}\rule{\width}{\height}}}%
%         \bitbox{#2}{#3}}
%       \definecolor{lightcyan}{rgb}{0.84,1,1}
%       \definecolor{lightgreen}{rgb}{0.64,1,0.71}
%       \definecolor{lightred}{rgb}{1,0.7,0.71}
%       \begin{bytefield}[bitheight=\widthof{~Sign~},
%                         boxformatting={\centering\small}]{32}
%         \bitheader[endianness=big]{31,23,0} \\
%         \colorbitbox{lightcyan}{1}{\rotatebox{90}{Sign}} &
%         \colorbitbox{lightgreen}{8}{Exponent} &
%         \colorbitbox{lightred}{23}{Mantissa}
%       \end{bytefield}
% \end{verbatim}
%
% \begin{bffigure}
%       \newcommand{\colorbitbox}[3]{^^A
%         \rlap{\bitbox{#2}{\color{#1}\rule{\width}{\height}}}^^A
%         \bitbox{#2}{#3}}^^A
%       \definecolor{lightcyan}{rgb}{0.84,1,1}
%       \definecolor{lightgreen}{rgb}{0.64,1,0.71}
%       \definecolor{lightred}{rgb}{1,0.7,0.71}
%       \begin{bytefield}[bitheight=\widthof{~Sign~},
%                         boxformatting={\centering\small}]{32}
%         \bitheader[endianness=big]{31,23,0} \\
%         \colorbitbox{lightcyan}{1}{\rotatebox{90}{Sign}} &
%         \colorbitbox{lightgreen}{8}{Exponent} &
%         \colorbitbox{lightred}{23}{Mantissa}
%       \end{bytefield}
% \end{bffigure}
%
% \paragraph{Omitted bit numbers}
% It is occasionally convenient to show a wide bit field in which the
% middle numbers are replaced with an ellipsis.  The trick to
% typesetting such a thing with \pkgname{bytefield} is to point the
% |bitformatting| option to a macro that conditionally modifies the
% given bit number before outputting it.  One catch is that
% \pkgname{bytefield} measures the height of the string ``|1234567890|''
% using the current bit formatting, so that needs to be a valid input.
% (If \optname{bitwidth} is set to ``|auto|'', then ``|99i|'' also has
% to be a valid input, but we're not using ``|auto|'' here.)  The
% following example shows how to \emph{conditionally} modify the bit
% number: If the number is |1234567890|, it is used as is; numbers
% greater than~9 are increased by 48; numbers less than~4 are
% unmodified; the number~6 is replaced by an ellipsis; and all other
% numbers are discarded.
%
% \begin{verbatim}
%       \newcommand{\fakesixtyfourbits}[1]{%
%         \tiny
%         \ifnum#1=1234567890
%           #1
%         \else
%           \ifnum#1>9
%             \count32=#1
%             \advance\count32 by 48
%             \the\count32%
%           \else
%             \ifnum#1<4
%               #1%
%             \else
%               \ifnum#1=6
%                 $\cdots$%
%               \fi
%             \fi
%           \fi
%         \fi
%       }
%       \begin{bytefield}[%
%           bitwidth=\widthof{\tiny Fwd~},
%           bitformatting=\fakesixtyfourbits,
%           endianness=big]{16}
%         \bitheader{0-15} \\
%         \bitbox{1}{\tiny F/E} & \bitbox{1}{\tiny T0} & \bitbox{1}{\tiny T1}
%         & \bitbox{1}{\tiny Fwd} & \bitbox{12}{Data value}
%       \end{bytefield}
% \end{verbatim}
%
% \begin{bffigure}
%       \newcommand{\fakesixtyfourbits}[1]{%
%         \tiny
%         \ifnum#1=1234567890
%           #1
%         \else
%           \ifnum#1>9
%             \count32=#1
%             \advance\count32 by 48
%             \the\count32^^A
%           \else
%             \ifnum#1<4
%               #1%
%             \else
%               \ifnum#1=6
%                 $\cdots$^^A
%               \fi
%             \fi
%           \fi
%         \fi
%       }
%       \begin{bytefield}[^^A
%           bitwidth=\widthof{\tiny Fwd~},
%           bitformatting=\fakesixtyfourbits,
%           endianness=big]{16}
%         \bitheader{0-15} \\
%         \bitbox{1}{\tiny F/E} & \bitbox{1}{\tiny T0} & \bitbox{1}{\tiny T1}
%         & \bitbox{1}{\tiny Fwd} & \bitbox{12}{Data value}
%       \end{bytefield}
% \end{bffigure}
%
% \paragraph{Memory-map diagrams}
% While certainly not the intended purpose of the \pkgname{bytefield}
% package, one can utilize word boxes with empty \meta{sides} and
% word labels to produce memory-map diagrams:
%
% \begin{verbatim}
%       \newcommand{\descbox}[2]{\parbox[c][3.8\baselineskip]{0.95\width}{%
%         \raggedright #1\vfill #2}}
%       \begin{bytefield}[bitheight=4\baselineskip]{32}
%         \begin{rightwordgroup}{Partition 4}
%           \bitbox[]{8}{\texttt{0xFFFFFFFF} \\[2\baselineskip]
%             \texttt{0xC0000000}} &
%           \bitbox{24}{\descbox{1\,GB area for VxDs, memory manager,
%             file system code; shared by all processes.}{Read/writable.}}
%         \end{rightwordgroup} \\
%         \begin{rightwordgroup}{Partition 3}
%           \bitbox[]{8}{\texttt{0xBFFFFFFF} \\[2\baselineskip]
%             \texttt{0x80000000}} &
%           \bitbox{24}{\descbox{1\,GB area for memory-mapped files,
%             shared system \textsc{dll}s, file system code; shared by all
%             processes.}{Read/writable.}}
%         \end{rightwordgroup} \\
%         \begin{rightwordgroup}{Partition 2}
%           \bitbox[]{8}{\texttt{0x7FFFFFFF} \\[2\baselineskip]
%             \texttt{0x00400000}} &
%           \bitbox{24}{\descbox{$\sim$2\,GB area private to process,
%              process code, and data.}{Read/writable.}}
%         \end{rightwordgroup} \\
%         \begin{rightwordgroup}{Partition 1}
%           \bitbox[]{8}{\texttt{0x003FFFFF} \\[2\baselineskip]
%             \texttt{0x00001000}} &
%           \bitbox{24}{\descbox{4\,MB area for MS-DOS and Windows~3.1
%             compatibility.}{Read/writable.}} \\
%           \bitbox[]{8}{\texttt{0x00000FFF} \\[2\baselineskip]
%             \texttt{0x00000000}} &
%           \bitbox{24}{\descbox{4096~byte area for MS-DOS and Windows~3.1
%             compatibility.}{Protected---catches \textsc{null}
%             pointers.}}
%         \end{rightwordgroup}
%       \end{bytefield}
% \end{verbatim}
%
% \begin{bffigure}
%       \newcommand{\descbox}[2]{\parbox[c][3.8\baselineskip]{0.95\width}{%
%         \raggedright #1\vfill #2}}
%       \begin{bytefield}[bitheight=4\baselineskip]{32}
%         \begin{rightwordgroup}{Partition 4}
%           \bitbox[]{8}{\texttt{0xFFFFFFFF} \\[2\baselineskip]
%             \texttt{0xC0000000}} &
%           \bitbox{24}{\descbox{1\,GB area for VxDs, memory manager,
%             file system code; shared by all processes.}{Read/writable.}}
%         \end{rightwordgroup} \\
%         \begin{rightwordgroup}{Partition 3}
%           \bitbox[]{8}{\texttt{0xBFFFFFFF} \\[2\baselineskip]
%             \texttt{0x80000000}} &
%           \bitbox{24}{\descbox{1\,GB area for memory-mapped files,
%             shared system \textsc{dll}s, file system code; shared by all
%             processes.}{Read/writable.}}
%         \end{rightwordgroup} \\
%         \begin{rightwordgroup}{Partition 2}
%           \bitbox[]{8}{\texttt{0x7FFFFFFF} \\[2\baselineskip]
%             \texttt{0x00400000}} &
%           \bitbox{24}{\descbox{$\sim$2\,GB area private to process, process
%             code, and data.}{Read/writable.}}
%         \end{rightwordgroup} \\
%         \begin{rightwordgroup}{Partition 1}
%           \bitbox[]{8}{\texttt{0x003FFFFF} \\[2\baselineskip]
%             \texttt{0x00001000}} &
%           \bitbox{24}{\descbox{4\,MB area for MS-DOS and Windows~3.1
%             compatibility.}{Read/writable.}} \\
%           \bitbox[]{8}{\texttt{0x00000FFF} \\[2\baselineskip]
%             \texttt{0x00000000}} &
%           \bitbox{24}{\descbox{4096~byte area for MS-DOS and Windows~3.1
%             compatibility.}{Protected---catches \textsc{null}
%             pointers.}}
%         \end{rightwordgroup}
%       \end{bytefield}
% \end{bffigure}
%
% The following variation uses variable-height regions in the memory map:
% \changes{v2.1}{2012/09/02}{Included in the documentation a variable-height
%   memory-map example suggested by Martin Demling}
%
% \begin{verbatim}
%       \newcommand{\descbox}[2]{\parbox[c][3.8\baselineskip]{0.95\width}{%
%       % facilitates the creation of memory maps. Start address at the bottom,
%       % end address at the top.
%       % syntax:
%       %   \memsection{end address}{start address}{height in lines}{text in box}
%       \newcommand{\memsection}[4]{%
%         % define the height of the memsection
%         \bytefieldsetup{bitheight=#3\baselineskip}%
%         \bitbox[]{10}{%
%           \texttt{#1}%      print end address
%           \\
%           %   do some spacing
%           \vspace{#3\baselineskip}
%           \vspace{-2\baselineskip}
%           \vspace{-#3pt}
%           \texttt{#2}%      print start address
%         }%
%         \bitbox{16}{#4}%    print box with caption
%       }
%
%       \begin{bytefield}{24}
%               \memsection{ffff ffff}{0040 0000}{15}{-- free --}\\
%               \begin{rightwordgroup}{internal memory}
%                       \memsection{003f ffff}{002f c000}{4}{Special Function
%                         Registers}\\
%                       \memsection{002f bfff}{0007 0000}{3}{-- reserved --}\\
%                       \memsection{0006 ffff}{0000 0000}{8}{Internal Flash}
%               \end{rightwordgroup}\\
%       \end{bytefield}
% \end{verbatim}
%
% \begin{bffigure}
%       \newcommand{\memsection}[4]{^^A
%       \bytefieldsetup{bitheight=#3\baselineskip}^^A
%       \bitbox[]{10}{^^A
%       \texttt{#1}^^A
%       \\ \vspace{#3\baselineskip}\vspace{-2\baselineskip}\vspace{-#3pt}^^A
%       \texttt{#2}^^A
%       }^^A
%       \bitbox{16}{#4}^^A
%       }
%       \begin{bytefield}{24}
%               \memsection{ffff ffff}{0040 0000}{15}{-- free --}\\
%               \begin{rightwordgroup}{internal memory}
%                       \memsection{003f ffff}{002f c000}{4}{Special Function Registers}\\
%                       \memsection{002f bfff}{0007 0000}{3}{-- reserved --}\\
%                       \memsection{0006 ffff}{0000 0000}{8}{Internal Flash}
%               \end{rightwordgroup}\\
%       \end{bytefield}
% \end{bffigure}
%
%
% \subsection{Putting it all together}
%
% The following code showcases most of \pkgname{bytefield}'s features in a
% single figure.
%
% \begin{verbatim}
%       \begin{bytefield}[bitheight=2.5\baselineskip]{32}
%         \bitheader{0,7,8,15,16,23,24,31} \\
%         \begin{rightwordgroup}{\parbox{6em}{\raggedright These words were taken
%           verbatim from the TCP header definition (RFC~793).}}
%           \bitbox{4}{Data offset} & \bitbox{6}{Reserved} &
%             \bitbox{1}{\tiny U\\R\\G} & \bitbox{1}{\tiny A\\C\\K} &
%             \bitbox{1}{\tiny P\\S\\H} & \bitbox{1}{\tiny R\\S\\T} &
%             \bitbox{1}{\tiny S\\Y\\N} & \bitbox{1}{\tiny F\\I\\N} &
%             \bitbox{16}{Window} \\
%           \bitbox{16}{Checksum} & \bitbox{16}{Urgent pointer}
%         \end{rightwordgroup} \\
%         \wordbox[lrt]{1}{Data octets} \\
%         \skippedwords \\
%         \wordbox[lrb]{1}{} \\
%         \begin{leftwordgroup}{\parbox{6em}{\raggedright Note that we can display,
%           for example, a misaligned 64-bit value with clever use of the
%           optional argument to \texttt{\string\wordbox} and
%           \texttt{\string\bitbox}.}}
%           \bitbox{8}{Source} & \bitbox{8}{Destination} &
%             \bitbox[lrt]{16}{} \\
%           \wordbox[lr]{1}{Timestamp} \\
%           \begin{rightwordgroup}{\parbox{6em}{\raggedright Why two Length fields?
%             No particular reason.}}
%             \bitbox[lrb]{16}{} & \bitbox{16}{Length}
%         \end{leftwordgroup} \\
%             \bitbox{6}{Key} & \bitbox{6}{Value} & \bitbox{4}{Unused} &
%               \bitbox{16}{Length}
%           \end{rightwordgroup} \\
%         \wordbox{1}{Total number of 16-bit data words that follow this
%           header word, excluding the subsequent checksum-type value} \\
%         \bitbox{16}{Data~1} & \bitbox{16}{Data~2} \\
%         \bitbox{16}{Data~3} & \bitbox{16}{Data~4} \\
%         \bitbox[]{16}{$\vdots$ \\[1ex]} &
%           \bitbox[]{16}{$\vdots$ \\[1ex]} \\
%         \bitbox{16}{Data~$N-1$} & \bitbox{16}{Data~$N$} \\
%         \bitbox{20}{\[ \mbox{A5A5}_{\mbox{\scriptsize H}} \oplus
%           \left(\sum_{i=1}^N \mbox{Data}_i \right) \bmod 2^{20} \]} &
%           \bitbox{12}{Command} \\
%         \wordbox{2}{64-bit random number}
%       \end{bytefield}
% \end{verbatim}
%
% \noindent
% Figure~\ref{fig:complex-diagram} shows the resulting protocol diagram.
%
% \begin{figure}[phtb]
%    \centering
%    \begin{lrbox}{\protocoldiagram}
%       \begin{bytefield}[bitheight=2.5\baselineskip]{32}
%         \bitheader{0,7,8,15,16,23,24,31} \\
%         \begin{rightwordgroup}{\parbox{6em}{\raggedright These words were taken
%           verbatim from the TCP header definition (RFC~793).}}
%           \bitbox{4}{Data offset} & \bitbox{6}{Reserved} &
%             \bitbox{1}{\tiny U\\R\\G} & \bitbox{1}{\tiny A\\C\\K} &
%             \bitbox{1}{\tiny P\\S\\H} & \bitbox{1}{\tiny R\\S\\T} &
%             \bitbox{1}{\tiny S\\Y\\N} & \bitbox{1}{\tiny F\\I\\N} &
%             \bitbox{16}{Window} \\
%           \bitbox{16}{Checksum} & \bitbox{16}{Urgent pointer}
%         \end{rightwordgroup} \\
%         \wordbox[lrt]{1}{Data octets} \\
%         \skippedwords \\
%         \wordbox[lrb]{1}{} \\
%         \begin{leftwordgroup}{\parbox{6em}{\raggedright Note that we can display,
%           for example, a misaligned 64-bit value with clever use of the
%           optional argument to \texttt{\string\wordbox} and
%           \texttt{\string\bitbox}.}}
%           \bitbox{8}{Source} & \bitbox{8}{Destination} & \bitbox[lrt]{16}{} \\
%           \wordbox[lr]{1}{Timestamp} \\
%           \begin{rightwordgroup}{\parbox{6em}{\raggedright Why two Length fields?
%             No particular reason.}}
%             \bitbox[lrb]{16}{} & \bitbox{16}{Length}
%         \end{leftwordgroup} \\
%             \bitbox{6}{Key} & \bitbox{6}{Value} & \bitbox{4}{Unused} &
%               \bitbox{16}{Length}
%           \end{rightwordgroup} \\
%         \wordbox{1}{Total number of 16-bit data words that follow this
%           header word, excluding the subsequent checksum-type value} \\
%         \bitbox{16}{Data~1} & \bitbox{16}{Data~2} \\
%         \bitbox{16}{Data~3} & \bitbox{16}{Data~4} \\
%         \bitbox[]{16}{$\vdots$ \\[1ex]} & \bitbox[]{16}{$\vdots$ \\[1ex]} \\
%         \bitbox{16}{Data~$N-1$} & \bitbox{16}{Data~$N$} \\
%         \bitbox{20}{\[ \mbox{A5A5}_{\mbox{\scriptsize H}} \oplus
%           \left(\sum_{i=1}^N \mbox{Data}_i \right) \bmod 2^{20} \]} &
%           \bitbox{12}{Command} \\
%         \wordbox{2}{64-bit random number}
%       \end{bytefield}
%     \end{lrbox}
%     \makebox[0pt][c]{\usebox{\protocoldiagram}}
%   \caption{Complex protocol diagram drawn with the \pkgname{bytefield}
%     package}
%   \label{fig:complex-diagram}
% \end{figure}
%
%
% \subsection{Upgrading from older versions}
% \label{sec:upgrading}
%
% \pkgname{bytefield}'s user interface changed substantially with the
% introduction of version~2.0.  Because documents written for
% \pkgname{bytefield}~v1.\textit{x} will not build properly under later
% versions of the package, this section explains how to convert documents
% to the new interface.
%
% \begin{decl}
%   \SpecialUsageIndex{\wordgroupr}
%   |\wordgroupr| \\
%   \SpecialUsageIndex{\endwordgroupr}
%   |\endwordgroupr|
% \end{decl}
%
% These have been replaced with the |rightwordgroup| environment to make
% their invocation more \LaTeX-like.  Use |\begin{rightwordgroup}|
% instead of |\wordgroupr| and |\end{rightwordgroup}| instead of
% |\endwordgroupr|.
%
% \begin{decl}
%   \SpecialUsageIndex{\wordgroupl}
%   |\wordgroupl| \\
%   \SpecialUsageIndex{\endwordgroupl}
%   |\endwordgroupl|
% \end{decl}
%
% These have been replaced with the |leftwordgroup| environment to make
% their invocation more \LaTeX-like.  Use |\begin{leftwordgroup}|
% instead of |\wordgroupl| and |\end{leftwordgroup}| instead of
% |\endwordgroupl|.
%
% \begin{decl}
%   \SpecialUsageIndex{\bitwidth}
%   |\bitwidth|
% \end{decl}
%
% Instead of changing bit widths with
% |\setlength{\bitwidth}{|\meta{width}|}|, use
% |\bytefieldsetup{bitwidth=|\meta{width}|}|.
%
% \begin{decl}
%   \SpecialUsageIndex{\byteheight}
%   |\byteheight|
% \end{decl}
%
% Instead of changing bit heights with
% |\setlength{\byteheight}{|\meta{height}|}|, use
% |\bytefieldsetup{bitheight=|\meta{height}|}| (and note the change from
% ``|byte|'' to ``|bit|'' for consistency with \optname{bitwidth}).
%
% \begin{decl}
%   \SpecialUsageIndex{\curlyspace}
%   |\curlyspace| \\
%   \SpecialUsageIndex{\labelspace}
%   |\labelspace|
% \end{decl}
%
% Instead of using |\setlength{\curlyspace}{|\meta{dist}|}| and
% |\setlength{\labelspace}{|\meta{dist}|}| to alter the horizontal space
% that appears before and after a curly brace, use
% |\bytefieldsetup{curlyspace=|\meta{dist}|}| and
% |\bytefieldsetup{labelspace=|\meta{dist}|}|.  Note that, as described
% in Section~\ref{sec:basic-cmds}, left and right spacing can be set
% independently if desired.
%
% \begin{decl}
%   \SpecialUsageIndex{\curlyshrinkage}
%   |\curlyshrinkage|
% \end{decl}
%
% Instead of using |\setlength{\curlyshrinkage}{|\meta{dist}|}| to
% reduce the vertical space occupied by a curly brace, use
% |\bytefieldsetup{curlyshrinkage=|\meta{dist}|}|.  Note that, as
% described in Section~\ref{sec:basic-cmds}, left and right curly-brace
% height can be reduced independently if desired.
%
% \begin{decl}
%   \SpecialUsageIndex{\bitwidth}
%   |\bitwidth| \oarg{endianness} \marg{bit-positions}
% \end{decl}
%
% The meaning of |\bitwidth|'s optional argument changed with
% \pkgname{bytefield}~v2.1.  In older versions of the package, the
% optional argument was one of ``|l|'' or ``|b|'' for, respectively,
% little-endian or big-endian bit ordering.  Starting with version~2.1,
% the optional argument can be any of the parameters described in
% Section~\ref{sec:options} (but practically only |bitformatting|,
% |endianness|, and |lsb|).  Hence, ``|l|'' should be replaced with
% |endianness=little| and ``|b|'' should be replaced with
% |endianness=big|.  Although more verbose, these new options can be
% specified once for the entire document by listing them as package
% options or as arguments to |\bytefieldsetup|.
%
% \bigskip
%
% As a crutch to help build older documents with minimal modification,
% \pkgname{bytefield} provides a |compat1| package option that restores
% the old interface.  This option, invoked with
% |\usepackage[compat1]{bytefield}|, may disappear in a future version
% of the package and should therefore not be relied upon as a long-term
% approach to using \pkgname{bytefield}.
%
%
% \StopEventually{^^A
%
%   \section{Future work}
%
%   \pkgname{bytefield} is my first \LaTeX\ package, and, as such, there
%   are a number of macros that could probably have been implemented a
%   lot better.  For example, \pkgname{bytefield} is somewhat wasteful
%   of \meta{dimen} registers (although it did get a lot better with
%   version~1.1 and again with version~1.3).  The package should really
%   get a major overhaul now that I've gotten better at \TeX\slash
%   \LaTeX\ programming.  One minor improvement I'd like to make in the
%   package is to move left, small curly braces closer to the bit field.
%   In the following figure, notice how distant the small curly appears
%   from the bit-field body:
%
%   \begin{bffigure}
%         \bytefieldsetup{bitheight=4ex}
%         \begin{bytefield}{16}
%           \begin{leftwordgroup}{Too distant}
%             \wordbox{1}{Something}
%           \end{leftwordgroup} \\
%           \begin{leftwordgroup}{Looks okay}
%             \wordbox{4}{Something else}
%           \end{leftwordgroup}
%         \end{bytefield}
%   \end{bffigure}
%
%   \noindent
%   The problem is that the curly braces are left-aligned relative to
%   each other, while they should be right-aligned.
% }
%
%
% \section{Implementation}
%
% \newcommand{\usermacro}{\marginpar{\vspace*{1ex}{\strut\hspace{18pt}\huge$\star$}\vspace*{-1ex}}}
%
% This section contains the complete source code for \pkgname{bytefield}.
% Most users will not get much out of it, but it should be of use to
% those who need more precise documentation and those who want to extend
% (or debug~\smiley) the \pkgname{bytefield} package.
%
% In this section, macros marked in the margin with a ``{\Large
%   $\star$}'' are intended to be called by the user (and were described
% in Section~\ref{sec:usage}).  All other macros are used only
% internally by \pkgname{bytefield}.
%
% \subsection{Required packages}
%
% Although |\widthof| and |\heightof| were introduced in June~1998,
% te\kern-1.5pt\TeX~2.0---still in widespread use at the time of this
% writing (2005)---ships with an earlier |calc.sty| in the |source|
% directory.  Because a misconfigured system may find the |source|
% version of |calc.sty| we explicitly specify a later date when loading
% the |calc| package.
% \changes{v1.2a}{2005/07/31}{Specified an explicit package date when
%   loading the \pkgname{calc} package to avoid loading an outdated
%   version.  Thanks to Kevin Quick for discovering that outdated versions
%   of \pkgname{calc} are still being included in \TeX{} distributions.}
%    \begin{macrocode}
\RequirePackage{calc}[1998/07/07]
\RequirePackage{keyval}
%    \end{macrocode}
%
%
% \subsection{Utility macros}
%
% The following macros in this section are used by the box-drawing macros
% and the ``skipped words''-drawing macros.
%
% \begin{macro}{\bf@newdimen}
% \begin{macro}{\allocationnumber}
% \changes{v1.1}{2002/09/15}{Bug fix: Added \cs{bf@newdimen} to greatly
%   reduce the likelihood of ``\texttt{No room for a new \cs{dimen}}''
%   errors (reported by Vitaly A. Repin)}
% |\newdimen| defines new \meta{dimen}s globally.  |\bf@newdimen|
% defines them locally.  It simply merges \LaTeXe's |\newdimen| and
% |\alloc@| macros while omitting |\alloc@|'s ``|\global|'' declaration.
%    \begin{macrocode}
\def\bf@newdimen#1{\advance\count11 by 1
  \ch@ck1\insc@unt\dimen
  \allocationnumber=\count11
  \dimendef#1=\allocationnumber
  \wlog{\string#1=\string\dimen\the\allocationnumber\space (locally)}%
}
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
% \begin{macro}{\bf@newdimen}
% If we're using $\varepsilon$-\TeX\ then we have access to many more
% \meta{dimen}s than in original \TeX\@.  The \textsf{etex} package
% makes these extra \meta{dimen}s available via the |\locdimen| macro.
% At the start of the document we check if |\locdimen| is defined and,
% if so, replace the preceding definition of |\bf@newdimen| with
% \textsf{etex}'s |\locdimen|.
% \changes{v1.3}{2010/10/31}{Added support for $\varepsilon$-\TeX's larger
%   local \meta{dimen} pool (code provided by Heiko Oberdiek)}
%    \begin{macrocode}
\AtBeginDocument{%
  \begingroup\expandafter\expandafter\expandafter\endgroup
  \expandafter\ifx\csname locdimen\endcsname\relax
  \else
    \let\bf@newdimen\locdimen
  \fi
}
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\bytefield@height}
% \begin{macro}{\ifcounting@words}
% When |\ifcounting@words| is \textsc{True}, add the height of the next
% \texttt{picture} environment to |\bytefield@height|.  We set
% |\counting@wordstrue| at the beginning of each word, and
% |\counting@wordsfalse| after each |\bitbox|, |\wordbox|, or |\skippedwords|
% picture.
%    \begin{macrocode}
\newlength{\bytefield@height}
\newif\ifcounting@words
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
% \begin{macro}{\inc@bytefield@height}
% We have to define a special macro to increment |\bytefield@height|
% because the \pkgname{calc} package's |\addtolength| macro doesn't seem
% to see the global value.  So we |\setlength| a temporary (to get
% \pkgname{calc}'s nice infix features) and |\advance| |\bytefield@height|
% by that amount.
%    \begin{macrocode}
\newlength{\bytefield@height@increment}
\DeclareRobustCommand{\inc@bytefield@height}[1]{%
  \setlength{\bytefield@height@increment}{#1}%
  \global\advance\bytefield@height by \bytefield@height@increment}
%    \end{macrocode}
% \end{macro}
%
% \subsection{Top-level environment}
%
% \begin{macro}{\entire@bytefield@picture}
% Declare a box for containing the entire bytefield.  By storing
% everything in a box and then typesetting it later (at the
% |\end{bytefield}|), we can center the bit field, put a box around it,
% and do other operations on the entire figure.
%    \begin{macrocode}
\newsavebox{\entire@bytefield@picture}
%    \end{macrocode}
% \end{macro}
%
% \begin{environment}{bytefield}
% \usermacro
% \begin{macro}{\bits@wide}
% \begin{macro}{\old@nl}
% \begin{macro}{\amp}
% The |bytefield| environment contains the layout of bits in a sequence
% of words.  This is the main environment defined by the
% \pkgname{bytefield} package.  The argument is the number of bits wide
% the bytefield should be.  We turn |&| into a space character so the
% user can think of a \pkgname{bytefield} as being analogous to a
% \texttt{tabular} environment, even though we're really setting the
% bulk of the picture in a single column.  (Row labels go in separate
% columns, however.)
%    \begin{macrocode}
\newenvironment{bytefield}[2][]{%
  \bf@bytefieldsetup{#1}%
  \def\bits@wide{#2}%
  \let\old@nl=\\%
  \let\amp=&%
  \catcode`\&=10
  \openup -1pt
  \setlength{\bytefield@height}{0pt}%
  \setlength{\unitlength}{1pt}%
  \global\counting@wordstrue
  \begin{lrbox}{\entire@bytefield@picture}%
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
% \begin{macro}{\\}
% \usermacro
% We redefine |\\| within the |bytefield| environment to make it aware
% of curly braces that surround the protocol diagram.
% \changes{v2.1}{2011/06/12}{Augmented the definition
%   of~\texttt{\string\string\string\\} to accept an optional argument,
%   just like in a \texttt{tabular} environment}
%    \begin{macrocode}
  \renewcommand{\\}[1][0pt]{%
    \unskip
    \vspace{##1}%
    \amp\show@wordlabelr\cr
    \ignorespaces\global\counting@wordstrue\make@lspace\amp}%
%    \end{macrocode}
% \end{macro}
%    \begin{macrocode}
  \vbox\bgroup\ialign\bgroup##\amp##\amp##\cr\amp
}{%
  \amp\show@wordlabelr\cr\egroup\egroup
  \end{lrbox}%
  \usebox{\entire@bytefield@picture}}
%    \end{macrocode}
% \end{environment}
%
% \subsection{Box-drawing macros}
%
% \subsubsection{Drawing (proper)}
%
% \begin{macro}{\bf@bitformatting}
% Format a bit number in the bit header.  |\bf@bitformatting| may be
% redefined to take either a single argument (\`a~la |\textbf|) or no
% argument (\`a~la |\small|).
% \changes{v1.4}{2011/01/15}{Introduced this macro at Steven R. King's
%   request to enable users to alter the bit header's font size}
%    \begin{macrocode}
\newcommand*{\bf@bitformatting}{\tiny}
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\bf@boxformatting}
% Format the text within a bit box or word box.  |\bf@boxformatting|
% takes either a single argument (\`a~la |\textbf|) or no argument
% (\`a~la |\small|).  The text that follows |\bf@boxformatting| is
% guaranteed to be a group that ends in |\par|, so if
% |\bf@boxformatting| accepts an argument, the macro should be defined
% with |\long| (e.g.,~with |\newcommand| but not with |\newcommand*|).
%    \begin{macrocode}
\newcommand*{\bf@boxformatting}{\centering}
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\bf@bitwidth}
% Define the width of a single bit.  Note that this is wide enough to
% display a two-digit number without it running into adjacent numbers.
% For larger words, be sure to |\setlength| this larger.
%    \begin{macrocode}
\newlength{\bf@bitwidth}
\settowidth{\bf@bitwidth}{\bf@bitformatting{99i}}
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\bf@bitheight}
% This is the height of a single bit within the bit field.
%    \begin{macrocode}
\newlength{\bf@bitheight}
\setlength{\bf@bitheight}{4ex}
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\units@wide}
% \begin{macro}{\units@tall}
% These are scratch variables for storing the width and height (in
% points) of the box we're about to draw.
%    \begin{macrocode}
\newlength{\units@wide}
\newlength{\units@tall}
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
% \begin{macro}{\bitbox}
% \usermacro
% Put some text (|#3|) in a box that's a given number of bits (|#2|)
% wide and one byte tall.
% An optional argument (|#1|) specifies which lines to draw---|[l]|eft,
% |[r]|ight, |[t]|op, and/or |[b]|ottom (default: |lrtb|).
%    \begin{macrocode}
\DeclareRobustCommand{\bitbox}[3][lrtb]{%
  \setlength{\units@wide}{\bf@bitwidth * #2}%
  \bf@parse@bitbox@arg{#1}%
  \draw@bit@picture{\strip@pt\units@wide}{\strip@pt\bf@bitheight}{#3}}
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\wordbox}
% \usermacro
% Put some text (|#3|) in a box that's a given number of bytes (|#2|)
% tall and one word (|\bits@wide| bits) wide.
% An optional argument (|#1|) specifies which lines to draw---|[l]|eft,
% |[r]|ight, |[t]|op, and/or |[b]|ottom (default: |lrtb|).
%    \begin{macrocode}
\DeclareRobustCommand{\wordbox}[3][lrtb]{%
  \setlength{\units@wide}{\bf@bitwidth * \bits@wide}%
  \setlength{\units@tall}{\bf@bitheight * #2}%
  \bf@parse@bitbox@arg{#1}%
  \draw@bit@picture{\strip@pt\units@wide}{\strip@pt\units@tall}{#3}}
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\draw@bit@picture}
% Put some text (|#3|) in a box that's a given number of units (|#1|)
% wide and a given number of units (|#2|) tall.
% We format the text with a |\parbox| to enable word-wrapping and explicit
% line breaks.  In addition, we define |\height|, |\depth|, |\totalheight|,
% and |\width| (\`a la |\makebox| and friends), so the user can utilize
% those for special effects (e.g., a |\rule| that fills the entire box).
% As an added bonus, we define |\widthunits| and |\heightunits|, which
% are the width and height of the box in multiples of |\unitlength|
% (i.e., |#1| and |#2|, respectively).
%    \begin{macrocode}
\DeclareRobustCommand{\draw@bit@picture}[3]{%
  \begin{picture}(#1,#2)%
%    \end{macrocode}
% \begin{macro}{\height}
% \usermacro
% \begin{macro}{\depth}
% \usermacro
% \begin{macro}{\totalheight}
% \usermacro
% \begin{macro}{\width}
% \usermacro
% \begin{macro}{\widthunits}
% \usermacro
% \begin{macro}{\heightunits}
% \usermacro
%   First, we plot the user's text, with all sorts of useful lengths
%   predefined.
%    \begin{macrocode}
    \put(0,0){\makebox(#1,#2){\parbox{#1\unitlength}{%
      \bf@newdimen\height
      \bf@newdimen\depth
      \bf@newdimen\totalheight
      \bf@newdimen\width
      \height=#2\unitlength
      \depth=0pt%
      \totalheight=#2\unitlength
      \width=#1\unitlength
      \def\widthunits{#1}%
      \def\heightunits{#2}%
      \bf@boxformatting{#3\par}}}}%
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
%   Next, we draw each line individually.  I suppose we could make a special
%   case for ``all lines'' and use a |\framebox| above, but the following
%   works just fine.
%    \begin{macrocode}
    \ifbitbox@top
      \put(0,#2){\line(1,0){#1}}
    \fi
    \ifbitbox@bottom
      \put(0,0){\line(1,0){#1}}
    \fi
    \ifbitbox@left
      \put(0,0){\line(0,1){#2}}
    \fi
    \ifbitbox@right
      \put(#1,0){\line(0,1){#2}}
    \fi
  \end{picture}%
%    \end{macrocode}
%   Finally, we indicate that we're no longer at the beginning of a word.
%   The following code structure (albeit with different arguments to
%   |\inc@bytefield@height|) is repeated in various places throughout this
%   package.  We document it only here, however.
%    \begin{macrocode}
  \ifcounting@words
    \inc@bytefield@height{\unitlength * \real{#2}}%
    \global\counting@wordsfalse
  \fi
  \ignorespaces}
%    \end{macrocode}
% \end{macro}
%
% \subsubsection{Parsing arguments}
%
% The macros in this section are used to parse the optional argument
% to |\bitbox| or |\wordbox|, which is some subset of
% $\{ |l|, |r|, |t|, |b| \}$.
%
% \begin{macro}{\ifbitbox@top}
% \begin{macro}{\ifbitbox@bottom}
% \begin{macro}{\ifbitbox@left}
% \begin{macro}{\ifbitbox@right}
% These macros are set to \textsc{True} if we're to draw the
% corresponding edge on the subsequent |\bitbox| or |\wordbox|.
%    \begin{macrocode}
\newif\ifbitbox@top
\newif\ifbitbox@bottom
\newif\ifbitbox@left
\newif\ifbitbox@right
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \begin{macro}{\bf@parse@bitbox@arg}
% This main parsing macro merely resets the above conditionals and
% calls a helper function, |\bf@parse@bitbox@sides|.
%    \begin{macrocode}
\def\bf@parse@bitbox@arg#1{%
  \bitbox@topfalse
  \bitbox@bottomfalse
  \bitbox@leftfalse
  \bitbox@rightfalse
  \bf@parse@bitbox@sides#1X}
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\bf@parse@bitbox@sides}
% The helper function for |\bf@parse@bitbox@arg| parses a single letter,
% sets the appropriate conditional to \textsc{True}, and calls itself
% tail-recursively until it sees an ``|X|''.
%    \begin{macrocode}
\def\bf@parse@bitbox@sides#1{%
  \ifx#1X%
  \else
    \ifx#1t%
      \bitbox@toptrue
    \else
      \ifx#1b%
        \bitbox@bottomtrue
      \else
        \ifx#1l%
          \bitbox@lefttrue
        \else
          \ifx#1r%
            \bitbox@righttrue
          \fi
        \fi
      \fi
    \fi
    \expandafter\bf@parse@bitbox@sides
  \fi}
%    \end{macrocode}
% \end{macro}
%
% \subsection{Skipped words}
%
% \begin{macro}{\units@high}
% This is the height of each diagonal line in the |\skippedwords|
% graphic.  Note that |\units@high|~$=$ |\units@tall|~$-$
% \textit{optional argument to} |\skippedwords|.
%    \begin{macrocode}
\newlength{\units@high}
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\skippedwords}
% \usermacro
% Output a fancy graphic representing skipped words.  The optional argument
% is the vertical space between the two diagonal lines (default: |2ex|).
%    \begin{macrocode}
\DeclareRobustCommand{\skippedwords}[1][2ex]{%
  \setlength{\units@wide}{\bf@bitwidth * \bits@wide}%
  \setlength{\units@high}{1pt * \ratio{\units@wide}{6.0pt}}%
  \setlength{\units@tall}{#1 + \units@high}%
  \edef\num@wide{\strip@pt\units@wide}%
  \edef\num@tall{\strip@pt\units@tall}%
  \edef\num@high{\strip@pt\units@high}%
  \begin{picture}(\num@wide,\num@tall)
    \put(0,\num@tall){\line(6,-1){\num@wide}}
    \put(\num@wide,0){\line(-6,1){\num@wide}}
    \put(0,0){\line(0,1){\num@high}}
    \put(\num@wide,\num@tall){\line(0,-1){\num@high}}
  \end{picture}%
  \ifcounting@words
    \inc@bytefield@height{\unitlength * \real{\num@tall}}%
    \global\counting@wordsfalse
  \fi}
%    \end{macrocode}
% \end{macro}
%
% \subsection{Bit-position labels}
%
% \begin{macro}{\bf@bit@endianness}
% \pkgname{bytefield} can label bit headers in either little-endian ($0,
% 1, 2,~\ldots, N-1$) or big-endian ($N-1, N-2, N-3,~\ldots, 0$) fashion.
% The |\bf@bit@endianness| macro specifies which to use, either ``|l|''
% for little-endian (the default) or ``|b|'' for big-endian.
%    \begin{macrocode}
\newcommand*{\bf@bit@endianness}{l}
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\bf@first@bit}
% Normally, bits are numbered starting from zero.  However,
% |\bf@first@bit| can be altered (usually locally) to begin numbering
% from a different value.
%    \begin{macrocode}
\newcommand*{\bf@first@bit}{0}
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\bitheader}
% \usermacro
% Output a header of numbered bit positions. The optional argument (|#1|)
% is ``|l|'' for little-endian (default) or ``|b|'' for big-endian.
% The required argument (|#2|) is a list of bit positions to label.
% It is composed of comma-separated ranges of numbers, for example,
% ``|0-31|'', ``|0,7-8,15-16,23-24,31|'', or even something odd like
% ``|0-7,15-23|''.  Ranges must be specified in increasing order; use the
% \optname{lsb} option to reverse the labels' direction.
% \changes{v2.1}{2011/06/17}{Changed the optional argument to accept
%   \meta{key}!=\meta{value} pairs instead of just ``\texttt{l}'' and
%  ``\texttt{b}''}
%    \begin{macrocode}
\DeclareRobustCommand{\bitheader}[2][]{%
  \bf@parse@bitbox@arg{lrtb}%
  \setlength{\units@wide}{\bf@bitwidth * \bits@wide}%
  \setlength{\units@tall}{\heightof{\bf@bitformatting{1234567890}}}%
  \setlength{\units@high}{\units@tall * -1}%
  \bf@process@bitheader@opts{#1}%
  \begin{picture}(\strip@pt\units@wide,\strip@pt\units@tall)%
                 (0,\strip@pt\units@high)
    \bf@parse@range@list#2,X,
  \end{picture}%
  \ifcounting@words
    \inc@bytefield@height{\unitlength * \real{\strip@pt\units@tall}}%
    \global\counting@wordsfalse
  \fi
  \ignorespaces}
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\bf@parse@range@list}
% This is helper function~\#1 for |\bitheader|.  It parses a
% comma-separated list of ranges, calling |\bf@parse@range| on each
% range.
% \changes{v1.1}{2002/06/24}{Bug fix: Swapped order of arguments to
%   \texttt{\string\bslash\space ifx} test (suggested by Hans-Joachim
%   Widmaier)}
%    \begin{macrocode}
\def\bf@parse@range@list#1,{%
  \ifx X#1
  \else
    \bf@parse@range#1-#1-#1\relax
    \expandafter\bf@parse@range@list
  \fi}
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\header@xpos}
% \begin{macro}{header@val}
% \begin{macro}{max@header@val}
% Define some miscellaneous variables to be used internally by
% |\bf@parse@range|---$x$ position of header, current label to output,
% and maximum label to output ($+ 1$).
%    \begin{macrocode}
\newlength{\header@xpos}
\newcounter{header@val}
\newcounter{max@header@val}
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \begin{macro}{\bf@parse@range}
% This is helper function~\#2 for |\bitheader|.  It parses a
% hyphen-separated pair of numbers (or a single number) and displays the
% number at the correct bit position.
% \changes{v2.1}{2011/06/16}{Added code due to Renaud Pacalet for shifting
%   the bit header by a distance corresponding to
%   \texttt{\string\string\string\bf@first@bit}, used for typesetting
%   registers split across rows}
%    \begin{macrocode}
\def\bf@parse@range#1-#2-#3\relax{%
  \setcounter{header@val}{#1}
  \setcounter{max@header@val}{#2 + 1}
  \loop
    \ifnum\value{header@val}<\value{max@header@val}%
      \if\bf@bit@endianness b%
        \setlength{\header@xpos}{%
          \bf@bitwidth * (\bits@wide - \value{header@val} + \bf@first@bit - 1)}%
      \else
        \setlength{\header@xpos}{\bf@bitwidth * (\value{header@val} - \bf@first@bit)}%
      \fi
      \put(\strip@pt\header@xpos,0){%
        \makebox(\strip@pt\bf@bitwidth,\strip@pt\units@tall){%
          \bf@bitformatting{\theheader@val}}}
      \addtocounter{header@val}{1}
  \repeat}
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\bf@process@bitheader@opts}
% \begin{macro}{\KV@bytefield@l}
% \begin{macro}{\KV@bytefield@b}
% \begin{macro}{\KV@bytefield@l@default}
% \begin{macro}{\KV@bytefield@b@default}
% This is helper function~\#3 for |\bitheader|.  It processes the
% optional argument to |\bitheader|.
%    \begin{macrocode}
\newcommand*{\bf@process@bitheader@opts}{%
  \let\KV@bytefield@l=\KV@bitheader@l
  \let\KV@bytefield@b=\KV@bitheader@b
  \let\KV@bytefield@l@default=\KV@bitheader@l@default
  \let\KV@bytefield@b@default=\KV@bitheader@b@default
  \setkeys{bytefield}%
}
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \begin{macro}{\KV@bitheader@l}
% \begin{macro}{\KV@bitheader@b}
% For backwards compatibility we also accept the (now deprecated) |l| as
% a synonym for |endianness=little| and |b| as a synonym for
% |endianness=big|.  A typical document will specify an |endianness|
% option not as an argument to |\bitheader| but rather as a package
% option that applies to the entire document.  If the \optname{compat1}
% option was provided to \pkgname{bytefield} (determined below by the
% existence of the |\curlyshrinkage| control word), we suppress the
% deprecation warning message.
%    \begin{macrocode}
\define@key{bitheader}{l}[true]{%
  \expandafter\ifx\csname curlyshrinkage\endcsname\relax
    \PackageWarning{bytefield}{%
      The "l" argument to \protect\bitheader\space is deprecated.\MessageBreak
      Instead, please use "endianness=little", which can\MessageBreak
      even be declared globally for the entire document.\MessageBreak
      This warning occurred}%
  \fi
  \def\bf@bit@endianness{l}%
}
\define@key{bitheader}{b}[true]{%
  \expandafter\ifx\csname curlyshrinkage\endcsname\relax
    \PackageWarning{bytefield}{%
      The "b" argument to \protect\bitheader\space is deprecated.\MessageBreak
      Instead, please use "endianness=big", which can\MessageBreak
      even be declared globally for the entire document.\MessageBreak
      This warning occurred}%
  \fi
  \def\bf@bit@endianness{b}%
}
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
%
% \subsection{Word labels}
%
% \subsubsection{Curly-brace manipulation}
%
% \begin{macro}{\bf@leftcurlyshrinkage}
% \begin{macro}{\bf@rightcurlyshrinkage}
% Reduce the height of a left (right) curly brace by
% |\bf@leftcurlyshrinkage| (|\bf@rightcurlyshrinkage|) so its ends don't
% overlap whatever is above or below it.  The default value (5\,pt.) was
% determined empirically and shouldn't need to be changed.  However, on
% the off-chance the user employs a math font with very different curly
% braces from Computer Modern's, |\bf@leftcurlyshrinkage| and
% |\bf@rightcurlyshrinkage| can be modified.
%    \begin{macrocode}
\def\bf@leftcurlyshrinkage{5pt}
\def\bf@rightcurlyshrinkage{5pt}
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
% \begin{macro}{\bf@leftcurlyspace}
% \begin{macro}{\bf@rightcurlyspace}
% \begin{macro}{\bf@leftlabelspace}
% \begin{macro}{\bf@rightlabelspace}
% Define the amount of space to insert before a curly brace and before a
% word label (i.e.,~after a curly brace).
%    \begin{macrocode}
\def\bf@leftcurlyspace{1ex}
\def\bf@rightcurlyspace{1ex}
\def\bf@leftlabelspace{0.5ex}
\def\bf@rightlabelspace{0.5ex}
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \begin{macro}{\bf@leftcurly}
% \begin{macro}{\bf@rightcurly}
% Define the symbols to use as left and right curly braces.  These
% symbols must be extensible math symbols (i.e.,~they will immediately
% follow |\left| or |\right| in math mode).
%    \begin{macrocode}
\let\bf@leftcurly=\{
\let\bf@rightcurly=\}
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
% \begin{macro}{\curly@box}
% Define a box in which to temporarily store formatted curly braces.
% \changes{v1.2}{2004/06/14}{Bug fix: Defined \cs{curly@box} globally
%   (suggested by Stefan Ulrich)}
%    \begin{macrocode}
\newbox{\curly@box}
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\store@rcurly}
% \begin{macro}{\curly@height}
% \begin{macro}{\half@curly@height}
% \begin{macro}{\curly@shift}
% \begin{macro}{\old@axis}
% Store a ``\}'' that's |#2| tall in box |#1|.
% The only unintuitive thing here is that we have to redefine
% |\fontdimen22|---axis height---to~0\,pt.\ before typesetting the curly
% brace.  Otherwise, the brace would be vertically off-center by a few
% points.  When we're finished, we reset it back to its old value.
%    \begin{macrocode}
\def\store@rcurly#1#2{%
  \begingroup
    \bf@newdimen\curly@height
    \setlength{\curly@height}{#2 - \bf@rightcurlyshrinkage}%
    \bf@newdimen\half@curly@height
    \setlength{\half@curly@height}{0.5\curly@height}%
    \bf@newdimen\curly@shift
    \setlength{\curly@shift}{\bf@rightcurlyshrinkage}%
    \setlength{\curly@shift}{\half@curly@height + 0.5\curly@shift}%
    \global\sbox{#1}{\raisebox{\curly@shift}{%
      $\xdef\old@axis{\the\fontdimen22\textfont2}$%
      $\fontdimen22\textfont2=0pt%
       \left.
       \vrule height\half@curly@height
              width 0pt
              depth\half@curly@height\right\bf@rightcurly$%
      $\fontdimen22\textfont2=\old@axis$}}%
  \endgroup
}
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \begin{macro}{\store@lcurly}
% \begin{macro}{\curly@height}
% \begin{macro}{\half@curly@height}
% \begin{macro}{\curly@shift}
% These are the same as |\store@rcurly|, etc.\ but using a ``\{''
% instead of a ``\}''.
%    \begin{macrocode}
\def\store@lcurly#1#2{%
  \begingroup
    \bf@newdimen\curly@height
    \setlength{\curly@height}{#2 - \bf@leftcurlyshrinkage}%
    \bf@newdimen\half@curly@height
    \setlength{\half@curly@height}{0.5\curly@height}%
    \bf@newdimen\curly@shift
    \setlength{\curly@shift}{\bf@leftcurlyshrinkage}%
    \setlength{\curly@shift}{\half@curly@height + 0.5\curly@shift}%
    \global\sbox{#1}{\raisebox{\curly@shift}{%
      $\xdef\old@axis{\the\fontdimen22\textfont2}$%
      $\fontdimen22\textfont2=0pt%
       \left\bf@leftcurly
       \vrule height\half@curly@height
              width 0pt
              depth\half@curly@height\right.$%
      $\fontdimen22\textfont2=\old@axis$}}%
  \endgroup
}
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \subsubsection{Right-side labels}
%
% \begin{macro}{\show@wordlabelr}
% This macro is output in the third column of every row of the
% |\ialign|ed bytefield table.  It's normally a no-op, but
% |\end{rightwordgroup}| defines it to output the word label and then
% reset itself to a no-op.
%    \begin{macrocode}
\def\show@wordlabelr{}
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\wordlabelr@start}
% \begin{macro}{\wordlabelr@end}
% Declare the starting and ending height (in points) of the set of rows
% to be labeled on the right.
%    \begin{macrocode}
\newlength{\wordlabelr@start}
\newlength{\wordlabelr@end}
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
% \begin{environment}{rightwordgroup}
% \usermacro
% Label the words defined between |\begin{rightwordgroup}| and |\end{rightwordgroup}|
% on the right side of the bit field.
% The argument is the text of the label.  The label is typeset to the
% right of a large curly brace, which groups the words together.
%    \begin{macrocode}
\newenvironment{rightwordgroup}[1]{%
%    \end{macrocode}
% We begin by ending the group that |\begin{rightwordgroup}| created.  This
% lets the |rightwordgroup| environment span rows (because we're technically
% no longer within the environment).
%    \begin{macrocode}
  \endgroup
%    \end{macrocode}
% \begin{macro}{\wordlabelr@start}
% \begin{macro}{\wordlabelr@text}
% |\begin{rightwordgroup}| merely stores the starting height in |\wordlabelr@start|
% and the user-supplied text in |\wordlabelr@text|.  |\end{rightwordgroup}|
% does most of the work.
%    \begin{macrocode}
  \global\wordlabelr@start=\bytefield@height
  \gdef\wordlabelr@text{#1}%
  \ignorespaces
}{%
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \begin{macro}{\wordlabelr@end}
% Because we already ended the group that |\begin{rightwordgroup}| created
% we now have to begin a group for |\end{rightwordgroup}| to end.
%    \begin{macrocode}
  \begingroup
  \global\wordlabelr@end=\bytefield@height
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\show@wordlabelr}
% Redefine |\show@wordlabelr| to output |\bf@rightcurlyspace| space,
% followed by a large curly brace (in |\curlybox|), followed by
% |\bf@rightlabelspace| space, followed by the user's text (previously
% recorded in |\wordlabelr@text|).  We typeset |\wordlabelr@text| within
% a |tabular| environment, so \LaTeX\ will calculate its width
% automatically.
%    \begin{macrocode}
  \gdef\show@wordlabelr{%
    \sbox{\word@label@box}{%
      \begin{tabular}[b]{@{}l@{}}\wordlabelr@text\end{tabular}}%
    \settowidth{\label@box@width}{\usebox{\word@label@box}}%
    \setlength{\label@box@height}{\wordlabelr@end-\wordlabelr@start}%
    \store@rcurly{\curly@box}{\label@box@height}%
    \bf@newdimen\total@box@width
    \setlength{\total@box@width}{%
      \bf@rightcurlyspace +
      \widthof{\usebox{\curly@box}} +
      \bf@rightlabelspace +
      \label@box@width}%
    \begin{picture}(\strip@pt\total@box@width,0)
      \put(0,0){%
        \hspace*{\bf@rightcurlyspace}%
        \usebox{\curly@box}%
        \hspace*{\bf@rightlabelspace}%
        \makebox(\strip@pt\label@box@width,\strip@pt\label@box@height){%
          \usebox{\word@label@box}}}
    \end{picture}%
%    \end{macrocode}
% The last thing |\show@wordlabelr| does is redefine itself back to a no-op.
%    \begin{macrocode}
    \gdef\show@wordlabelr{}}%
%    \end{macrocode}
% \end{macro}
% \begin{macro}{\@currenvir}
% Because of our meddling with |\begingroup| and |\endgroup|, the
% current environment is all messed up.  We therefore force the
% |\end{rightwordgroup}| to succeed, even if it doesn't match the preceding
% |\begin|.
%    \begin{macrocode}
  \def\@currenvir{rightwordgroup}%
  \ignorespaces
}
%    \end{macrocode}
% \end{macro}
% \end{environment}
%
% \subsubsection{Left-side labels}
%
% \begin{macro}{\wordlabell@start}
% \begin{macro}{\wordlabell@end}
% Declare the starting and ending height (in points) of the set of rows
% to be labeled on the left.
%    \begin{macrocode}
\newlength{\wordlabell@start}
\newlength{\wordlabell@end}
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
% \begin{macro}{\total@box@width}
% Declare the total width of the next label to typeset on the left of
% the bit field, that is, the aggregate width of the text box, curly
% brace, and spaces on either side of the curly brace.
%    \begin{macrocode}
\newlength{\total@lbox@width}
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\make@lspace}
% This macro is output in the first column of every row of the
% |\ialign|ed bytefield table.  It's normally a no-op, but
% |\begin{leftwordgroup}| defines it to output enough space for the next word
% label and then reset itself to a no-op.
%    \begin{macrocode}
\gdef\make@lspace{}
%    \end{macrocode}
% \end{macro}
%
% \begin{environment}{leftwordgroup}
% \usermacro
% This environment is essentially the same as the |rightwordgroup|
% environment but puts the label on the left.  However, the following
% code is not symmetric to that of |rightwordgroup|.  The problem is that we
% encounter |\begin{leftwordgroup}| after entering the second
% (i.e.,~figure) column, which doesn't give us a chance to reserve space
% in the first (i.e.,~left label) column.  When we reach the
% |\end{leftwordgroup}|, we know the height of the group of words we wish
% to label.  However, if we try to label the words in the subsequent
% first column, we won't know the vertical offset from the ``cursor'' at
% which to start drawing the label, because we can't know the height of
% the subsequent row until we reach the second
% column.\footnote{Question: Is there a way to push the label up to the
%   \emph{top} of the subsequent row, perhaps with
%   \texttt{\string\vfill}?}
%
% Our solution is to allocate space for the box the next time we enter a
% first column.  As long as space is eventually allocated, the column
% will expand to fit that space.  |\end{leftwordgroup}| outputs the label
% immediately.  Even though |\end{leftwordgroup}| is called at the end of
% the \emph{second} column, it |\put|s the label at a sufficiently
% negative $x$ location for it to overlap the first column.  Because
% there will eventually be enough space to accomodate the label, we know
% that the label won't overlap the bit field or extend beyond the bit-field
% boundaries.
%    \begin{macrocode}
\newenvironment{leftwordgroup}[1]{%
%    \end{macrocode}
% We begin by ending the group that |\begin{rightwordgroup}| created.  This
% lets the |leftwordgroup| environment span rows (because we're technically
% no longer within the environment).
%    \begin{macrocode}
  \endgroup
%    \end{macrocode}
% \begin{macro}{\wordlabell@start}
% \begin{macro}{\wordlabell@text}
% We store the starting height and label text, which are needed by the
% |\end{leftwordgroup}|.
%    \begin{macrocode}
  \global\wordlabell@start=\bytefield@height
  \gdef\wordlabell@text{#1}%
%    \end{macrocode}
% \end{macro}
% \end{macro}
% Next, we typeset a draft version of the label into |\word@label@box|,
% which we measure (into |\total@lbox@width|) and then discard.
% We can't typeset the final version of the label until we reach the
% |\end{leftwordgroup}|, because that's when we learn the height of the
% word group.  Without knowing the height of the word group, we don't
% how how big to make the curly brace.  In the scratch version, we make
% the curly brace 5\,cm.~tall.  This should be more than large enough to
% reach the maximum curly-brace width, which is all we really care about
% at this point.
%    \begin{macrocode}
  \sbox{\word@label@box}{%
    \begin{tabular}[b]{@{}l@{}}\wordlabell@text\end{tabular}}%
  \settowidth{\label@box@width}{\usebox{\word@label@box}}%
  \store@lcurly{\curly@box}{5cm}%
  \setlength{\total@lbox@width}{%
    \bf@leftcurlyspace +
    \widthof{\usebox{\curly@box}} +
    \bf@leftlabelspace +
    \label@box@width}%
  \global\total@lbox@width=\total@lbox@width
%    \end{macrocode}
% \begin{macro}{\make@lspace}
% Now we know how wide the box is going to be (unless, of course,
% the user is using some weird math font that scales the width of a curly
% brace proportionally to its height).  So we redefine |\make@lspace| to
% output |\total@lbox@width|'s worth of space and then redefine itself
% back to a no-op.
%    \begin{macrocode}
  \gdef\make@lspace{%
    \hspace*{\total@lbox@width}%
    \gdef\make@lspace{}}%
  \ignorespaces
}{%
%    \end{macrocode}
% \end{macro}
% Because we already ended the group that |\begin{leftwordgroup}| created
% we have to start the |\end{leftwordgroup}| by beginning a group for
% |\end{leftwordgroup}| to end.
%    \begin{macrocode}
  \begingroup
%    \end{macrocode}
% The |\end{leftwordgroup}| code is comparatively straightforward.  We
% calculate the final height of the word group, and then output the
% label text, followed by |\bf@leftlabelspace| space, followed by a
% curly brace (now that we know how tall it's supposed to be), followed
% by |\bf@leftcurlyspace| space.  The trick, as described earlier, is
% that we typeset the entire label in the second column, but in a $0
% \times 0$ |picture| environment and with a negative horizontal offset
% (|\starting@point|), thereby making it overlap the first column.
%    \begin{macrocode}
  \global\wordlabell@end=\bytefield@height
  \bf@newdimen\starting@point
  \setlength{\starting@point}{%
    -\total@lbox@width - \bf@bitwidth*\bits@wide}%
  \sbox{\word@label@box}{%
    \begin{tabular}[b]{@{}l@{}}\wordlabell@text\end{tabular}}%
  \settowidth{\label@box@width}{\usebox{\word@label@box}}%
  \setlength{\label@box@height}{\wordlabell@end-\wordlabell@start}%
  \store@lcurly{\curly@box}{\label@box@height}%
  \begin{picture}(0,0)
    \put(\strip@pt\starting@point,0){%
      \makebox(\strip@pt\label@box@width,\strip@pt\label@box@height){%
        \usebox{\word@label@box}}%
      \hspace*{\bf@leftlabelspace}%
      \usebox{\curly@box}%
      \hspace*{\bf@leftcurlyspace}}
  \end{picture}%
%    \end{macrocode}
% \begin{macro}{\@currenvir}
% Because of our meddling with |\begingroup| and |\endgroup|, the
% current environment is all messed up.  We therefore force the
% |\end{leftwordgroup}| to succeed, even if it doesn't match the preceding
% |\begin|.
%    \begin{macrocode}
  \def\@currenvir{leftwordgroup}%
  \ignorespaces}
%    \end{macrocode}
% \end{macro}
% \end{environment}
%
% \subsubsection{Scratch space}
%
% \begin{macro}{\label@box@width}
% \begin{macro}{\label@box@height}
% \begin{macro}{\word@label@box}
% Declare some scratch storage for the width, height, and contents of
% the word label we're about to output.
%    \begin{macrocode}
\newlength{\label@box@width}
\newlength{\label@box@height}
\newsavebox{\word@label@box}
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \subsection{Compatibility mode}
%
% \begin{macro}{\bf@enter@compatibility@mode@i}
% \pkgname{bytefield}'s interface changed substantially with the move to
% version~2.0.  To give version~1.\textit{x} users a quick way to build
% their old documents, we provide a version~1.\textit{x} compatibility
% mode.  We don't enable this by default because it exposes a number of
% extra length registers (a precious resource) and because we want to
% encourage users to migrate to the new interface.
%    \begin{macrocode}
\newcommand{\bf@enter@compatibility@mode@i}{%
%    \end{macrocode}
% \begin{macro}{\bitwidth}
% \begin{macro}{\byteheight}
% \begin{macro}{\curlyspace}
% \begin{macro}{\labelspace}
% \begin{macro}{\curlyshrinkage}
% Define a handful of lengths that the user was allowed to |\setlength|
% explicitly in \pkgname{bytefield}~1.\textit{x}.
%    \begin{macrocode}
  \PackageInfo{bytefield}{Entering version 1 compatibility mode}%
  \newlength{\bitwidth}%
  \newlength{\byteheight}%
  \newlength{\curlyspace}%
  \newlength{\labelspace}%
  \newlength{\curlyshrinkage}%
%    \end{macrocode}
%    \begin{macrocode}
  \settowidth{\bitwidth}{\tiny 99i}%
  \setlength{\byteheight}{4ex}%
  \setlength{\curlyspace}{1ex}%
  \setlength{\labelspace}{0.5ex}%
  \setlength{\curlyshrinkage}{5pt}%
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
% \begin{macro}{\newbytefield}
% \begin{macro}{\endnewbytefield}
% \begin{environment}{bytefield}
% Redefine the |bytefield| environment in terms of the existing
% (new-interface) |bytefield| environment.  The difference is that the
% redefinition utilizes all of the preceding lengths.
%    \begin{macrocode}
  \let\newbytefield=\bytefield
  \let\endnewbytefield=\endbytefield
  \renewenvironment{bytefield}[1]{%
    \begin{newbytefield}[%
      bitwidth=\bitwidth,
      bitheight=\byteheight,
      curlyspace=\curlyspace,
      labelspace=\labelspace,
      curlyshrinkage=\curlyshrinkage]{##1}%
  }{%
    \end{newbytefield}%
  }
%    \end{macrocode}
% \end{environment}
% \end{macro}
% \end{macro}
% \begin{macro}{\wordgroupr}
% \begin{macro}{\endwordgroupr}
% \begin{macro}{\wordgroupl}
% \begin{macro}{\endwordgroupl}
% Define |\wordgroupr|, |\endwordgroupr|, |\wordgroupl|, and
% |\endwordgroupl| in terms of the new |rightwordgroup| and
% |leftwordgroup| environments.
%    \begin{macrocode}
  \def\wordgroupr{\begin{rightwordgroup}}
  \def\endwordgroupr{\end{rightwordgroup}}
  \def\wordgroupl{\begin{leftwordgroup}}
  \def\endwordgroupl{\end{leftwordgroup}}
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \begin{macro}{\bytefieldsetup}
% Disable |\bytefieldsetup| in compatibility mode because it doesn't
% work as expected.  (Every use of the compatibility-mode |bytefield|
% environment overwrites all of the figure-formatting values.)
%    \begin{macrocode}
  \renewcommand{\bytefieldsetup}[1]{%
    \PackageError{bytefield}{%
      The \protect\bytefieldsetup\space macro is not available in\MessageBreak
      version 1 compatibility mode%
    }{%
      Remove [compat1] from the \protect\usepackage{bytefield} line to
      make \protect\bytefieldsetup\MessageBreak
      available to this document.\space\space (The document may also need
      to be modified to use\MessageBreak
      the new bytefield interface.)
    }%
  }%
}
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
% \begin{macro}{\wordgroupr}
% \begin{macro}{\endwordgroupr}
% \begin{macro}{\wordgroupl}
% \begin{macro}{\endwordgroupl}
% Issue a helpful error message for the commands that were removed in
% \pkgname{bytefield}~v2.0.  While this won't help users who first
% invalid action is to modify a no-longer-extant length register such as
% |\bitwidth| or |\byteheight|, it may benefit at least a few users who
% didn't realize that the \pkgname{bytefield} interface has changed
% substantially with version~2.0.
%    \begin{macrocode}
\newcommand{\wordgroupr}{%
  \PackageError{bytefield}{%
    Macros \protect\wordgroupr, \protect\wordgroupl, \protect\endwordgroupr,
    \MessageBreak
    and \protect\endwordgroupl\space no longer exist%
  }{%
    Starting with version 2.0, bytefield uses \protect\begin{wordgroupr}...
    \MessageBreak
    \protect\end{wordgroupr} and \protect\begin{wordgroupl}...%
    \protect\end{wordgroupl}\MessageBreak
    to specify word groups and a new \protect\bytefieldsetup\space macro to
    \MessageBreak
    change bytefield's various formatting parameters.%
  }%
}
\let\endwordgroupr=\wordgroupr
\let\wordgroupl=\wordgroupr
\let\endwordgroupl=\wordgroupr
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
% \end{macro}
%
%
% \subsection{Option processing}
%
% We use the \textsf{keyval} package to handle option processing.
% Because all of \pkgname{bytefield}'s options have local impact, options
% can be specified either as package arguments or through the use of the
% |\bytefieldsetup| macro.
%
% \begin{macro}{\KV@bytefield@bitwidth}
% \begin{macro}{\bf@bw@arg}
% \begin{macro}{\bf@auto}
% Specify the width of a bit number in the bit header.  If the special
% value ``|auto|'' is given, set the width to the width of a formatted
% ``|99i|''.
%    \begin{macrocode}
\define@key{bytefield}{bitwidth}{%
  \def\bf@bw@arg{#1}%
  \def\bf@auto{auto}%
  \ifx\bf@bw@arg\bf@auto
    \settowidth{\bf@bitwidth}{\bf@bitformatting{99i}}%
  \else
    \setlength{\bf@bitwidth}{#1}%
  \fi
}
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \begin{macro}{\KV@bytefield@bf@bitheight}
% Specify the height of a bit in a |\bitbox| or |\wordbox|.
%    \begin{macrocode}
\define@key{bytefield}{bitheight}{\setlength{\bf@bitheight}{#1}}
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\KV@bytefield@bitformatting}
% Specify the style of a bit number in the bit header.  This should be
% passed an expression that takes either one argument (e.g.,~|\textit|)
% or no arguments (e.g.,~|{\small\bfseries}|).
%    \begin{macrocode}
\define@key{bytefield}{bitformatting}{\def\bf@bitformatting{#1}}
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\KV@bytefield@boxformatting}
% Specify a style to be applied to the contents of every bit box and
% word box. This should be passed an expression that takes either one
% argument (e.g.,~|\textit|) or no arguments
% (e.g.,~|{\small\bfseries}|).
%    \begin{macrocode}
\define@key{bytefield}{boxformatting}{\def\bf@boxformatting{#1}}
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\KV@bytefield@leftcurly}
% \begin{macro}{\KV@bytefield@rightcurly}
% Specify the symbol to use for bracketing a left or right word group.
% This must be an extensible math delimiter (i.e.,~something that can
% immediately follow |\left| or |\right| in math mode).
%    \begin{macrocode}
\define@key{bytefield}{leftcurly}{\def\bf@leftcurly{#1}}
\define@key{bytefield}{rightcurly}{\def\bf@rightcurly{#1}}
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
% \begin{macro}{\KV@bytefield@leftcurlyspace}
% \begin{macro}{\KV@bytefield@rightcurlyspace}
% \begin{macro}{\KV@bytefield@curlyspace}
% Specify the amount of space between the bit fields in a word group and
% the adjacent left or right curly brace.  The \optname{curlyspace}
% option is a shortcut that puts the same space before both left and
% right curly braces.
%    \begin{macrocode}
\define@key{bytefield}{leftcurlyspace}{\def\bf@leftcurlyspace{#1}}
\define@key{bytefield}{rightcurlyspace}{\def\bf@rightcurlyspace{#1}}
\define@key{bytefield}{curlyspace}{%
  \def\bf@leftcurlyspace{#1}%
  \def\bf@rightcurlyspace{#1}%
}
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \begin{macro}{\KV@bytefield@leftlabelspace}
% \begin{macro}{\KV@bytefield@rightlabelspace}
% \begin{macro}{\KV@bytefield@labelspace}
% Specify the amount of space between a left or right word group's curly
% brace and the associated label text.  The |labelspace| option is a
% shortcut that puts the same space after both left and right curly
% braces.
%    \begin{macrocode}
\define@key{bytefield}{leftlabelspace}{\def\bf@leftlabelspace{#1}}
\define@key{bytefield}{rightlabelspace}{\def\bf@rightlabelspace{#1}}
\define@key{bytefield}{labelspace}{%
  \def\bf@leftlabelspace{#1}%
  \def\bf@rightlabelspace{#1}%
}
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \begin{macro}{\KV@bytefield@leftcurlyshrinkage}
% \begin{macro}{\KV@bytefield@rightcurlyshrinkage}
% \begin{macro}{\KV@bytefield@curlyshrinkage}
% Specify the number of points by which to reduce the height of a curly
% brace (left, right, or both) so its ends don't overlap whatever's
% above or below it.
%    \begin{macrocode}
\define@key{bytefield}{leftcurlyshrinkage}{\def\bf@leftcurlyshrinkage{#1}}
\define@key{bytefield}{rightcurlyshrinkage}{\def\bf@rightcurlyshrinkage{#1}}
\define@key{bytefield}{curlyshrinkage}{%
  \def\bf@leftcurlyshrinkage{#1}%
  \def\bf@rightcurlyshrinkage{#1}%
}
%    \end{macrocode}
% \end{macro}
% \end{macro}
% \end{macro}
%
% \begin{macro}{\KV@bytefield@endianness}
% \begin{macro}{\bf@parse@endianness}
% Set the default endianness to either little-endian or big-endian.
%    \begin{macrocode}
\define@key{bytefield}{endianness}{\bf@parse@endianness{#1}}
%    \end{macrocode}
%    \begin{macrocode}
\newcommand{\bf@parse@endianness}[1]{%
  \def\bf@little{little}%
  \def\bf@big{big}%
  \def\bf@arg{#1}%
  \ifx\bf@arg\bf@little
    \def\bf@bit@endianness{l}%
  \else
    \ifx\bf@arg\bf@big
      \def\bf@bit@endianness{b}%
    \else
      \PackageError{bytefield}{%
        Invalid argument "#1" to the endianness option%
      }{%
        The endianness option must be set to either "little" or
        "big".\MessageBreak
        Please specify either endianness=little or endianness=big.
      }%
    \fi
  \fi
}
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
% \begin{macro}{\KV@bytefield@lsb}
% Specify a numerical value for the least significant bit of a word.
%    \begin{macrocode}
\define@key{bytefield}{lsb}{\def\bf@first@bit{#1}}
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\bytefieldsetup}
% \usermacro
% \changes{v2.0}{2011/01/18}{Introduced this macro to provide a more
%   convenient way of configuring \pkgname{bytefield}'s parameters}
% \begin{macro}{\bf@bytefieldsetup}
% Reconfigure values for various \pkgname{bytefield} parameters.
% Internally to the package we use the |\bf@bytefieldsetup| macro
% instead of |\bytefieldsetup|.  This enables us to redefine
% |\bytefieldsetup| when entering version~1 compatibility mode without
% impacting the rest of \pkgname{bytefield}.
%    \begin{macrocode}
\newcommand{\bf@bytefieldsetup}{\setkeys{bytefield}}
\let\bytefieldsetup=\bf@bytefieldsetup
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
% We define only a single option that can be used only as a package
% option, not as an argument to |\bytefieldsetup|: |compat1| instructs
% \pkgname{bytefield} to enter version~1 compatibility mode---at the
% cost of a number of additional length registers and the inability to
% specify parameters in the argument to the |bytefield| environment.
%    \begin{macrocode}
\DeclareOption{compat1}{\bf@enter@compatibility@mode@i}
%    \end{macrocode}
%
% \begin{macro}{\bf@package@options}
% \begin{macro}{\next}
% We want to use |\bf@bytefieldsetup| to process \pkgname{bytefield}
% package options.  Unfortunately, |\DeclareOption| doesn't handle
% \meta{key}=\meta{value} arguments.  Hence, we use |\DeclareOption*| to
% catch \emph{all} options, each of which it appends to
% |\bf@package@options|.  |\bf@package@options| is passed to
% |\bf@bytefieldsetup| only at the beginning of the document so that the
% options it specifies (a)~can refer to ex-heights and (b)~override the
% default values, which are also set at the beginning of the document.
%    \begin{macrocode}
\def\bf@package@options{}
\DeclareOption*{%
  \edef\next{%
    \noexpand\g@addto@macro\noexpand\bf@package@options{,\CurrentOption}%
  }%
  \next
}
\ProcessOptions\relax
\expandafter\bf@bytefieldsetup\expandafter{\bf@package@options}
%    \end{macrocode}
% \end{macro}
% \end{macro}
%
% \Finale
\endinput