\input texinfo  @c -*-texinfo-*-
@c %**start of header
@setfilename udunits2lib.info
@settitle The UDUNITS-2 C API Guide
@c %**end of header
@dircategory Software libraries
@direntry
* UDUNITS-2: (udunits2lib).     The Unidata units library.
@end direntry
@syncodeindex fn cp
@syncodeindex tp cp
@copying
@include LICENSE
@end copying

@titlepage
@title The UDUNITS-2 C API
@author Steven R. Emmerson

@page
@vskip 0pt plus 1filll
@insertcopying
@end titlepage

@contents

@ifnottex
@node Top, Synopsis, (dir), (dir)
@top UDUNITS-2
This manual describes how to use the C API of the UDUNITS-2 library.
Among other things, the library allows C code to obtain a binary
representation of a unit of a physical quantity, to
operate on such units, and to convert numeric values between compatible
units.

The library comes with an extensive database of units all referenced to the
SI system of units.

@insertcopying
@end ifnottex

@menu
* Synopsis::            Terse usage display
* Why::                 What's a unit package good for?
* Unit-Systems::        Explanation of unit-systems and how to get one
* Parsing::             Converting strings into units
* Syntax::              Syntax for string representation of units
* Formatting::          Converting units into strings
* Value Conversion::    Converting values between units
* Prefixes::            Defining unit prefixes
* Mapping::             Mapping between units and identifiers
* Operations::          Operations on units
* Time::                Handling time
* Errors::              Error-handling
* Database::            The units database
* Types::               Data types
* Complete Index::      Complete index
@end menu

@node Synopsis, Why, Top, Top
@chapter Synopsis
@cindex synopsis

Coding:
@example
#include <udunits2.h>
@quotation
@multitable {ut_error_message_handler} {ut_get_dimensionless_unit_one(}
@item ut_system* @tab ut_read_xml(const char* @var{path});
@item ut_system* @tab ut_new_system(void);
@item void      @tab ut_free_system(ut_system* @var{system});
@item ut_system* @tab ut_get_system(const ut_unit* @var{unit});
@item ut_unit*   @tab ut_get_dimensionless_unit_one(const ut_system* @var{system});
@item ut_unit*   @tab ut_get_unit_by_name(const ut_system* @var{system}, const char* @var{name});
@item ut_unit*   @tab ut_get_unit_by_symbol(const ut_system* @var{system}, const char* @var{symbol});
@item ut_status @tab  ut_set_second(const ut_unit* @var{second});
@item ut_status @tab  ut_add_name_prefix(ut_system* @var{system}, const char* @var{name}, double @var{value});
@item ut_status @tab  ut_add_symbol_prefix(ut_system* @var{system}, const char* @var{symbol}, double @var{value});
@item ut_unit*   @tab ut_new_base_unit(ut_system* @var{system});
@item ut_unit*   @tab ut_new_dimensionless_unit(ut_system* @var{system});
@item ut_unit*   @tab ut_clone(const ut_unit* @var{unit});
@item void      @tab ut_free(ut_unit* @var{unit});
@item const char* @tab ut_get_name(const ut_unit* @var{unit}, ut_encoding @var{encoding});
@item ut_status @tab ut_map_name_to_unit(const char* @var{name}, const ut_encoding @var{encoding}, const ut_unit* @var{unit});
@item ut_status @tab ut_unmap_name_to_unit(ut_system* @var{system}, const char* @var{name}, const ut_encoding @var{encoding});
@item ut_status @tab ut_map_unit_to_name(const ut_unit* @var{unit}, const char* @var{name}, ut_encoding @var{encoding});
@item ut_status @tab ut_unmap_unit_to_name(const ut_unit* @var{unit}, ut_encoding @var{encoding});
@item const char* @tab ut_get_symbol(const ut_unit* @var{unit}, ut_encoding @var{encoding});
@item ut_status @tab ut_map_symbol_to_unit(const char* @var{symbol}, const ut_encoding @var{encoding}, const ut_unit* @var{unit});
@item ut_status @tab ut_unmap_symbol_to_unit(ut_system* @var{system}, const char* @var{symbol}, const ut_encoding @var{encoding});
@item ut_status @tab ut_map_unit_to_symbol(const ut_unit* @var{unit}, const char* @var{symbol}, ut_encoding @var{encoding});
@item ut_status @tab ut_unmap_unit_to_symbol(const ut_unit* @var{unit}, ut_encoding @var{encoding});
@item int @tab ut_is_dimensionless(const ut_unit* @var{unit});
@item int @tab ut_same_system(const ut_unit* @var{unit1}, const ut_unit* @var{unit2});
@item int @tab ut_compare(const ut_unit* @var{unit1}, const ut_unit* @var{unit2});
@item int @tab ut_are_convertible(const ut_unit* @var{unit1}, const ut_unit* @var{unit2});
@item cv_converter* @tab ut_get_converter(ut_unit* @var{from}, ut_unit* @var{to});
@item ut_unit* @tab ut_scale(double @var{factor}, const ut_unit* @var{unit});
@item ut_unit* @tab ut_offset(const ut_unit* @var{unit}, double @var{offset});
@item ut_unit* @tab ut_offset_by_time(const ut_unit* @var{unit}, double @var{origin});
@item ut_unit* @tab ut_multiply(const ut_unit* @var{unit1}, const ut_unit* @var{unit2});
@item ut_unit* @tab ut_invert(const ut_unit* @var{unit});
@item ut_unit* @tab ut_divide(const ut_unit* @var{numer}, const ut_unit* @var{denom});
@item ut_unit* @tab ut_raise(const ut_unit* @var{unit}, int @var{power});
@item ut_unit* @tab ut_root(const ut_unit* @var{unit}, int @var{root});
@item ut_unit* @tab ut_log(double @var{base}, const ut_unit* @var{reference});
@item ut_unit* @tab ut_parse(const ut_system* @var{system}, const char* @var{string}, ut_encoding @var{encoding});
@item char*   @tab ut_trim(char* @var{string}, ut_encoding @var{encoding});
@item int @tab ut_format(const ut_unit* @var{unit}, char* @var{buf}, size_t @var{size}, unsigned @var{opts});
@item ut_status @tab ut_accept_visitor(const ut_unit* @var{unit}, const ut_visitor* @var{visitor}, void* @var{arg});
@item double @tab ut_encode_date(int @var{year}, int @var{month}, int @var{day});
@item double @tab ut_encode_clock(int @var{hours}, int @var{minutes}, double @var{seconds});
@item double @tab ut_encode_time(int @var{year}, int @var{month}, int @var{day}, int @var{hour}, int @var{minute}, double @var{second});
@item void @tab ut_decode_time(double @var{value}, int* @var{year}, int* @var{month}, int* @var{day}, int* @var{hour}, int* @var{minute}, double* @var{second}, double* @var{resolution});
@item ut_status @tab ut_get_status(void);
@item void @tab ut_set_status(ut_status @var{status});
@item int @tab ut_handle_error_message(const char* @var{fmt}, ...);
@item ut_error_message_handler@tab ut_set_error_message_handler(ut_error_message_handler @var{handler});
@item int @tab ut_write_to_stderr(const char* @var{fmt}, va_list @var{args});
@item int @tab ut_ignore(const char* @var{fmt}, va_list @var{args});
@item 
@item float @tab cv_convert_float(const cv_converter* @var{converter}, float @var{value});
@item double @tab cv_convert_double(const cv_converter* @var{converter}, double @var{value});
@item float* @tab cv_convert_floats(const cv_converter* @var{converter}, const float* @var{in}, size_t @var{count}, float* @var{out});
@item double* @tab cv_convert_doubles(const cv_converter* @var{converter}, const double* @var{const} in, @var{size_t} count, @var{double}* out);
@item void @tab cv_free(cv_converter* @var{conv});
@end multitable
@end quotation
@end example

Compiling:
@example
c89 -I @emph{includedir} ...
@end example

Where @emph{includedir} is the installation-directory for C header
files (e.g., @code{/usr/local/include}).

Linking:
@example
c89 ... -L@emph{libdir} -ludunits2 -lexpat ... -lm
@end example

Where @emph{libdir} is the installation-directory for object code
libraries (e.g., @code{/usr/local/lib}).

@node Why, Unit-Systems, Synopsis, Top
@chapter What's a Unit Package Good For?

The existance of a software package is justified by what you can do with it.
The three main things you can do with the UDUNIT-2 package are
@enumerate
@item
@ref{Value Conversion,
Convert numeric values between compatible units}.
@item
Convert a string representation of a unit into a binary one --- enabling
the programatic manipulation of units.
There are three ways to do this:
@itemize
@item
@ref{Extracting,Get the unit} from a @ref{unit-system}.
This requires that
you know the unit's name or symbol and that the unit is in a unit-system.  
@item
@ref{Parsing,Parse a string representation of the unit into its binary
representation}.
This requires that the string be parsable by @code{@ref{ut_parse()}}.
@item
@ref{Operations,Explicity construct the unit from subcomponent units
using unit operations}.
@end itemize
@item
@ref{Formatting,Convert a binary representation of a unit into a string} ---
enabling the printing and storing of units in a human-readable form.
@end enumerate
While the above might seem to be trivial activities, their 
general availability at the time might have helped prevent
the @uref{http://en.wikipedia.org/wiki/Mars_Climate_Orbiter,
Mars Climate Orbiter} fiasco.

@anchor{unit-system}
@node Unit-Systems, Value Conversion, Why, Top
@chapter Unit-Systems
@cindex unit-system
@cindex system of units

A unit-system is a set of units that are all defined in terms of the same
set of
@cindex unit, base
@cindex base unit
base units.
In the SI system of units, for example, the base units are the
meter, kilogram, second, ampere, kelvin, mole, and candela.
(For definitions of these base units, see
@uref{http://@/physics.nist.gov/@/cuu/@/Units/@/current.html}.)

In the UDUNITS-2 package, every accessible unit belongs to one and only one
unit-system.  It is not possible to convert numeric values between units
of different unit-systems.  Similarly, units belonging to different
unit-systems always compare unequal.

There are several categories of operations on unit-systems:

@menu
* Obtaining::   How to obtain a unit-system.
* Extracting::  Getting a unit from a unit-system.
* Adding::      Adding new units to a unit-system.
* Prefixes::    Add new unit-prefixes to a unit-system.
* Misc::        Miscelaneous unit-system operations.
@end menu

@node Obtaining, Extracting, , Unit-Systems
@section Obtaining a Unit-System
@cindex database, unit, obtaining predefined
@cindex unit database, obtaining predefined
@cindex unit-system, obtaining predefined
@cindex units, obtaining predefined
@cindex @code{ut_read_xml()}, discussion of
Typically, you would obtain a unit-system of predefined units by reading
the default unit database using @code{@ref{ut_read_xml()}}
with a @code{NULL} pathname argument.
If this doesn't quite match your needs, then there are alternatives.
Together with the typical solution, the means for obtaining a useful
unit-system are (in order of increasing difficulty):

@itemize @bullet
@item
Obtain the default unit-system using @code{@ref{ut_read_xml(),ut_read_xml}(NULL)}.
@item
Copy and customize the unit database and then
call @code{@ref{ut_read_xml()}} with the pathname of the customized
database to obtain a customized unit-system.
@item
Same as either of the above but then adding new units to the unit-system using
@code{@ref{ut_new_base_unit()}} and
@code{@ref{ut_new_dimensionless_unit()}}.
@item
Same as the above but also deriving new units using
@ref{Operations, unit operations}
and then adding them to the unit-system using
@ref{Mapping, unit mapping}.
@item
Same as the above but starting with an empty unit-system obtained from
@code{@ref{ut_new_system()}},
in which case you will definitely have to start with
@code{@ref{ut_new_base_unit()}} and
@code{@ref{ut_new_dimensionless_unit()}}.
@end itemize

You should pass every unit-system pointer to @code{@ref{ut_free_system()}}
when you no longer need the corresponding unit-system.

@anchor{ut_read_xml()}
@deftypefun @code{ut_system*} ut_read_xml @code{(const char* @var{path})}
Reads the XML-formatted unit-database specified by @var{path}
and returns the corresponding unit-sytem.
If @var{path} is @code{NULL}, then the pathname specified by the
environment variable @code{UDUNITS2_XML_PATH} is used if set; otherwise,
the compile-time pathname of the installed, default, unit database is
used.
You should pass the returned pointer to @code{ut_free_system()} when you
no longer need the unit-system.
If an error occurs,
then this function writes an error-message using
@code{@ref{ut_handle_error_message()}}
and returns @code{NULL}.
Also, @code{@ref{ut_get_status()}} will return one of the following:

@table @code
@item UT_OPEN_ARG
@var{path} is non-@code{NULL} but the file couldn't be opened.
See @code{errno} for the reason.
@item UT_OPEN_ENV
@var{path} is @code{NULL} and environment variable
@code{UDUNITS2_XML_PATH} is set but the file couldn't be opened.
See @code{errno} for the reason.
@item UT_OPEN_DEFAULT
@var{path} is @code{NULL}, environment variable
@code{UDUNITS2_XML_PATH} is unset, and the installed, default, unit
database couldn't be opened.
See @code{errno} for the reason.
@item UT_OS
Operating-system error.  See @code{errno}.
@item UT_PARSE
The database file couldn't be parsed.
@end table
@end deftypefun

@anchor{ut_new_system()}
@deftypefun @code{ut_system*} ut_new_system @code{(void)}
Creates and returns a new unit-system.
On success, the unit-system will be empty
except for the dimensionless unit one.
You should pass the returned pointer to @code{ut_free_system()} when you
no longer need the unit-system.
If an error occurs,
then this function writes an error-message using
@code{@ref{ut_handle_error_message()}}
and returns @code{NULL}.
Also, @code{@ref{ut_get_status()}} will return the following:

@table @code
@item UT_OS
Operating-system error.  See @code{errno}.
@end table
@end deftypefun

@node Extracting, Adding, Obtaining, Unit-Systems
@section Extracting Units from a Unit-System

@strong{NOTE:} This section covers low-level access to the indidual units of a
@ref{unit-system}.
General parsing of arbitrary unit specifications is coverted in
the section @ref{Parsing}.

A @ref{unit-system} contains mappings from identifiers to units (and
vice versa).
Consequently, once you have a unit-system, you can easily
obtain a unit for which you know the name or symbol using
the function @code{@ref{ut_get_unit_by_name()}} or
@code{@ref{ut_get_unit_by_symbol()}}.

@cindex getting a unit by its name
@cindex unit, getting by name
@anchor{ut_get_unit_by_name()}
@deftypefun @code{ut_unit*} ut_get_unit_by_name @code{(const ut_system* @var{system}, const char* @var{name})}
Returns the unit to which @var{name} maps from the unit-system referenced by
@var{system} or @code{NULL} if no such unit exists.
Name comparisons are case-insensitive.
If this function returns @code{NULL}, then
@code{@ref{ut_get_status()}} will return 
one of the following:

@table @code
@item UT_SUCCESS
@var{name} doesn't map to a unit of @var{system}.
@item UT_BAD_ARG
@var{system} or @var{name} is @code{NULL}.
@end table
@end deftypefun

@cindex getting a unit by its symbol
@cindex unit, getting by symbol
@anchor{ut_get_unit_by_symbol()}
@deftypefun @code{ut_unit*} ut_get_unit_by_symbol @code{(const ut_system* @var{system}, const char* @var{symbol})}
Returns the unit to which @var{symbol} maps from the unit-system referenced by
@var{system} or @code{NULL} if no such unit exists.
Symbol comparisons are case-sensitive.
If this function returns @code{NULL}, then
@code{@ref{ut_get_status()}} will return 
one of the following:

@table @code
@item UT_SUCCESS
@var{symbol} doesn't map to a unit of @var{system}.
@item UT_BAD_ARG
@var{system} or @var{symbol} is @code{NULL}.
@end table
@end deftypefun

@anchor{ut_get_dimensionless_unit_one()}
@deftypefun @code{ut_unit*} ut_get_dimensionless_unit_one @code{(const ut_system* @var{system})}
Returns the dimensionless unit one of the unit-system referenced by
@var{system}.
While not necessary, the returned pointer may be passed to @code{ut_free()}
when you no longer need the unit.
If @var{system} is @code{NULL}, then this function writes an error-message 
using @code{@ref{ut_handle_error_message()}}
and returns @code{NULL}.
Also, @code{@ref{ut_get_status()}} will return @code{UT_BAD_ARG}.
@end deftypefun

@node Adding, Prefixes, Extracting, Unit-Systems
@section Adding Units to a Unit-System
@cindex adding units to a unit-system
@cindex unit, adding to a unit-system
@cindex unit-system, adding a unit to

If you use @code{@ref{ut_read_xml()}}, then you should not normally need to
add any new units to a unit-system.

Because you get units via their names or symbols, adding a unit to a
unit-system actually means mapping one or more identifiers (i.e., names
or symbols) to the unit.
Thereafter, you can use @code{@ref{ut_get_unit_by_name()}} and
@code{@ref{ut_get_unit_by_symbol()}} to retrieve the unit.
The mapping of identifiers to units is covered @ref{Mapping,here}.

Having said that, it is possible to create a new base or dimensionless
unit within a unit-system using @code{@ref{ut_new_base_unit()}} or
@code{@ref{ut_new_dimensionless_unit()}}---you'll just also have to map
identifiers to the newly-created unit in order to be able to retrieve
it later by identifier.

@anchor{ut_new_base_unit()}
@deftypefun @code{ut_unit*} ut_new_base_unit @code{(ut_system* @var{system})}
Creates and adds a new base-unit to the unit-system referenced by @var{system}.
This function returns the new base-unit.
You should pass the returned pointer to @code{ut_free()} when you
no longer need the unit.
If an error occurs,
then this function writes an error-message using
@code{@ref{ut_handle_error_message()}}
and returns @code{NULL}.
Also, @code{@ref{ut_get_status()}} will return one of the following:

@table @code
@item UT_BAD_ARG
@var{system} is @code{NULL}.
@item UT_OS
Operating-system failure.  See @code{errno}.
@end table
If you use @code{@ref{ut_read_xml()}}, then you should not normally need to call this
function.
@end deftypefun

@anchor{ut_new_dimensionless_unit()}
@deftypefun @code{ut_unit*} ut_new_dimensionless_unit @code{(ut_system* @var{system})}
Creates and adds a new dimensionless-unit to the unit-system referenced by @var{system}.
This function returns the new dimensionless-unit.
You should pass the returned pointer to @code{ut_free()} when you
no longer need the unit.
If an error occurs,
then this function writes an error-message using
@code{@ref{ut_handle_error_message()}}
and returns @code{NULL}.
Also, @code{@ref{ut_get_status()}} will return one of the following:

@table @code
@item UT_BAD_ARG
@var{system} is @code{NULL}.
@item UT_OS
Operating-system failure.  See @code{errno}.
@end table
If you use @code{@ref{ut_read_xml()}}, then you should not normally need to call this
function.
@end deftypefun

@node Prefixes, Misc, Adding, Unit-Systems
@section Adding Unit-Prefixes to a Unit-System
@cindex prefixes, adding to a unit-system
@cindex adding prefixes to a unit-system
@cindex unit-system, adding prefixes to a

A prefix is a word or symbol that is appended to the beginning of a word or
symbol that represents a unit in order to modify the value of that unit.  For
example, the prefix ``kilo'' in the word ``kiloamperes'' changes the value
from one ampere to one-thousand amperes.

If you use @code{@ref{ut_read_xml()}}, then you should not normally need to
add any new prefixes to a unit-system.

@anchor{ut_add_name_prefix()}
@deftypefun @code{@ref{ut_status}} ut_add_name_prefix @code{(ut_system* @var{system}, const char* @var{name}, double @var{value})}
Adds the name-prefix @var{name} with the value @var{value}
to the unit-system @var{system}.
A name-prefix is something like ``mega'' or ``milli''.
Comparisons between name-prefixes are case-insensitive.
This function returns one of the following:

@table @code
@item UT_SUCCESS
Success.
@item UT_BAD_ARG
@var{system} or @var{name} is @code{NULL}, or @var{value} is @code{0}.
@item UT_EXISTS
@var{name} already maps to a different value.
@item UT_OS
Operating-system failure.  See @code{errno}.
@end table
@end deftypefun

@anchor{ut_add_symbol_prefix()}
@deftypefun @code{@ref{ut_status}} ut_add_symbol_prefix @code{(ut_system* @var{system}, const char* @var{symbol}, double @var{value})}
Adds the symbol-prefix @var{symbol} with the value @var{value}
to the unit-system @var{system}.
A symbol-prefix is something like ``M'' or ``m''.
Comparisons between symbol-prefixes are case-sensitive.
This function returns one of the following:

@table @code
@item UT_SUCCESS
Success.
@item UT_BAD_ARG
@var{system} or @var{symbol} is @code{NULL}, or @var{value} is @code{0}.
@item UT_EXISTS
@var{symbol} already maps to a different value.
@item UT_OS
Operating-system failure.  See @code{errno}.
@end table
@end deftypefun

@node Misc, , Prefixes, Unit-Systems
@section Miscelaneous Operations on Unit-Systems

@anchor{ut_free_system()}
@deftypefun @code{void} ut_free_system @code{(ut_system* @var{system})}
Frees the unit-system referenced by @var{system}.  All unit-to-identifier and
identifier-to-unit mappings are removed.  Use of @code{system} after this
function returns results in undefined behavior.
@end deftypefun

@anchor{ut_set_second()}
@deftypefun @code{@ref{ut_status}} ut_set_second @code{(const ut_unit* @var{second})}
Sets the ``second'' unit of a unit-system.  This function must be called before
the first call to @code{ut_offset_by_time()} for a unit in the same unit-system.
@code{@ref{ut_read_xml()}} calls this function if the
unit-system it's reading contains a unit named ``second''.
This function returns one of the following:

@table @code
@item UT_SUCCESS
The ``second'' unit of @var{system} was successfully set.
@item UT_EXISTS
The ``second'' unit of @var{system} is set to a different unit.
@item UT_BAD_ARG
@var{second} is @code{NULL}.
@end table
@end deftypefun

@node Value Conversion, Parsing, Unit-Systems, Top
@chapter Converting Values Between Units
@cindex converting values between units
@cindex unit conversion

You can convert numeric values in one unit to equivalent values in
another, compatible unit by means of a converter.
For example

@example
#include <udunits2.h>
...
    ut_unit*      from = ...;
    ut_unit*      to = ...;
    cv_converter* converter = ut_get_converter(from, to);
    double       fromValue = ...;
    double       toValue = cv_convert_double(converter, fromValue);

    cv_free(converter);
@end example

The converter API is declared in the header-file @code{<converter.h>},
which is automatically included by the UDUNITS-2 header-file
(@code{<udunits2.h>}) so you don't need to explicitly include it.

@anchor{ut_are_convertible()}
@deftypefun @code{int} ut_are_convertible @code{(const ut_unit* @var{unit1}, uconst t_unit* @var{unit2})}
Indicates if numeric values in unit @var{unit1} are convertible to numeric
values in unit @var{unit2} via @ref{ut_get_converter()}.
In making this determination, dimensionless units are ignored.
This function returns a non-zero value if conversion is possible;
otherwise, @code{0} is returned and @ref{ut_get_status()} will return one
of the following:

@table @code
@item UT_BAD_ARG
@var{unit1} or @var{unit2} is @code{NULL}.
@item UT_NOT_SAME_SYSTEM
@var{unit1} and @var{unit2} belong to different @ref{unit-system}s.
@item UT_SUCCESS
Conversion between the units is not possible (e.g., @var{unit1} refers
to a meter and @var{unit2} refers to a kilogram.
@end table
@end deftypefun

@anchor{ut_get_converter()}
@deftypefun @code{cv_converter*} ut_get_converter @code{(ut_unit* const @var{from}, ut_unit* const @var{to})}
Creates and returns a converter of numeric values in the @var{from} unit
to equivalent values in the @var{to} unit.
You should pass the returned pointer to @code{cv_free()} when you
no longer need the converter.
If an error occurs,
then this function writes an error-message using
@code{@ref{ut_handle_error_message()}}
and returns @code{NULL}.
Also, @code{@ref{ut_get_status()}} will return one of the following:

@table @code
@item UT_BAD_ARG
@var{from} or @var{to} is @code{NULL}.
@item UT_NOT_SAME_SYSTEM
The units @var{from} and @var{to} don't belong to the same unit-system.
@item UT_MEANINGLESS
The units belong to the same unit-system but conversion between them is
meaningless (e.g., conversion between seconds and kilograms is meaningless).
@item UT_OS
Operating-system failure.  See @code{errno}.
@end table
@end deftypefun

@anchor{cv_convert_float()}
@deftypefun @code{float} cv_convert_float @code{(const cv_converter* @var{converter}, const float @var{value})}
Converts the single floating-point value @var{value} and
returns the new value.
@end deftypefun

@anchor{cv_convert_double()}
@deftypefun @code{double} cv_convert_double @code{(const cv_converter* @var{converter}, const double @var{value})}
Converts the single double-precision value @var{value} and
returns the new value.
@end deftypefun

@anchor{cv_convert_floats()}
@deftypefun @code{float*} cv_convert_floats @code{(const cv_converter* @var{converter}, const float* @var{in}, size_t @var{count}, float* @var{out})}
Converts the @var{count} floating-point values starting at @var{in}, writing
the new values starting at @var{out} and, as a convenience,
returns @var{out}.
The input and output arrays may overlap or be identical.
@end deftypefun

@anchor{cv_convert_doubles()}
@deftypefun @code{double*} cv_convert_doubles @code{(const cv_converter* @var{converter}, const double* @var{in}, size_t @var{count}, double* @var{out})}
Converts the @var{count} double-precision values starting at @var{in}, writing
the new values starting at @var{out} and, as a convenience,
returns @var{out}.
The input and output arrays may overlap or be identical.
@end deftypefun

@anchor{cv_free()}
@deftypefun @code{void} cv_free @code{(cv_converter* @var{conv})};
Frees resources associated with the converter referenced by @var{conv}.
You should call this function when you no longer need the converter.
Use of @var{conv} upon return results in undefined behavior.
@end deftypefun

@node Parsing, Syntax, Value Conversion, Top
@chapter Parsing a String into a Unit
@cindex parsing a string into a unit
@cindex string, parsing into a unit

Here's an example of parsing a string representation of a unit into
its binary representation:

@example
#include <stdlib.h>
#include <udunits2.h>
...
    ut_system*   unitSystem = @ref{ut_read_xml(),ut_read_xml(NULL)};
    const char* string = "kg.m2/s3";
    ut_unit*     watt = @ref{ut_parse(),ut_parse}(unitSystem, string, UT_ASCII);

    if (watt == NULL) @{
        /* Unable to parse string. */
    @}
    else @{
        /* Life is good. */
    @}
@end example

@anchor{ut_parse()}
@deftypefun @code{ut_unit*} ut_parse @code{(const ut_system* @var{system}, const char* @var{string}, ut_encoding @var{encoding})}
Returns the binary unit representation corresponding to the string unit
representation @var{string} in the character-set @var{encoding} using the
unit-system @var{system}.
@var{string} must have no leading or trailing whitespace (see
@code{@ref{ut_trim()}}).
If an error occurs, then this function returns @code{NULL} and
@code{@ref{ut_get_status()}} will return one of the following:

@table @code
@item UT_BAD_ARG
@var{system} or @var{string} is @code{NULL}.
@item UT_SYNTAX
@var{string} contained a syntax error.
@item UT_UNKNOWN
@var{string} contained an unknown identifier.
@item UT_OS
Operating-system failure.  See @code{errno} for the reason.
@end table
@end deftypefun

@anchor{ut_trim()}
@deftypefun @code{size_t} ut_trim @code{(char* @var{string}, ut_encoding @var{encoding})}
Removes all leading and trailing whitespace from the NUL-terminated string
@var{string}.  Returns @var{string}, which is modified if it contained leading
or trailing whitespace.
@end deftypefun

@node Syntax, Formatting, Parsing, Top
@chapter Unit Syntax
@cindex unit syntax
@cindex syntax, unit

For the most part, the UDUNITS-2 package follows the syntax for unit-strings
promulgated by the US National Institute for Standards and Technology (NIST).
Details, of which, can be found at @uref{http://physics.nist.gov/cuu/Units/index.html}.
The one general exception to this is the invention of a syntax for
``offset''-units (e.g., the definition of the degree Celsius is ``K @@
273.15'').

@menu
* Examples::  Examples of unit specifications
* Grammar::   Formal unit grammar
@end menu

@node Examples, Grammar, , Syntax
@section Unit Specification Examples
@cindex unit specification examples
@cindex examples, unit specification

@quotation
@multitable {Logarithmic} {(5 meter)/(30 second)} {lg(re mW)} {"lg" is base 10, "ln" is base e, and "lb" is base 2}
@headitem String Type @tab Using Names @tab Using Symbols @tab Comment
@item Simple @tab meter @tab m
@item Raised @tab meter^2 @tab m2 @tab higher precedence than multiplying or
dividing
@item Product @tab newton meter @tab N.m
@item Quotient @tab meter per second @tab m/s
@item Scaled @tab 60 second @tab 60 s
@item Prefixed @tab kilometer @tab km
@item Offset @tab kelvin from 273.15 @tab K @@ 273.15 @tab lower
precedence than multiplying or dividing
@item Logarithmic @tab lg(re milliwatt) @tab lg(re mW) @tab "lg" is base 10, 
"ln" is base e, and "lb" is base 2
@item Grouped @tab (5 meter)/(30 second) @tab (5 m)/(30 s)
@end multitable
@end quotation

The above may be combined, e.g., "0.1 lg(re m/(5 s)^2) @@ 50".

You may also look at the @code{<def>} elements in @ref{Database,the
units database} to see examples of string unit specifications.

You may use the @code{@ref{Top, , udunits2, udunits2prog}} utility 
to experiment with string unit specifications.

@node Grammar, , Examples, Syntax
@section Unit Grammar
@cindex unit grammar
@cindex grammar, unit

Here is the unit-syntax understood by the UDUNITS-2 package.  Words printed
@emph{Thusly} indicate non-terminals; words printed THUSLY indicate terminals;
and words printed <thusly> indicate lexical elements.

@example
@emph{Unit-Spec: one of}
        nothing
        @emph{Shift-Spec}

@emph{Shift-Spec: one of}
        @emph{Product-Spec}
        @emph{Product-Spec} SHIFT REAL
        @emph{Product-Spec} SHIFT INT
        @emph{Product-Spec} SHIFT @emph{Timestamp}

@emph{Product-Spec: one of}
        @emph{Power-Spec}
        @emph{Product-Spec} @emph{Power-Spec}
        @emph{Product-Spec} MULTIPLY @emph{Power-Spec}
        @emph{Product-Spec} DIVIDE @emph{Power-Spec}

@emph{Power-Spec: one of}
        @emph{Basic-Spec}
        @emph{Basic-Spec} INT
        @emph{Basic-Spec} EXPONENT
        @emph{Basic-Spec} RAISE INT

@emph{Basic-Spec: one of}
        ID
        "(" @emph{Shift-Spec} ")"
        LOGREF @emph{Product_Spec} ")"
        @emph{Number}

@emph{Number: one of}
        INT
        REAL

@emph{Timestamp: one of}
        DATE
        DATE CLOCK
        DATE CLOCK CLOCK
        DATE CLOCK INT
        DATE CLOCK ID
        TIMESTAMP
        TIMESTAMP INT
        TIMESTAMP ID

SHIFT:
        <space>* <shift_op> <space>*

<shift_op>: one of
        "@@"
        "after"
        "from"
        "since"
        "ref"

REAL:
        the usual floating-point format

INT:
        the usual integer format

MULTIPLY: one of
        "-"
        "."
        "*"
        <space>+
        <centered middot>

DIVIDE:
        <space>* <divide_op> <space>*

<divide_op>: one of
        per
        PER
        "/"

EXPONENT:
        ISO-8859-9 or UTF-8 encoded exponent characters

RAISE: one of
        "^"
        "**"

ID: one of
        <id>
        "%"
        "'"
        "\""
        degree sign
        greek mu character

<id>:
        <alpha> <alphanum>*

<alpha>:
        [A-Za-z_]
        ISO-8859-1 alphabetic characters
        non-breaking space

<alphanum>: one of
        <alpha>
        <digit>

<digit>:
        [0-9]

LOGREF:
        <log> <space>* <logref>

<log>: one of
        "log"
        "lg"
        "ln"
        "lb"

<logref>:
        "(" <space>* <re> ":"? <space>*

DATE:
        <year> "-" <month> ("-" <day>)?

<year>:
        [+-]?[0-9]@{1,4@}

<month>:
        "0"?[1-9]|1[0-2]

<day>:
        "0"?[1-9]|[1-2][0-9]|"30"|"31"

CLOCK:
        <hour> ":" <minute> (":" <second>)?

TIMSTAMP:
        <year> (<month> <day>?)? "T" <hour> (<minute> <second>?)?

<hour>:
        [+-]?[0-1]?[0-9]|2[0-3]

<minute>:
        [0-5]?[0-9]

<second>:
        (<minute>|60) (\.[0-9]*)?

@end example

@node Formatting, Operations, Syntax, Top
@chapter Formatting a Unit into a String
@cindex formatting a unit into a string
@cindex unit, formatting into a string
@cindex string, formatting a unit into a

Use the @code{@ref{ut_format()}} function to obtain the string representation
of a binary unit.
For example, the following gets the definition of the unit "watt" in
ASCII characters using unit-symbols rather than unit-names:

@example
ut_unit*     watt = ...;
char        buf[128];
unsigned    opts = @ref{ut_encoding,UT_ASCII} | UT_DEFINITION;
int         len = @ref{ut_format(),ut_format}(watt, buf, sizeof(buf), opts);

if (len == -1) @{
    /* Couldn't get string */
@}
else if (len == sizeof(buf)) @{
    /* Entire buffer used: no terminating NUL */
@}
else @{
    /* Have string with terminating NUL */
@}
@end example

@anchor{ut_format()}
@deftypefun @code{int} ut_format @code{(const ut_unit* @var{unit}, char* @var{buf}, size_t @var{size}, unsigned @var{opts})}
Formats the unit @var{unit} (i.e., returns its string representation)
into the buffer pointed-to by @var{buf} of size @var{size}.
The argument @var{opts} specifies how the formatting is to be done and
is a bitwise OR of a @ref{ut_encoding} value and zero or more of the following:

@table @code
@item UT_NAMES
Use unit names instead of symbols.
@item UT_DEFINITION
The formatted string should be the definition of @var{unit} in terms
of basic-units instead of stopping any expansion at the highest level
possible.
@end table

On success, this function returns either the number of bytes -- excluding the
terminating @code{NUL} -- that were written into @code{buf} or the number of
bytes that @emph{would have been written}. The difference is due to the
the runtime @code{snprinf()} function that was used.

On failure, this function returns @code{-1} and @ref{ut_get_status()} will
return one of the following:

@table @code
@item UT_BAD_ARG
@var{unit} or @var{buf} is @code{NULL}, or 
@var{opts} contains the bit patterns of both @code{UT_LATIN1} and
@code{UT_UTF8}.
@item UT_CANT_FORMAT
@var{unit} can't be formatted in the desired manner (e.g., @var{opts}
contains @code{UT_ASCII} but @var{unit} doesn't have an identifier in
that character-set or @var{opts} doesn't contain UT_NAMES and a necessary
symbol doesn't exist).
@end table
@end deftypefun

@node Operations, Mapping, Formatting, Top
@chapter Unit Operations
@cindex unit operations
@cindex operations, unit

You can use unit operations to construct new units, get information
about units, or compare units.

@menu
* Unary::       Operations on a single unit
* Binary::      Operations on pairs of units
@end menu

@node Unary, Binary, , Operations
@section Unary Unit Operations
@cindex unary unit operations

@anchor{ut_free()}
@deftypefun @code{void} ut_free @code{(ut_unit* @var{unit})}
Frees resources associated with @var{unit}.
You should invoke this function on every unit that you no longer need.
Use of @var{unit} upon
return from this function results in undefined behavior.
@end deftypefun

@anchor{ut_scale()}
@deftypefun @code{ut_unit*} ut_scale @code{(double @var{factor}, const ut_unit* @var{unit})}
Returns a unit equivalent to another unit scaled by a numeric factor.
For example:
@example
const ut_unit*   meter = ...
const ut_unit*   kilometer = ut_scale(1000, meter);
@end example
The returned unit is equivalent to @var{unit} multiplied by
@var{factor}.
You should pass the returned pointer to @code{@ref{ut_free()}} when you
no longer need the unit.
@end deftypefun

@anchor{ut_offset()}
@deftypefun @code{ut_unit*} ut_offset @code{(const ut_unit* @var{unit}, double @var{offset})}
Returns a unit equivalent to another unit relative to a particular
origin.
For example:
@example
const ut_unit*   kelvin = ...
const ut_unit*   celsius = ut_offset(kelvin, 273.15);
@end example
The returned unit is equivalent to @var{unit} with an origin of
@var{offset}.
You should pass the returned pointer to @code{@ref{ut_free()}} when you
no longer need the unit.
If an error occurs, then this function returns @code{NULL} and
@code{@ref{ut_get_status()}} will return one of the following:

@table @code
@item UT_BAD_ARG
@var{unit} is @code{NULL}.
@item UT_OS
Operating-system error.  See @code{errno} for the reason.
@end table
@end deftypefun

@anchor{ut_offset_by_time()}
@deftypefun @code{ut_unit*} ut_offset_by_time @code{(const ut_unit* const @var{unit}, const double @var{origin})}
Returns a timestamp-unit equivalent to the time unit @var{unit}
referenced to the time-origin @var{origin} (as returned by
@code{@ref{ut_encode_time()}}).
For example:
@example
const ut_unit*   second = ...
const ut_unit*   secondsSinceTheEpoch =
    ut_offset_by_time(second, ut_encode_time(1970, 1, 1, 0, 0, 0.0));
@end example
Leap seconds are not taken into account.
You should pass the returned pointer to @code{@ref{ut_free()}} when you
no longer need the unit.
If an error occurs, then this function returns @code{NULL} and
@code{@ref{ut_get_status()}} will return one of the following:

@table @code
@item UT_BAD_ARG
@var{unit} is @code{NULL}.
@item UT_OS
Operating-system error.  See @code{errno} for the reason.
@item UT_MEANINGLESS
Creation of a timestamp unit based on @var{unit} is not meaningful.  It
might not be a time-unit, for example.
@item UT_NO_SECOND
The associated unit-system doesn't contain a ``second'' unit.  See
@code{@ref{ut_set_second()}}.
@end table

@strong{CAUTION:}
The timestamp-unit was created to be analogous to, for example, the degree
celsius---but for the time dimension.
I've come to believe, however, that creating
such a unit was a mistake, primarily because users try to use the unit in
ways for which it was not designed (such as converting dates in a
calendar whose year is exactly 365 days long).
Such activities are much better handled by a dedicated calendar package.
Please be careful about using timestamp-units.
See also the section on @ref{Time, The Handling of Time}.

@end deftypefun

@anchor{ut_invert()}
@deftypefun @code{ut_unit*} ut_invert @code{(const ut_unit* @var{unit})}
Returns the inverse (i.e., reciprocal) of the unit @var{unit}.
This convenience function is equal to 
@code{@ref{ut_raise(),ut_raise(@var{unit}@comma{}-1)}}.
You should pass the returned pointer to @code{@ref{ut_free()}} when you
no longer need the unit.
If an error occurs,
then this function writes an error-message using
@code{@ref{ut_handle_error_message()}}
and returns @code{NULL}.
Also, @code{@ref{ut_get_status()}} will return one of the following:

@table @code
@item UT_BAD_ARG
@var{unit} is @code{NULL}.
@item UT_OS
Operating-system error. See @code{errno} for the reason.
@end table
@end deftypefun

@anchor{ut_raise()}
@deftypefun @code{ut_unit*} ut_raise @code{(const ut_unit* @var{unit}, int @var{power})}
Returns the unit equal to unit @var{unit} raised to the power @var{power}.
You should pass the returned pointer to @code{ut_free()} when you
no longer need the unit.
If an error occurs,
then this function writes an error-message using
@code{@ref{ut_handle_error_message()}}
and returns @code{NULL}.
Also, @code{@ref{ut_get_status()}} will return one of the following:

@table @code
@item UT_BAD_ARG
@var{unit} is @code{NULL}.
@item UT_OS
Operating-system error. See @code{errno} for the reason.
@end table
@end deftypefun

@anchor{ut_root()}
@deftypefun @code{ut_unit*} ut_root @code{(const ut_unit* @var{unit}, int @var{root})}
Returns the unit equal to the @var{root} root of unit @var{unit}.
You should pass the returned pointer to @code{ut_free()} when you
no longer need the unit.
If an error occurs,
then this function writes an error-message using
@code{@ref{ut_handle_error_message()}}
and returns @code{NULL}.
Also, @code{@ref{ut_get_status()}} will return one of the following:

@table @code
@item UT_BAD_ARG
@var{unit} is @code{NULL}.
@item UT_MEANINGLESS
It's meaningless to take the given root of the given unit.
This could be because the
resulting unit would have fractional (i.e., non-integral) dimensionality,
or because the unit is, for example, a logarithmic unit.
@item UT_OS
Operating-system error. See @code{errno} for the reason.
@end table
@end deftypefun

@anchor{ut_log()}
@deftypefun @code{ut_unit*} ut_log @code{(double @var{base}, const ut_unit* @var{reference})}
Returns the logarithmic unit corresponding to the logarithmic base
@var{base} and a reference level specified as the unit @var{reference}.
For example, the following creates a decibel unit with a
one milliwatt reference level:
@example
     const ut_unit* milliWatt = ...;
     const ut_unit* bel_1_mW = ut_log(10.0, milliWatt);

     if (bel_1_mW != NULL) @{
         const ut_unit* decibel_1_mW = @ref{ut_scale(),ut_scale}(0.1, bel_1_mW);

         @ref{ut_free(),ut_free}(bel_1_mW);   /* no longer needed */

         if (decibel_1_mW != NULL) @{
             /* Have decibel unit with 1 mW reference */
             ...
             @ref{ut_free(),ut_free}(decibel_1_mW);
         @}                 /* "decibel_1_mW" allocated */
     @}
@end example
You should pass the returned pointer to @code{ut_free()} when you
no longer need the unit.
If an error occurs,
then this function writes an error-message using
@code{@ref{ut_handle_error_message()}}
and returns @code{NULL}.
Also, @code{@ref{ut_get_status()}} will return one of the following:

@table @code
@item UT_BAD_ARG
@var{reference} is @code{NULL}.
@item UT_OS
Operating-system error. See @code{errno} for the reason.
@item UT_BAD_ARG
@var{base} is invalid (e.g., it must be greater than one).
@end table
@end deftypefun

@anchor{ut_get_name()}
@deftypefun @code{const char*} ut_get_name @code{(const ut_unit* @var{unit}, ut_encoding @var{encoding})}
Returns the name to which the unit referenced by @var{unit} maps in the
character-encoding specified by @var{encoding}.
If this function returns @code{NULL}, then
@code{@ref{ut_get_status()}} will return
one of the following:

@table @code
@item UT_BAD_ARG
@var{name} is @code{NULL}.
@item UT_SUCCESS
@var{unit} doesn't map to a name in the given character-set.
@end table
@end deftypefun

@anchor{ut_get_symbol()}
@deftypefun @code{const char*} ut_get_symbol @code{(const ut_unit* @var{unit}, ut_encoding @var{encoding})}
Returns the symbol to which the unit referenced by @var{unit} maps in the
character-encoding specified by @var{encoding}.
If this function returns @code{NULL}, then
@code{@ref{ut_get_status()}} will return
one of the following:

@table @code
@item UT_BAD_ARG
@var{symbol} is @code{NULL}.
@item UT_SUCCESS
@var{unit} doesn't map to a symbol in the given character-set.
@end table
@end deftypefun

@anchor{ut_get_system()}
@deftypefun @code{ut_system*} ut_get_system @code{(const ut_unit* @var{unit})}
Returns the unit-system to which the unit referenced by @var{unit} belongs.
If @var{unit} is @code{NULL}, then this function writes an error-message
using @code{@ref{ut_handle_error_message()}}
and returns @code{NULL}.
Also, @code{@ref{ut_get_status()}} will return @code{UT_BAD_ARG}.
@end deftypefun

@anchor{ut_is_dimensionless()}
@deftypefun @code{int} ut_is_dimensionless @code{(const ut_unit* @var{unit})}
Indicates if unit @var{unit} is dimensionless (like ``radian'').
This function returns a non-zero value if the unit is dimensionfull;
otherwise, @code{0} is returned and
@ref{ut_get_status()} will return one of the following:

@table @code
@item UT_BAD_ARG
@var{unit1} is @code{NULL}.
@item UT_SUCCESS
The unit is dimensionless.
@end table
@end deftypefun

@anchor{ut_clone()}
@deftypefun @code{ut_unit*} ut_clone @code{(const ut_unit* @var{unit})}
Returns a copy of the unit referenced by @var{unit}.
You should pass the returned pointer to @code{ut_free()} when you
no longer need the unit.
If an error occurs,
then this function writes an error-message using
@code{@ref{ut_handle_error_message()}}
and returns @code{NULL}.
Also, @code{@ref{ut_get_status()}} will return one of the following:

@table @code
@item UT_BAD_ARG
@var{unit} is @code{NULL}.
@item UT_OS
Operating-system failure.  See @code{errno}.
@end table
If you use @code{@ref{ut_read_xml()}}, then you should not normally
need to call this function.
@end deftypefun

@anchor{ut_accept_visitor()}
@deftypefun @code{@ref{ut_status}} ut_accept_visitor @code{(const ut_unit* @var{unit}, const @ref{ut_visitor,ut_visitor}* @var{visitor}, void* @var{arg})}
Accepts the visitor @var{visitor} to the unit @var{unit}.
The argument @var{arg} is passed to the visitor's functions.
This function returns one of the following:

@table @code
@item UT_BAD_ARG
@var{visitor} or @var{unit} is @code{NULL}.
@item UT_VISIT_ERROR
An error occurred in @var{visitor} while visiting @var{unit}.
@item UT_SUCCESS
Success.
@end table
@end deftypefun

@anchor{ut_visitor}
@deftp {Data type} {ut_visitor} {int foo(int)} {int bar(int, int)}
You pass a pointer to a data object of this type if and when you call
@code{@ref{ut_accept_visitor()}}.
It contains the following pointers to functions that implement your
unit-visitor:

@table @code
@item @ref{ut_status} (*visit_basic)(const ut_unit* @var{unit}, void* @var{arg}); 
Visits the basic-unit @var{unit}.  A basic-unit is a base unit like
``meter'' or a non-dimensional but named unit like ``radian''.
This function returns @code{@ref{ut_status,UT_SUCCESS}} on and only on success.
@item @ref{ut_status} (*visit_product)(const ut_unit* @var{unit}, int @var{count}, const ut_unit* const* @var{basicUnits}, const int* @var{powers}, void* @var{arg}); 
Visits the product-unit @var{unit}.
The product-unit is a product of the @var{count}
basic-units referenced by @var{basicUnits}, each raised to their respective,
non-zero power in @var{powers}.
This function returns @code{@ref{ut_status,UT_SUCCESS}} on and only on success.
@item @ref{ut_status} (*visit_galilean)(const ut_unit* @var{unit}, double @var{scale}, const ut_unit* @var{underlyingUnit}, double @var{origin}, void* arg); 
Visits the Galilean-unit @var{unit}.
The Galilean-unit has the underlying unit @var{underlyingUnit} and either the
non-unity scale factor @var{scale} or the non-zero origin @var{origin}, or both.
This function returns @code{@ref{ut_status,UT_SUCCESS}} on and only on success.
@item @ref{ut_status} (*visit_timestamp)(const ut_unit* @var{unit}, const ut_unit* @var{timeUnit}, double @var{origin}, void* @var{arg}); 
Visits the timestamp-unit @var{unit}.
The timestamp-unit has the underlying unit of time @var{timeUnit}
and the @code{@ref{ut_encode_time()}}-encoded time-origin @var{origin}.
This function returns @code{@ref{ut_status,UT_SUCCESS}} on and only on success.
@item @ref{ut_status} (*visit_logarithmic)(const ut_unit* @var{unit}, double @var{base}, const ut_unit* @var{reference}, void* @var{arg}); 
Visits the logarithmic-unit @var{unit}.
The logarithmic-unit has the logarithmic base @var{base} and
the reference-level is specified by the unit @var{reference}.
This function returns @code{@ref{ut_status,UT_SUCCESS}} on and only on success.
@end table
@end deftp

@node Binary, , Unary, Operations
@section Binary Unit Operations
@cindex binary unit operations

Binary unit operations act on two units.

@strong{NOTE:} The functions
@code{@ref{ut_are_convertible()}} and @code{@ref{ut_get_converter()}} are also
binary unit operations but are documented elsewhere.

@anchor{ut_multiply()}
@deftypefun @code{ut_unit*} ut_multiply @code{(const ut_unit* @var{unit1}, const ut_unit* @var{unit2})}
Returns the result of multiplying unit @var{unit1} by unit @var{unit2}.
You should pass the pointer to @ref{ut_free()} when you no longer need the unit
On failure, this function returns @code{NULL} and @ref{ut_get_status()}
will return one of the following:

@table @code
@item UT_BAD_ARG
@var{unit1} or @var{unit2} is @code{NULL}.
@item UT_NOT_SAME_SYSTEM
@var{unit1} and @var{unit2} belong to different @ref{unit-system}s.
@item UT_OS
Operating-system error. See @var{errno} for the reason.
@end table
@end deftypefun

@anchor{ut_divide}
@deftypefun @code{ut_unit*} ut_divide @code{(const ut_unit* @var{numer}, const ut_unit* @var{denom})}
Returns the result of dividing unit @var{numer} by unit @var{denom}.
You should pass the pointer to @ref{ut_free()} when you no longer need the unit
On failure, this function returns @code{NULL} and @ref{ut_get_status()} 
will return one of the following:

@table @code
@item UT_BAD_ARG
@var{numer} or @var{denom} is @code{NULL}.
@item UT_NOT_SAME_SYSTEM
@var{unit1} and @var{unit2} belong to different @ref{unit-system}s.
@item UT_OS
Operating-system error. See @code{errno} for the reason.
@end table
@end deftypefun

@anchor{ut_compare()}
@deftypefun @code{int} ut_compare @code{(const ut_unit* @var{unit1}, const ut_unit* @var{unit2})}
Compares two units.  Returns a value less than, equal to, or greater than
zero as @var{unit1} is considered less than, equal to, or greater than
@var{unit2}, respectively.
Units from different @ref{unit-system}s never compare equal.
The value zero is also returned if both unit pointers are @code{NULL}.
@end deftypefun

@anchor{ut_same_system()}
@deftypefun @code{int} ut_same_system @code{(const ut_unit* @var{unit1}, const ut_unit* @var{unit2})}
Indicates if two units belong to the same unit-system.
This function returns a non-zero value if the two units belong to the
same @ref{unit-system}; otherwise, @code{0} is returned and
@ref{ut_get_status()} will return one of the following:

@table @code
@item UT_BAD_ARG
@var{unit1} or @var{unit2} is @code{NULL}.
@item UT_SUCCESS
The units belong to different @ref{unit-system}s.
@end table
@end deftypefun

@node Mapping, Time, Operations, Top
@chapter Mapping Between Identifiers and Units
@cindex mapping units
@cindex mapping identifiers
@cindex units, mapping to identifiers

Within a unit-system, you can map an identifier to a unit and vice
versa.
If an identifier maps to a unit, then the unit can be retrieved from the
unit-system via the identifier.
Similarly, if a unit maps to an identifier, then the unit can be printed
using the identifier.

There a two kinds of identifiers: names and symbols.

@menu
* Names::       Mapping between units and names.
* Symbols::     Mapping between units and symbols.
@end menu

@node Names, Symbols, , Mapping
@section Names
@cindex names

You can map a name to a unit and vice versa.  If you use
@code{@ref{ut_read_xml()}}, then you shouldn't normally need to do this.

@anchor{ut_map_name_to_unit()}
@deftypefun @code{@ref{ut_status}} ut_map_name_to_unit @code{(const char* @var{name}, const ut_encoding @var{encoding}, const ut_unit* @var{unit})}
Maps the name referenced by @var{name}, in character-set @var{encoding},
to the unit referenced by @var{unit}
in the unit-system that contains @var{unit}.
This function returns one of the following:

@table @code
@item UT_SUCCESS
Success.
@item UT_BAD_ARG
@var{name} or @var{unit} is @code{NULL}.
@item UT_OS
Operating-system failure.  See @code{errno}.
@item UT_EXISTS
@var{name} already maps to a different unit.
@end table
@end deftypefun

@anchor{ut_unmap_name_to_unit()}
@deftypefun @code{@ref{ut_status}} ut_unmap_name_to_unit @code{(ut_system* @var{system}, const char* @var{name}, const ut_encoding @var{encoding})}
Removes any mapping from name @var{name}, in character-set @var{encoding}, to a unit in unit-system @var{system}.
This function returns one of the following:

@table @code
@item UT_SUCCESS
Success.
@item UT_BAD_ARG
@var{system} or @var{name} is @code{NULL}.
@end table
@end deftypefun

@anchor{ut_map_unit_to_name()}
@deftypefun @code{@ref{ut_status}} ut_map_unit_to_name @code{(const ut_unit* @var{unit}, const char* @var{name}, ut_encoding @var{encoding})}
Maps the unit @var{unit} to the name @var{name}, which is in character-set
@var{encoding}, in the unit-system that contains the unit.
This function returns one of the following:

@table @code
@item UT_SUCCESS
Success.
@item UT_BAD_ARG
@var{unit} or @var{name} is @code{NULL}, or @var{name} is not in the
character-set @var{encoding}.
@item UT_OS
Operating-system failure.  See @code{errno}.
@item UT_EXISTS
@var{unit} already maps to a different name.
@end table
@end deftypefun

@anchor{ut_unmap_unit_to_name()}
@deftypefun @code{@ref{ut_status}} ut_unmap_unit_to_name @code{(const ut_unit* @var{unit}, ut_encoding @var{encoding})}
Removes any mapping from unit @var{unit} to a name in character-set
@var{encoding} from the unit-system that contains the unit.
This function returns one of the following:

@table @code
@item UT_SUCCESS
Success.
@item UT_BAD_ARG
@var{unit} is @code{NULL}.
@end table
@end deftypefun

@node Symbols, , Names, Mapping
@section Symbols
@cindex symbols

You can map a symbol to a unit and vice versa.  If you use
@code{@ref{ut_read_xml()}}, then you shouldn't normally need to do this.

@anchor{ut_map_symbol_to_unit()}
@deftypefun @code{@ref{ut_status}} ut_map_symbol_to_unit @code{(const char* @var{symbol}, const ut_encoding @var{encoding}, const ut_unit* @var{unit})}
Maps the symbol referenced by @var{symbol}, in character-set @var{encoding},
to the unit referenced by @var{unit}
in the unit-system that contains @var{unit}.
This function returns one of the following:

@table @code
@item UT_SUCCESS
Success.
@item UT_BAD_ARG
@var{symbol} or @var{unit} is @code{NULL}.
@item UT_OS
Operating-system failure.  See @code{errno}.
@item UT_EXISTS
@var{symbol} already maps to a different unit.
@end table
@end deftypefun

@anchor{ut_unmap_symbol_to_unit()}
@deftypefun @code{@ref{ut_status}} ut_unmap_symbol_to_unit @code{(ut_system* @var{system}, const char* @var{symbol}, const ut_encoding @var{encoding})}
Removes any mapping from symbol @var{symbol}, in character-set @var{encoding}, to a unit in unit-system @var{system}.
This function returns one of the following:

@table @code
@item UT_SUCCESS
Success.
@item UT_BAD_ARG
@var{system} or @var{symbol} is @code{NULL}.
@end table
@end deftypefun

@anchor{ut_map_unit_to_symbol()}
@deftypefun @code{@ref{ut_status}} ut_map_unit_to_symbol @code{(const ut_unit* @var{unit}, const char* @var{symbol}, ut_encoding @var{encoding})}
Maps the unit @var{unit} to the symbol @var{symbol}, which is in character-set
@var{encoding}, in the unit-system that contains the unit.
This function returns one of the following:

@table @code
@item UT_SUCCESS
Success.
@item UT_BAD_ARG
@var{unit} or @var{symbol} is @code{NULL}.
@item UT_BAD_ARG
Symbol @var{symbol} is not in the character-set @var{encoding}.
@item UT_OS
Operating-system failure.  See @code{errno}.
@item UT_EXISTS
@var{unit} already maps to a different symbol.
@end table
@end deftypefun

@anchor{ut_unmap_unit_to_symbol()}
@deftypefun @code{@ref{ut_status}} ut_unmap_unit_to_symbol @code{(const ut_unit* @var{unit}, ut_encoding @var{encoding})}
Removes any mapping from unit @var{unit} to a symbol in character-set
@var{encoding} from the unit-system that contains the unit.
This function returns one of the following:

@table @code
@item UT_SUCCESS
Success.
@item UT_BAD_ARG
@var{unit} is @code{NULL}.
@end table
@end deftypefun

@node Time, Errors, Mapping, Top
@chapter The Handling of Time
@cindex time, handling of

You should use a true calendar package rather than the UDUNITS-2 package 
to handle time.  Having said that, many people use the time-handling
capabilities of the UDUNITS-2 package because it supports "units" like
"@code{seconds since 1970-01-01}".  You should be aware, however, that the
hybrid Gregorian/Julian calendar used by the UDUNITS-2 package @emph{cannot be
changed}.  Dates on or after 1582-10-15 are assumed to be Gregorian dates;
dates before that are assumed to be Julian dates.  In particular,
the year 1 BCE is immediately followed by the year 1 CE.

In general, the UDUNITS-2 package handles time by encoding it as
double-precision value, which can then be acted upon arithmetically.

@anchor{ut_encode_time()}
@deftypefun @code{double} ut_encode_time @code{(int @var{year}, int @var{month}, int @var{day}, int @var{hour}, int @var{minute}, double @var{second})}
Encodes a time as a double-precision value.
This convenience function is equivalent to
@example
@ref{ut_encode_date(),ut_encode_date}(@var{year},@var{month},@var{day}) + @ref{ut_encode_clock(),ut_encode_clock}(@var{hour},@var{minute},@var{second})
@end example
@end deftypefun

@anchor{ut_encode_date()}
@deftypefun @code{double} ut_encode_date @code{(int @var{year}, int @var{month}, int @var{day})}
Encodes a date as a double-precision value.
You probably won't use this function.
Dates on or after 1582-10-15 are assumed to be Gregorian dates;
dates before that are assumed to be Julian dates.  In particular,
the year 1 BCE is immediately followed by the year 1 CE.
@end deftypefun

@anchor{ut_encode_clock()}
@deftypefun @code{double} ut_encode_clock @code{(int @var{hour}, int @var{minute}, double @var{second})}
Encodes a clock-time as a double-precision value.
You probably won't use this function.
@end deftypefun

@anchor{ut_decode_time()}
@deftypefun @code{void} ut_decode_time @code{(double @var{time}, int* @var{year}, int* @var{month}, int* @var{day}, int* @var{hour}, int* @var{minute}, double* @var{second}, double* @var{resolution})}
Decodes a time from a double-precision value into its individual components.
The variable referenced by @var{resolution} will be set to the resolution
(i.e., uncertainty) of the time in seconds.
@end deftypefun

@node Errors, Database, Time, Top
@chapter Error Handling
@cindex error handling

Error-handling in the units module has two aspects: the status of
the last operation performed by the module and the handling of error-messages:

@menu
* Status::      The status of the last operation.
* Messages::    The handling of error-messages.
@end menu

@node Status, Messages, , Errors
@section Status of Last Operation
@cindex status of last operation
@cindex module status

UDUNITS-2 functions set their status by calling @code{@ref{ut_set_status()}}.
You can use the function @code{@ref{ut_get_status()}} to retrieve that
status.

@anchor{ut_get_status()}
@deftypefun @code{@ref{ut_status}} ut_get_status @code{(void)}
Returns the value specified in the last call to
@code{@ref{ut_set_status()}}
@end deftypefun

@anchor{ut_set_status()}
@deftypefun @code{void} ut_set_status @code{(@ref{ut_status} @var{status})}
Set the status of the units module to @var{status}.
@end deftypefun

@anchor{ut_status}
@deftp {Data type} {ut_status}
This enumeration has the following values:

@table @code
@item UT_SUCCESS
Success
@item UT_BAD_ARG
An argument violates the the function's contract (e.g., it's @code{NULL}).
@item UT_EXISTS
Unit, prefix, or identifier already exists
@item UT_NO_UNIT
No such unit exists
@item UT_OS
Operating-system error.  See @code{errno} for the reason.
@item UT_NOT_SAME_SYSTEM
The units belong to different unit-systems
@item UT_MEANINGLESS
The operation on the unit or units is meaningless
@item UT_NO_SECOND
The unit-system doesn't have a unit named ``second''
@item UT_VISIT_ERROR
An error occurred while visiting a unit
@item UT_CANT_FORMAT
A unit can't be formatted in the desired manner
@item UT_SYNTAX
String unit representation contains syntax error
@item UT_UNKNOWN
String unit representation contains unknown word
@item UT_OPEN_ARG
Can't open argument-specified unit database
@item UT_OPEN_ENV
Can't open environment-specified unit database
@item UT_OPEN_DEFAULT
Can't open installed, default, unit database
@item UT_PARSE
Error parsing unit database
@end table
@end deftp

@node Messages, , Status, Errors
@section Error-Messages
@cindex messages, error
@cindex error-messages

@anchor{ut_handle_error_message()}
@deftypefun @code{int} ut_handle_error_message @code{(const char* @var{fmt}, ...)}
Handles the error-message corresponding to the format-string @var{fmt} and
any subsequent arguments referenced by it.
The interpretation of the formatting-string is identical to that of
the UNIX function @code{printf()}.
On success, this function returns the number of bytes in the
error-message; otherwise, this function returns @code{-1}.

Use the function @code{@ref{ut_set_error_message_handler()}} to change how
error-messages are handled.
@end deftypefun

@anchor{ut_set_error_message_handler()}
@deftypefun @code{@ref{ut_error_message_handler}} ut_set_error_message_handler @code{(@ref{ut_error_message_handler} @var{handler})}
Sets the function that handles error-messages and returns the previous
error-message handler.
The initial error-message handler is @code{@ref{ut_write_to_stderr()}}.
@end deftypefun

@anchor{ut_write_to_stderr()}
@deftypefun @code{int} ut_write_to_stderr @code{(const char* @var{fmt}, va_list @var{args})}
Writes the variadic error-message
corresponding to formatting-string @var{fmt} and arguments @var{args}
to the standard-error stream and appends a newline.
The interpretation of the formatting-string is identical to that of
the UNIX function @code{printf()}.
On success, this function returns the number of bytes in the
error-message; otherwise, this function returns @code{-1}.
@end deftypefun

@anchor{ut_ignore()}
@deftypefun @code{int} ut_ignore @code{(const char* @var{fmt}, va_list @var{args})}
Does nothing.
In particular, it ignores the variadic error-message
corresponding to formatting-string @var{fmt} and arguments @var{args}.
Pass this function to @code{@ref{ut_set_error_message_handler()}} 
when you don't want the unit module to print any error-messages.
@end deftypefun

@anchor{ut_error_message_handler}
@deftp {Data type} {ut_error_message_handler}
This is the type of an error-message handler.
It's definition is
@example
typedef int (*ut_error_message_handler)(const char* fmt, va_list args);
@end example
@end deftp

@node Database, Types, Errors, Top
@chapter The Units Database
@cindex units database
@cindex database, units

The database of units that comes with the UDUNITS-2 package is an
XML-formatted file that is based on the SI system of units.
It contains the names and symbols of most of the units that you will ever
encounter.
The pathname of the installed file is
@code{@emph{datadir}/udunits2.xml}, where @emph{datadir} is the
installation-directory for read-only, architecture-independent data
(e.g., @code{/usr/local/share}).
This pathname is the default that @code{@ref{ut_read_xml()}} uses.

Naturally, because the database is a regular file, it can be edited
to add new units or remove existing ones.
Be very careful about doing this, however, because you might lose
the benefit of exchanging unit-based information with others who
haven't modified their database.

@node Types, Complete Index, Database, Top
@chapter Data Types
@cindex data types
@cindex types, data

The data types @code{@ref{ut_visitor}}, @code{@ref{ut_status}}, and
@code{@ref{ut_error_message_handler}} are documented elsewhere.

@anchor{ut_encoding}
@deftp {Data type} {ut_encoding}
This enumeration has the following values:

@table @code
@item UT_ASCII
@uref{http://en.wikipedia.org/wiki/Ascii,US ASCII} character-set.
@item UT_ISO_8859_1
The @uref{http://en.wikipedia.org/wiki/Iso-8859-1,ISO-8859-1} character-set.
@item UT_LATIN1
Synonym for @code{UT_ISO_8859_1}.
@item UT_UTF8
The @uref{http://en.wikipedia.org/wiki/Utf-8,UTF-8} encoding of the Unicode
character-set.
@end table
@end deftp

@node Complete Index, , Types, Top
@unnumbered Index

@printindex cp

@bye