Variadic function: Difference between revisions

Revision as of 16:05, 26 December 2022

Latest revision as of 12:16, 25 July 2025

(40 intermediate revisions by 27 users not shown)

Line 1:

{{Short description|Function with variable number of arguments}}

{{Redirect|Varargs|the varargs.h library in C|varargs.h}}

In [[mathematics]] and in [[computer programming]], a '''variadic function''' is a [[function (programming)|function]] of indefinite [[arity]], i.e., one which accepts a variable number of [[argument (computer science)|argument]]s. Support for variadic functions differs widely among [[programming language]]s.

The term ''variadic'' is a [[neologism]], dating back to 1936–19371936/1937.<ref>Henry S. Leonard and H. N. Goodman, ''A calculus of individuals''. Abstract of a talk given at the Second Meeting of the Association for Symbolic Logic, held in Cambridge MA on December 28–30, 1936, [https://www.jstor.org/stable/2268861?seq=1#metadata_info_tab_contents], ''Journal of Symbolic Logic'' '''2'''(1) 1937, 63.</ref> The term was not widely used until the 1970s.

==Overview==

Line 12 ⟶ 13:

Variadic functions can expose [[Type safety|type-safety]] problems in some languages. For instance, C's {{code|printf}}, if used incautiously, can give rise to a class of security holes known as [[format string attack]]s. The attack is possible because the language support for variadic functions is not type-safe: it permits the function to attempt to pop more arguments off the [[Stack (abstract data type)#Hardware stacks|stack]] than were placed there, corrupting the stack and leading to unexpected behavior. As a consequence of this, the [[CERT Coordination Center]] considers variadic functions in C to be a high-severity security risk.<ref>{{cite book|last=Klemens|first=Ben|title=21st Century C: C Tips from the New School|publisher=O'Reilly Media, Inc.|date=2014|pages=224|isbn=978-1491904442}}</ref>

In [[functional programming]] languages, variadics can be considered complementary to the [[apply]] function, which takes a function and a list/sequence/array as arguments, and calls the function with the arguments supplied in that list, thus passing a variable number of arguments to the function.{{citation needed|date=February 2018}} In the functional language [[Haskell (programming language)|Haskell]], variadic functions can be implemented by returning a value of a [[type class]] {{code|T}}; if instances of {{code|T}} are a final return value {{code|r}} and a function {{code|1=(T t) => x -> t}}, this allows for any number of additional arguments {{code|x}}.{{Explain|reason=Either this should be rephrased in a simpler way, or better explained. It is not immediately understandable, especially for readers not knowing Haskell, which is likelymost thereaders, vasteven majorityfor ofthis readersarticle.|date=May 2018}}

A related subject in [[term rewriting]] research is called '''hedges''', or '''hedge variables'''.<ref>[https://arxiv.org/abs/1503.00336 CLP (H): Constraint Logic Programming for Hedges]</ref> Unlike variadics, which are functions with arguments, hedges are sequences of arguments themselves. They also can have constraints ('take no more than 4 arguments', for example) to the point where they are not variable-length (such as 'take exactly 4 arguments') - thus calling them ''variadics'' can be misleading. However they are referring to the same phenomenon, and sometimes the phrasing is mixed, resulting in names such as ''variadic variable'' (synonymous to hedge). Note the double meaning of the word ''variable'' and the difference between arguments and variables in functional programming and term rewriting. For example, a term (function) can have three variables, one of them a hedge, thus allowing the term to take three or more arguments (or two or more if the hedge is allowed to be empty).

Line 19 ⟶ 20:

===In C===

To portably implement variadic functions in the [[C (programming language)|C programming language]], the standard [[stdarg.h|{{code|stdarg.h}}]] header file is used. The older [[varargs.h|{{code|varargs.h}}]] header has been [[Deprecation|deprecated]] in favor of {{code|stdarg.h}}. In C++, the header file {{code|cstdarg}} is used.<ref>{{cite web|url=http://www.cplusplus.com/reference/clibrary/cstdarg/|title=<cstdarg> (stdarg.h) - C++ Reference|website=www.cplusplus.com|access-date=2007-10-02|archive-date=2012-10-31|archive-url=https://web.archive.org/web/20121031160547/http://www.cplusplus.com/reference/clibrary/cstdarg/|url-status=dead}}</ref>

<syntaxhighlight lang="C">

Line 45 ⟶ 46:

</syntaxhighlight>

This will compute the average of an arbitrary number of arguments. Note that the function does not know the number of arguments or their types. The above function expects that the types will be {{code|int}}, and that the number of arguments is passed in the first argument (this is a frequent usage but by no means enforced by the language or compiler). In some other cases, for example [[printf]], the number and types of arguments are figured out from a format string. In both cases, this depends on the programmer to supply the correct information. (Alternatively, a [[sentinel value]] like {{code|NULL}} or {{code|nullptr}} may be used to indicate the numberend of the parameter list.) If fewer arguments are passed in than the function believes, or the types of arguments are incorrect, this could cause it to read into invalid areas of memory and can lead to vulnerabilities like the [[format string attack]]. Depending on the system, even using {{code|NULL}} as a sentinel may encounter such problems; {{code|nullptr}} or a dedicated null pointer of the correct target type may be used to avoid them.

{{code|stdarg.h}} declares a type, {{code|va_list}}, and defines four macros: [[va_startva start|{{code|va_start}}]], [[va_argva arg|{{code|va_arg}}]], [[va_copyva copy|{{code|va_copy}}]], and [[va_endva end|{{code|va_end}}]]. Each invocation of {{code|va_start}} and {{code|va_copy}} must be matched by a corresponding invocation of {{code|va_end}}. When working with variable arguments, a function normally declares a variable of type {{code|va_list}} ({{code|ap}} in the example) that will be manipulated by the macros.

# {{code|va_start}} takes two arguments, a {{code|va_list}} object and a reference to the function's last parameter (the one before the ellipsis; the macro uses this to get its bearings). In [[C2xC23 (C standard revision)|C23]], the second argument will no longer be required and variadic functions will no longer need a named parameter before the ellipsis.{{r| g=note|n=KandR|r=Making the named parameter optional was needed since there was no way to specify a function taking an unspecified number of arguments in C23 after the removal of K&R style function definitions. Since C++ was already using this syntax for the same purpose, this change was also a way to increase compatibility between the languages.{{r|r=<ref>{{cite web|url=https://thephd.dev/c23-is-coming-here-is-what-is-on-the-menu#n2975---relax-requirements-for-variadic-parameter-lists|title=C23 is Finished: Here is What is on the Menu# §N2975 - Relax requirements for variadic parameter lists|date=31 July 2022 }}}}</ref>}}<ref name=N2975>{{cite web|url=https://open-std.org/JTC1/SC22/WG14/www/docs/n2975.pdf|title=WG14-N2975 : Relax requirements for variadic parameter lists, v3|date=2022-04-15|last1=Gilding|first1=Alex|last2=Meneide|first2=JeanHeyd}}</ref> It initialises the {{code|va_list}} object for use by {{code|va_arg}} or {{code|va_copy}}. The compiler will normally issue a warning if the reference is incorrect (e.g. a reference to a different parameter than the last one, or a reference to a wholly different object), but will not prevent compilation from completing normally.

# {{code|va_arg}} takes two arguments, a {{code|va_list}} object (previously initialised) and a type descriptor. It expands to the next variable argument, and has the specified type. Successive invocations of {{code|va_arg}} allow processing each of the variable arguments in turn. Unspecified behavior occurs if the type is incorrect or there is no next variable argument.

# {{code|va_end}} takes one argument, a {{code|va_list}} object. It serves to clean up. If one wanted to, for instance, scan the variable arguments more than once, the programmer would re-initialise your {{code|va_list}} object by invoking {{code|va_end}} and then {{code|va_start}} again on it.

# {{code|va_copy}} takes two arguments, both of them {{code|va_list}} objects. It clones the second (which must have been initialised) into the first. Going back to the "scan the variable arguments more than once" example, this could be achieved by invoking {{code|va_start}} on a first {{code|va_list}}, then using {{code|va_copy}} to clone it into a second {{code|va_list}}. After scanning the variable arguments a first time with {{code|va_arg}} and the first {{code|va_list}} (disposing of it with {{code|va_end}}), the programmer could scan the variable arguments a second time with {{code|va_arg}} and the second {{code|va_list}}. Don't forget to {{code|va_end}} needs to also be called on the cloned {{code|va_list}} before the containing function returns.

===In C#===

Line 82 ⟶ 83:

===In C++===

The basic variadic facility in C++ is largely identical to that in C. The only difference is in the syntax, where the comma before the ellipsis can be omitted. C++ allows variadic functions without [[named parametersparameter]]s but provides no way to access those arguments since <code>va_start</code> requires the name of the last fixed argument of the function. <!-- When C23 is released the text should be updated to reflect that this is a difference between the languages. When C++26 ports the one argument va_start to C++ (https://wg21.link/p2537r1p2537) this should be updated again. See also the C section of the article. -->

<syntaxhighlight lang="c++">

Line 137 ⟶ 138:

The [[CERT Coding Standards]] for C++ strongly prefers the use of [[variadic templates]] (parameter pack) in C++ over the C-style variadic function due to a lower risk of misuse.<ref>{{cite web |title=DCL50-CPP. Do not define a C-style variadic function |url=https://wiki.sei.cmu.edu/confluence/display/cplusplus/DCL50-CPP}}</ref>

=== In Fortran ===

Since the Fortran 90 revision, [[Fortran]] functions or subroutines can accept optional arguments:<ref>{{Cite web |title=Optional Arguments |url=https://www.intel.com/content/www/us/en/docs/fortran-compiler/developer-guide-reference/2023-0/optional-arguments.html |access-date=2025-03-18 |website=Intel |language=en}}</ref> the argument list is still fixed, but the ones that have the {{code|optional}} attribute can be omitted in the function/subroutine call. The intrinsic function {{code|present()}} can be used to detect the presence of an optional argument. The optional arguments can appear anywhere in the argument list.

<syntaxhighlight lang="fortran">program test

implicit none

real :: x

!> all arguments are passed:

call foo( 1, 2, 3.0, 4, x )

!< outputs 1 \ 2 \ 3.0 \ 4 \ 6.0 (the "\" denotes a newline)

!> the last 2 arguments are omitted:

call foo( 1, 2, 3.0 )

!< outputs 1 \ 2 \ 3.0

!> the 2nd and 4th arguments are omitted: the arguments that are positioned after

!> an omitted argument must be passed with a keyword:

call foo( 1, c=3.0, e=x )

!< outputs 1 \ 3.0 \ 6.0

!> alternatively, the Fortran 2023 revision has introduced the .NIL. pseudo constant

!> to denote an omitted argument

call foo( 1, .NIL., 3.0, .NIL., x )

!< outputs 1 \ 3.0 \ 6.0

contains

!> the subroutine foo() has 2 mandatory and 3 optional arguments

subroutine foo( a, b, c, d, e )

integer, intent(in) :: a

integer, intent(in), optional :: b

real, intent(in) :: c

integer, intent(in), optional :: d

real, intent(out), optional :: e

print*, a

if (present(b)) print*, b

print*, c

if (present(d)) print*, d

if (present(e)) then

e = 2*c

print*, c

end if

end subroutine

end program</syntaxhighlight>

'''Output:'''

<pre>

The sum of [1 2] is 3

The sum of [1 2 3] is 6

The sum of [1 2 3 4] is 10

</pre>

===In Go===

Line 223 ⟶ 280:

</syntaxhighlight>

=== In [[PascalLua (programming language)|PascalLua]] ===

[[Lua (programming language)|Lua]] functions may pass varargs to other functions the same way as other values using the {{Code|return}} keyword. tables can be passed into variadic functions by using, in Lua version 5.2 or higher<ref>{{Cite web |title=Lua 5.2 Reference Manual |url=https://www.lua.org/manual/5.2/manual.html#pdf-table.unpack |access-date=2023-02-05 |website=www.lua.org}}</ref> {{Code|table.unpack}}, or Lua 5.1 or lower<ref>{{Cite web |title=Lua 5.1 Reference Manual |url=https://www.lua.org/manual/5.1/manual.html#pdf-unpack |access-date=2023-02-05 |website=www.lua.org}}</ref> {{Code|unpack}}. Varargs can be used as a table by constructing a table with the vararg as a value.<syntaxhighlight lang="lua">

{{cleanup section|reason=Default and polymorphism is not variadic. I/O example is too long.|date=April 2021}}

function sum(...) --... designates varargs

[[Pascal (programming language)|Pascal]] has four built-in procedures which are defined as variadic, which, because of this special condition, are intrinsic to the compiler. These are the {{code|read}}, {{code|readln}}, {{code|write}}, and {{code|writeln}} procedures. However, there are alternate specifications allowing for [[default argument|''default'' arguments]] to procedures or functions which make them work variadically, as well as ''[[polymorphism (computer science)|polymorphism]]'' which allows a procedure or function to have different parameters.

local sum=0

for _,v in pairs({...}) do --creating a table with a varargs is the same as creating one with standard values

sum=sum+v

end

return sum

end

values={1,2,3,4}

The {{code|read[ln]}} and {{code|write[ln]}} procedures all have the same format:

sum(5,table.unpack(values)) --returns 15. table.unpack should go after any other arguments, otherwise not all values will be passed into the function.

function add5(...)

<pre>

return ...+5 --this is incorrect usage of varargs, and will only return the first value provided

read[ln] [( [file ,] variable [, variable ...] )] ;

end

write[ln] [( [file][, value [, value ...] )] ;

</pre>

entries={}

where

function process_entries()

local processed={}

* {{code|file}} is an optional file variable, which if omitted, defaults to {{code|input}} for {{code|read}} and {{code|readln}}, or defaults to {{code|output}} for {{code|write}} and {{code|writeln}};

for i,v in pairs(entries) do

* {{code|variable}} is a scalar such as a char (character), integer, or real (or for some compilers, certain record types or array types such as strings); and

processed[i]=v --placeholder processing code

* {{code|value}} is a variable or a constant.

end

return table.unpack(processed) --returns all entries in a way that can be used as a vararg

Example:

end

<syntaxhighlight lang="pascal" line>

var

f: text;

ch: char;

n,a,I,B: Integer;

S: String;

print(process_entries()) --the print function takes all varargs and writes them to stdout separated by newlines

begin

Write('Enter name of file to write results: ');

readln(s);

assign(f,S);

rewrite(f);

Write('What is your name? ');

readln(Input,S);

Write('Hello, ',S,'! Enter the number of calculations you want to do:');

writeln(output);

Write('? ');

readln(N);

Write('For each of the ',n,' formulas, enter ');

write('two integers separated by one or more spaces');

writeln;

for i := 1 to N do

begin

Write('Enter pair #',i,'? ');

read(a,b);

READLN;

WRITELN(Out,'A [',a,'] + B [',B,'] =',A+B);

end;

close(OUT);

end.

</syntaxhighlight>

===In Pascal===

In the above example, as far as the compiler is concerned, lines 9 and 13 are identical, because if {{code|input}} is the file variable being read into by a {{code|read}} or {{code|readln}} statement, the file variable may be omitted. Also, the compiler considers lines 15 and 20 to be identical, because if the file variable being written to is {{code|output}}, it can be omitted, which means (on line 20) since there are no arguments being passed to the procedure the parentheses listing arguments can be omitted. Line 26 shows the {{code|writeln}} statement can have any number of arguments, and they can be a quoted string, a variable, or even a formula result.

[[Pascal (programming language)|Pascal]] is standardized by [[International Organization for Standardization|ISO]] standards 7185 (“Standard Pascal”) and 10206 (“Extended Pascal”).

Neither standardized form of Pascal supports variadic routines, ''except'' for certain [[Intrinsic function|built-in routines]] ({{code|read|pascal}}/{{code|readLn|pascal}} and {{code|write|pascal}}/{{code|writeLn|pascal}}, and additionally in {{abbr|EP|Extended Pascal}} {{code|readStr|pascal}}/{{code|writeStr|pascal}}).

Nonetheless, ''dialects'' of Pascal implement mechanisms ''resembling'' variadic routines.

Object Pascal supports ''polymorphic'' procedures and functions, where different procedure(s) or function(s) can have the same name but are distinguished by the arguments supplied to them.

[[Delphi (programming language)|Delphi]] defines an {{code|array of const|delphi}} data type that may be associated with the ''last'' [[formal parameter]].

Within the routine definition the {{code|array of const|delphi}} is an {{code|array of TVarRec|delphi}}, an [[Array (data type)|array]] of [[variant record]]s.<ref name="delphi">{{cite web|url=https://docwiki.embarcadero.com/RADStudio/Alexandria/en/Parameters_(Delphi)#Variant_Open_Array_Parameters|title=Parameters (Delphi)|access-date=2023-08-28}}</ref>

The {{code|VType|delphi}} member of the aforementioned {{code|record|delphi}} data type allows inspection of the argument’s data type and subsequent appropriate handling.

The [[Free Pascal Compiler]] supports Delphi’s variadic routines, too.<ref name="fpc">{{cite web|url=https://www.freepascal.org/docs-html/3.2.0/ref/refsu68.html|title=Free Pascal - Reference guide|access-date=2023-08-28}}</ref>

This implementation, however, technically requires a ''single'' argument, that is an {{code|array|pascal}}.

Pascal also supports ''default'' arguments, where the value of an argument, if not provided, is given a default value.

Pascal imposes the restriction that arrays need to be homogenous.

This requirement is circumvented by utilizing a variant record.

For the first example, polymorphism, consider the following:

The [[GNU Pascal]] defines a real variadic formal parameter specification using an ellipsis ({{code|...|pascal}}), but as of 2022 no portable mechanism to use such has been defined.<ref name="gpc">{{cite web|url=https://www.gnu-pascal.de/gpc/Special-Parameters.html|title=The GNU Pascal Manual|access-date=2023-08-28}}</ref>

<syntaxhighlight lang="pascal" line>

function add(a1,a2:integer):Integer; begin add := a1+a2 end;

function add(r1,r2:real):real; begin add := a1+a2 end;

function add(a1:integer;r2:real):real; begin add := real(a1)+a2 end;

function add(r1:real,a2:integer):real; begin add := a1+real(a2) end;

</syntaxhighlight>

In the above example, if {{code|add}} as called with two integer values, the function declared on line 1 would be called; if one of the arguments is an integer and one is real, either the function on line 3 or 4 is called depending on which is integer. If both are real, the function on line 2 is called.

For default parameters, consider the following:

<syntaxhighlight lang="pascal" line>

const

Three = 3;

var

K: Integer;

function add(i1: integer = 0;

i2: integer = 0;

i3: integer = 0;

i4: integer = 0;

i5: integer = 0;

i6: integer = 0;

i7: integer = 0;

i8: integer = 0): integer;

begin

add := i1+i2+i3+I4+I5+i6+I7+I8;

end;

begin

K := add; { K is 0}

K := add(K,1); { K is 1}

K := add(1,2); { K is 3}

K := add(1,2,Three); { K is 6, etc.}

end.

</syntaxhighlight>

Both GNU Pascal and FreePascal allow externally declared functions to use a variadic formal parameter specification using an ellipsis ({{code|...|pascal}}).

On Line 6, (and the lines below) the parameter {{code|1== 0}} tells the compiler, "if no argument is provided, presume the argument to be zero." On line 19, no arguments were given, so the function returns {{code|0}}. On line 20, either a number or a variable can be supplied for any argument, and as shown on line 22, a constant.

===In [[PHP]]===

[[PHP]] does not care about types of variadic arguments unless the argument is typed.

<syntaxhighlight lang="php">

Line 376 ⟶ 381:

====Unflattened slurpy====

These parameters are declared with two asterisks (<code>**</code>) and they do not flatten any iterable arguments within the list, but keep the arguments more or less as-is:

<syntaxhighlight lang="perl6">

Line 400 ⟶ 405:

zaz(1, 2, 3); # [3]

zaz(1, 2, 3, "hello"); # [3 "hello"]

zaz(1, 2, [4, 5]); # [4, 5], single argurmentargument fills up array

zaz(1, 2, 3, [4, 5]); # [3, [4, 5]], behaving as **@

zaz(1, 2, 3, [4, 5], [6]); # [3, [4, 5], [6]], behaving as **@

Line 526 ⟶ 531:

==External links==

* [http://rosettacode.org/wiki/Variadic_function Variadic function]. [[Rosetta Code]] task showing the implementation of variadic functions in over fifty120 programming languages.

* [https://web.archive.org/web/20070927215504/http://www.codeproject.com/cpp/argfunctions.asp?df=100&forumid=15556&exp=0&select=503481 Variable Argument Functions] — A tutorial on Variable Argument Functions for C++

* [https://www.gnu.org/software/hello/manual/libc/Variadic-Functions.html GNU libc manual]

[[Category:Subroutines]]

[[Category:Programming language comparisons]]

<!-- Hidden categories below -->

[[Category:Articles with example C code]]

[[Category:Articles with example C++ code]]

[[Category:Articles with example C Sharp code]]

[[Category:Articles with example Haskell code]]

[[Category:Articles with example Java code]]

[[Category:Articles with example JavaScript code]]

[[Category:Articles with example Pascal code]]

[[Category:Articles with example Perl code]]

[[Category:Articles with example Python (programming language) code]]

[[Category:Articles with example Ruby code]]

[[Category:Articles with example Rust code]]

[[Category:Articles with example Swift code]]

[[Category:Articles with example Tcl code]]