Difference between revisions of "Unit"

From Free Pascal wiki
Jump to navigationJump to search
(→‎Read more: "See also" is used much more in this and other wikis than "Read more" Fixed hyperlink.)
Line 124: Line 124:
 
If you want to have different behaviour, you can explicity prepend ''unitname.''identifier to specify the unit you want to use, or reorder the units. The former is often the clearest option.
 
If you want to have different behaviour, you can explicity prepend ''unitname.''identifier to specify the unit you want to use, or reorder the units. The former is often the clearest option.
  
== Read more ==
+
== See also ==
  
* [http://www.freepascal.org/docs-html/ref/refsu84.html Unit scope] (FPC html doc)
+
* [http://www.freepascal.org/docs-html/ref/refsu87.html Unit scope] (FPC html doc)
 
* [http://www.freepascal.org/docs-html/user/userse11.html Compiling a unit] (FPC html doc)
 
* [http://www.freepascal.org/docs-html/user/userse11.html Compiling a unit] (FPC html doc)
  
 
[[category:Pascal]]
 
[[category:Pascal]]

Revision as of 09:22, 6 May 2014

Deutsch (de) English (en) español (es) suomi (fi) français (fr) português (pt) русский (ru)

A unit is a source code file (or the binary compiled from that file) which was written using the Pascal programming language, and that is designed to be a single module in an application or an object module.

Aim

A unit may be used where functionality needs to be provided to an application program or to other units. It allows writing code that performs that functionality once and have it used in many places. This can reduce the possibility of errors and increase code reusage.

A binary unit may be used where a unit author wishes to provide certain functionality to be used in a Pascal program but does not wish to provide the source code that performs that functionality.

Units were also used on older versions of Pascal when it was necessary on computers with limited resources to be able to load routines as needed rather than keeping every routine of the executable program in memory all of the time. A unit that needs to access e. g. procedures and data types in another unit must specify those units it needs to access in a Uses statement (but linking is done without the need to write a makefile as in C).

A unit may also be used to declare a series of global constants or variables for use by the entire application, without actually containing any executable code. This is similar to the common keyword in the Fortran programming language.

Format

A unit is defined with the unit keyword followed by the unit-identifier. The unit-identifier (in the following example the unit's name is “minimalunit”) should match the filename it is written in. A minimal working example (which does nothing) is:

unit minimalunit;
interface
	(* here comes stuff that the unit publicly offers *)
implementation
	(* here comes the implementation of offered stuff and *)
	(* optionally internal stuff (only known in the unit) *)
end.

where the part after interface corresponds to the public-part of other languages and implementation does so to private.

A more advanced basic structure is

unit advancedunit;
interface

implementation

initialization
	(* here may be placed code that is *)
	(* executed as the unit gets loaded *)

finalization
	(* code executed at program end *)

end.

The optional initialization and finalization blocks may be followed by code that is executed on program start/end.

Detailed unit example

A step-by-step example:

unit randomunit;
(* this unit does something *)

(* public *)
interface

type
	(* the type TRandomNumber gets globally known *)
	(* since it is included somewhere (uses-statement) *)
	TRandomNumber = integer;

(* of course the const- and var-blocks are possible, too *)

(* a list of procedure/function signatures makes *)
(* them useable from outside of the unit *)
function getRandomNumber(): TRandomNumber;

(* an implementation of a function/procedure *)
(* must not be in the interface-part *)

(* private *)
implementation

var
	(* var in private-part *)
	(* => only modifiable inside from this unit *)
	chosenRandomNumber: TRandomNumber;

function getRandomNumber(): TRandomNumber;
begin
	(* return value *)
	getRandomNumber := chosenRandomNumber;
end;

(* initialization is the part executed *)
(* when the unit is loaded/included *)
initialization
begin
	(* choose our random number *)
	chosenRandomNumber := 3;
	(* chosen by fair-dice-roll *)
	(* guaranteed to be random *)
end;

(* finalization is worked off at program end *)
finalization
begin
	(* this unit says 'bye' at program halt *)
	writeln('bye');
end;
end.

To include a unit just use the uses-statement.

program chooseNextCandidate;
uses
	(* include a unit *)
	randomunit;

begin
	writeln('next candidate: no. ' + getRandomNumber());
end.

When compiling, FPC checks this program for unit dependencies. It has to be able to find the unit "randomunit".

The simplest way to satisfy this is to have a unit whose name matches the file name it is written in (appending .pas is OK and encouraged). The unit file may be in the same directory where the program source is in or in the unit path FPC looks for units.

Unit precedence

When multiple units are described in a use clause, conflicts can occur with identifiers (procedures, types, functions etc) that have the same name in multiple units. In FPC, the last unit "wins" and provides the code for the unit.

If you want to have different behaviour, you can explicity prepend unitname.identifier to specify the unit you want to use, or reorder the units. The former is often the clearest option.

See also