CnPack Forum


 
Subject: where is class and variable modeling name convention
forum1
普通灌水员
Rank: 2



UID 64926
Digest Posts 0
Credits 93
Posts 30
点点分 93
Reading Access 10
Registered 2011-5-13
Status Offline
Post at 2011-12-9 02:16  Profile | Blog | P.M. 
where is class and variable modeling name convention

Not any class and variable modeling name convention. Check resharper model name convention

private Object => %_nameClass%
public Object =>  %NameClass%



See this template. I thing you fully understate what is this? If you make this functionality, your program will be the first that support into Delphi.

<CodeStyleSettings>
  <CSS>
    <FormatSettings />
    <Naming2 />
  </CSS>
  <CSharp>
    <FormatSettings>
      <MODIFIERS_ORDER IsNull="False">
        <Item>public</Item>
        <Item>protected</Item>
        <Item>internal</Item>
        <Item>private</Item>
        <Item>new</Item>
        <Item>abstract</Item>
        <Item>virtual</Item>
        <Item>override</Item>
        <Item>sealed</Item>
        <Item>static</Item>
        <Item>readonly</Item>
        <Item>extern</Item>
        <Item>unsafe</Item>
        <Item>volatile</Item>
      </MODIFIERS_ORDER>
    </FormatSettings>
    <UsingsSettings />
    <Naming2>
      <EventHandlerPatternLong>$object$_$event$</EventHandlerPatternLong>
      <EventHandlerPatternShort>$event$Handler</EventHandlerPatternShort>
    </Naming2>
  </CSharp>
  <HTML>
    <FormatSettings />
  </HTML>
  <JavaScript>
    <FormatSettings />
    <Naming2>
      <UserRule Name="JS_LOCAL_VARIABLE" Inspect="True" Prefix="" Suffix="" Style="aaBb" />
      <UserRule Name="JS_FUNCTION" Inspect="True" Prefix="" Suffix="" Style="aaBb" />
      <UserRule Name="JS_PARAMETER" Inspect="True" Prefix="" Suffix="" Style="aaBb" />
      <UserRule Name="JS_LABEL" Inspect="True" Prefix="" Suffix="" Style="aaBb" />
      <UserRule Name="JS_GLOBAL_VARIABLE" Inspect="True" Prefix="" Suffix="" Style="AaBb" />
      <UserRule Name="JS_OBJECT_PROPERTY_OF_FUNCTION" Inspect="True" Prefix="" Suffix="" Style="aaBb" />
      <UserRule Name="JS_CONSTRUCTOR" Inspect="True" Prefix="" Suffix="" Style="AaBb" />
    </Naming2>
  </JavaScript>
  <VB>
    <FormatSettings />
    <ImportsSettings />
    <Naming2>
      <EventHandlerPatternLong>$object$_$event$</EventHandlerPatternLong>
      <EventHandlerPatternShort>$event$Handler</EventHandlerPatternShort>
    </Naming2>
  </VB>
  <Web>
    <Naming2 />
  </Web>
  <XML>
    <FormatSettings />
  </XML>
  <Xaml>
    <Naming2>
      <UserRule Name="XAML_FIELD" Inspect="True" Prefix="_" Suffix="" Style="aaBb" />
      <UserRule Name="NAMESPACE_ALIAS" Inspect="True" Prefix="" Suffix="" Style="AaBb" />
      <UserRule Name="XAML_RESOURCE" Inspect="True" Prefix="" Suffix="" Style="AaBb" />
    </Naming2>
  </Xaml>
  <GenerateMemberBody />
  <Naming2>
    <EventHandlerPatternLong>$object$_On$event$</EventHandlerPatternLong>
    <EventHandlerPatternShort>$event$Handler</EventHandlerPatternShort>
    <PredefinedRule Inspect="True" Prefix="" Suffix="" Style="AaBb" ElementKind="TypesAndNamespaces" />
    <PredefinedRule Inspect="True" Prefix="I" Suffix="" Style="AaBb" ElementKind="Interfaces" />
    <PredefinedRule Inspect="True" Prefix="T" Suffix="" Style="AaBb" ElementKind="TypeParameters" />
    <PredefinedRule Inspect="True" Prefix="" Suffix="" Style="AaBb" ElementKind="MethodPropertyEvent" />
    <PredefinedRule Inspect="True" Prefix="" Suffix="" Style="aaBb" ElementKind="Locals" />
    <PredefinedRule Inspect="True" Prefix="" Suffix="" Style="aaBb" ElementKind="LocalConstants" />
    <PredefinedRule Inspect="True" Prefix="" Suffix="" Style="aaBb" ElementKind="Parameters" />
    <PredefinedRule Inspect="True" Prefix="" Suffix="" Style="AaBb" ElementKind="PublicFields" />
    <PredefinedRule Inspect="True" Prefix="_" Suffix="" Style="aaBb" ElementKind="PrivateInstanceFields" />
    <PredefinedRule Inspect="True" Prefix="_" Suffix="" Style="aaBb" ElementKind="PrivateStaticFields" />
    <PredefinedRule Inspect="True" Prefix="" Suffix="" Style="AaBb" ElementKind="Constants" />
    <PredefinedRule Inspect="True" Prefix="" Suffix="" Style="AaBb" ElementKind="PrivateConstants" />
    <PredefinedRule Inspect="True" Prefix="" Suffix="" Style="AaBb" ElementKind="StaticReadonly" />
    <PredefinedRule Inspect="True" Prefix="" Suffix="" Style="AaBb" ElementKind="PrivateStaticReadonly" />
    <PredefinedRule Inspect="True" Prefix="" Suffix="" Style="AaBb" ElementKind="EnumMember" />
    <PredefinedRule Inspect="True" Prefix="" Suffix="" Style="AaBb" ElementKind="Other" />
  </Naming2>
</CodeStyleSettings>

[ Last edited by  forum1 at 2012-2-26 18:28 ]
Top
forum1
普通灌水员
Rank: 2



UID 64926
Digest Posts 0
Credits 93
Posts 30
点点分 93
Reading Access 10
Registered 2011-5-13
Status Offline
Post at 2012-2-26 18:12  Profile | Blog | P.M. 
class and variable modeling convention

class and variable modeling convention

If my variable are named with wrong modeling name, variable must look like this




[ Last edited by  forum1 at 2012-2-26 18:28 ]
Top
forum1
普通灌水员
Rank: 2



UID 64926
Digest Posts 0
Credits 93
Posts 30
点点分 93
Reading Access 10
Registered 2011-5-13
Status Offline
Post at 2012-2-26 18:40  Profile | Blog | P.M. 
[size=+2]

Delphi Identifier Naming Conventions
[size=-2]How to name all those variables and components you use in your daily Delphi programming.
The name of an identifier (variable, component, ...) plays a fundamental role in (Delphi) programming.  
The Hungarian naming convention is quite useful — it's one technique among many that helps programmers produce better code faster. Since most of the API headers and documentation Microsoft has published over the last years have used Hungarian notation names for identifiers, many programmers have adopted one variation or another of this scheme for naming their identifiers.
In Delphi, using Hungarian notation, variable names begin with one or more lowercase letters that denote the variable type, thus providing an inherent identification. For example, the prefix h is used to identify a handle, as in hWnd or hDlg, referring to window and dialog box handles, respectively.
All components should be given descriptive names. No components should be left with their default names assigned by Delphi (Class name without the T letter plus a number). Components are to have a lowercase prefix to designate their type.
The following table defines some of suggested Delphi object name prefixes:
Category: in Types and classes
ObjectPrefixExample
ExceptionEEMyError = Class(Exception)
Classes and TypesTTMyClass = Class(TObject)
Fields in classesffVisible
EventsOnOnMouseDown
Category: VariablesTypePrefixExample
stringssSurName
BooleanbbIsThisOK
IntegeriiBirthYear
PointerppMePointer
DateTimedtdtBirthday
CurrencycurcurSallary
Category: ComponentsTypePrefixExample
FormfrmfrmMain
ButtonbtnbtnOK
LabellbllblName
EditededPassword
ComboBoxcbocboFont
ListBoxlblbFiles
TabletbltblMaster
QueryqryqrySumChild
DataBAsedbdbDataBase
PaintBoxpbpbMyPicture
MediaPlayermpmpMP3Player
OpenDialogOpenDialogOpenDialog
Top
forum1
普通灌水员
Rank: 2



UID 64926
Digest Posts 0
Credits 93
Posts 30
点点分 93
Reading Access 10
Registered 2011-5-13
Status Offline
Post at 2012-2-26 18:41  Profile | Blog | P.M. 
Object Pascal Style Guide

Original by Charles Calvert



This article documents a standard style for formatting Delphi code which is to be included in the JEDI Visual Component Library (JVCL). This document is a modified/annotated version of the article originally written by Charles Calvert (with his permission to do so). The original article can be obtained from the Codegear Community site. As documented elsewhere, it's not necessary for you to format your code according to the rules of this document before submitting it to Project JEDI, although that would be great. Instead the reformatting is done by the JVCL unit owners where necessary. We accept feedback in the form of corrections or suggestions. Send your communications to the JVCL team or Charlie Calvert. Note that with a few exceptions this document does not include coding guideliness, only formatting guideliness. A separate document for coding guideliness is in the making.



This article documents a standard style for formatting Delphi code. It is based on the conventions developed by the Delphi team.

Object Pascal is a beautifully designed language. One of its great virtues is its readability. These standards are designed to enhance that readability of Object Pascal code. When developers follow the simple conventions laid out in this guide, they will be promoting standards that benefit all Delphi developers by using a uniform style that is easy to read. Efforts to enforce these standards will increase the value of a developer's source code, particularly during maintenance and debugging cycles.



It goes without saying that these are conventions based primarily on matters of taste. Though we believe in, and admire the style promoted in these pages, we support them not necessarily because we believe they are right and others are wrong, but because we believe in the efficacy of having a standard which most developers follow. The human mind adapts to standards, and finds ways to quickly recognize familiar patterns, thereby assimilating meaning quickly and effortlessly. It is the desire to create a standard that will make reading code as simple as possible for the largest number of people that is behind this effort. If at first our guidelines seem strange to you, we ask you to try them for a while, and then we are sure you will grow used to them over time.



Do not post this specification on other web sites. Instead, simply link to either this version of the document or the original one on Charlie's website.


Contents



1.0 Introduction

    1.1 Background
    1.2 Acknowledgments

2.0 Source Files

    2.1 Source-File Naming
    2.2 Source-File Organization

        2.2.1 unit declaration
        2.2.2 uses declarations
        2.2.3 class/interface declarations

3.0 Naming Conventions

    3.1 Unit Naming
    3.2 Class/Interface Naming
    3.3 Field Naming
    3.4 Method Naming
    3.5 Local Variable Naming
    3.6 Reserved Words
    3.7 Type Declarations

4.0 White Space Usage

    4.1 Blank Lines
    4.2 Blank Spaces

        4.2.1 A single blank space (not tab) should be used
        4.2.2 Blanks should not be used

    4.3 Indentation
    4.4 Continuation Lines

5.0 Comments

    5.1 Block Comments
    5.2 Single-Line Comments

6.0 Classes

    6.1 Class Body Organization
    6.2 Method Declarations
    6.3 Data Store Declarations

7.0 Interfaces

    7.1 Interface Body Organization

8.0 Statements

    8.1 Simple Statements

        8.1.1 Assignment and expression statements
        8.1.2 Local variable declarations
        8.1.3 Array declarations

    8.2 Compound Statements

        8.2.3 if statement
        8.2.4 for statement
        8.2.5 while statement
        8.2.6 repeat until statement
        8.2.7 case statement
        8.2.8 try statement

9.0 Miscellanous

    9.1 Const, Var and Type
    9.2 Conditional compilation
    9.3 Resource strings
    9.4 Exceptions
    9.5 Categories and routine separation
    9.6 Assembler
    9.7 Local routines
    9.8 Parameter Declarations
    9.9 Initialization of global variables

1.0 Introduction

This document is not an attempt to define a grammar for the Object Pascal language. For instance, it is illegal to place a semicolon before an else statement; the compiler simply won't let you do it. As a result, I do not lay that rule out in this style guide. This document is meant to define the proper course of action in places where the language gives you a choice. I usually remain mute on matters that can only be handled one way.
1.1 Background

The guidelines presented here are based on the public portions of the Delphi source. The Delphi source should follow these guidelines precisely. If you find cases where the source varies from these guidelines, then these guidelines, and not the errant source code, should be considered your standard. Nevertheless, you should use the source as a supplement to these guidelines, at least so far as it can help you get a general feel for how your code should look.
1.2 Acknowledgments

The format of this document and some of its language is based on work done to define a style standard for the Java language. Java has had no influence on the rules for formatting Object Pascal source, but documents found on the Sun web site formed the basis for this document. In particular the style and format of this document were heavily influenced by "A Coding Style Guide for Java WorkShop and Java Studio Programming" by Achut Reddy. That document can be found at the following URL: http://www.sun.com/workshop/java/wp-coding



The Delphi team also contributed heavily to the generation of this document, and indeed, it would not have been possible to create it without their help.



Many of the modifications to this document where at least partly based on feedback by Marcel van Brakel and Mike Lischke. Other people who contributed are Robert Marquardt and Matthias Thoma.

2.0 Source Files

Object Pascal source is divided up primarily into units and Delphi Project files, which both follow the same conventions. A Delphi Project file has a DPR extension. It is the main source file for a project. Any units used in the project will have a PAS extension. Additional files, such as batch files, html files, or DLLs, may play a role in a project, but this paper only treats the formatting of DPR and PAS files.
2.1 Source-File Naming

Object Pascal supports long file names. If you are appending several words to create a single name, then it is best to use capital letters for each word in the name: MyFile.pas. This is known as InfixCaps, or Camel Caps. Extensions should be in lower case. For historical reasons, the Delphi source itself often confines itself to 8:3 naming patterns, but developers no longer need feel constrained by those limits, even if turning in source that might be used by the Delphi team.



If you are translating a C/C++ header file, then your Pascal header translation will usually have the same name as the file you are translating, except it should have a PAS extension. For instance, Windows.h would become Windows.pas. If the rules of Pascal grammar force you to combine multiple header files into a single unit, then use the name of the base unit into which you are folding the other files. For instance, if you fold WinBase.h into Windows.h, then call the resulting file Windows.pas.



All JEDI Visal Component Library source-files must be prefixed with 'Jv'. Also, since these files are to be ported to Linux be carefull that you use the same capitalization everywhere a source-file is referenced (Linux filenames are, as opposed to Win32, case-sensitive).
2.2 Source-File Organization

All Object Pascal units should contain the following elements in the following order:



    Copyright/ID block comment
    CVS Id marker

    Unit Name

    Include files

    Interface section

    Additional defines

    Uses clause

    Implementation

    Uses clause

    A closing end and a period.

    At least one blank line should separate each of these elements.

     



Additional elements can be structured in the order you find most appropriate, except that the top of the file should always list the copyright first, the unit name second, then any conditional defines, compiler directives or include statements, then the uses clause. For the JVCL the following header is used. Replace JvGraphics with the appropriate unit name. The Last Modified date is kept up to date by the unit owner and should always match the last modified date in the filesystem.



{-----------------------------------------------------------------------------
The contents of this file are subject to the Mozilla Public License
Version 1.1 (the "License"); you may not use this file except in compliance
with the License. You may obtain a copy of the License at
http://www.mozilla.org/MPL/MPL-1.1.html

Software distributed under the License is distributed on an "AS IS" basis,
WITHOUT WARRANTY OF ANY KIND, either expressed or implied. See the License for
the specific language governing rights and limitations under the License.

The Original Code is: JvActions.Pas, released on 2002-10-04.

The Initial Developer of the Original Code is Sébastien Buysse [sbuysse att buypin dott com]
Portions created by Sébastien Buysse are Copyright (C) 2002 Sébastien Buysse.
All Rights Reserved.

Contributor(s): -

You may retrieve the latest version of this file at the Project JEDI's JVCL home page,
located at http://jvcl.sourceforge.net

Known Issues:
-----------------------------------------------------------------------------}
// $Id: StyleGuide.htm 12496 2009-09-09 21:31:04Z obones $

unit JvActions;



Compiler directives are not directly included in the source files. Instead a global JVCL.INC include file is used which defines all standard directives. Ex.



{$I JVCL.INC}

interface



If overriding directives are needed they can be included below this include but this must be avoided. If overriding directives are needed these must be documented. You should strive to override directives only a local scope. For example, for a single procedure.



{$S-,W-,R-}
{$C PRELOAD}

interface

uses
  Windows, Messages, Classes, Controls, Forms, Graphics, StdCtrls, ExtCtrls, CommCtrl;



It does not matter if you place a type section before a const section, or if you mix type and const sections up in any order you choose.

The implementation should list the word implementation first, then the uses clause, then any include statements or other directives:

implementation

uses
  Consts, SysUtils, ActnList, ImgList;

{$R BUTTONS.RES}

2.2.1 Unit declaration

Every source file should contain a unit declaration. The word unit is a reserved word, so it should be in lower case. The name of the unit should be in mixed upper and lowercase, and must be the same as the name used by the operating system's file system. Example:



unit JvMyUnit;



This unit would be called MyUnit.pas when an entry is placed in the file system.
2.2.2 uses declarations

Inside units, a uses declaration should begin with the word uses, in lowercase. Add the names of the units, following the capitalization conventions used in the declaration found inside the units:



uses MyUnit;



Each unit must be separated from its neighbor by a comma, and the last unit should have a semicolon after it:



uses
  Windows, SysUtils, Classes, Graphics, Controls, Forms, TypInfo;



The uses clause is always started on the next line and units are written down one after another, wrapping at 100 columns. Furthermore, you should separate the standard Delphi units, JVCL units and Platform dependent units. Finally, it is prefered to list to units in alpabetical order unless the order is important (this should never be the case but sometimes is, an example is the Windows unit which - by convention - should always be listed first). An example follows (comments shouldn't be included):



uses
  {$IFDEF WIN32}
  Windows, ActiveX,    // Windows units
  {$ENDIF}
  {$IFDEF LINUX}
  ..Linux specific units go here
  {$ENDIF}
  Math, SysUtils,      // Standard Delphi platform independent units
  JVCLBase, JVCLStrings; // JVCL units

2.2.3 class/interface declarations

A class declaration begins with two spaces, followed by an identifier prefaced by TJv. Identifiers should begin with a capital letter, and should have capital letters for each embedded word (InfixCaps). Never use tab characters in your Object Pascal source. Example:



TJvMyClass



Follow the identifier with a space, then an equals sign, then the word class (all in lower case), furthermore a parenthesis, the name of the ancestor class, and closing parenthesis. The class keyword and ancestor should be separated with a space:



TJvMyClass = class (TObject)



Scoping directives should be two spaces in from the margin, and declared in the order shown in this example:



  TJvMyClass = class (TObject)
  private
  protected
  public
  published
  end;



Data should always be declared only in the private section, and its identifier should be prefaced by an F. All type declarations should be four spaces in from the margin:



  TJvMyClass = class (TObject)
  private
    FMyData: Integer;
  protected
    function GetData: Integer;
    procedure SetData(Value: Integer);
  public
  published
    property MyData: Integer read GetData write SetData;
  end;



Interfaces follow the same rules as class declarations, except you should omit any scoping directives or private data, and should use the word interface rather than class.
Property accessor functions should be in the protected section so that a user deriving a class from yours can use them. Should there be no accessor (either read or write) then the field itself should be in the protected section.

3.0 Naming Conventions

Except for reserved words and directives, which are in all lowercase, all Pascal identifiers should use InfixCaps, which means the first letter should be a capital, and any embedded words in an identifier should be in caps, as well as any acronym that is embedded:



MyIdentifier MyFTPClass



The major exception to this rule is in the case of header translations, which should always follow the conventions used in the header. For instance, write WM_LBUTTONDOWN, not wm_LButtonDown.



Except in header translations, do not use underscores to separate words. Class names should be nouns or noun phrases. Interface or class names depend on the salient purpose of the interface.



GOOD type names:

    AddressForm
    JvArrayIndexOutOfBoundsException

BAD type names:

    ManageLayout (verb phrase)
    delphi_is_new_to_me (underscores)

It seems to be unavoidable but every now and then someone suggests using hungarian, or a similar, notation for identifier naming. Although good arguments can be provided in favor of hungarian notation, at least as many arguments can be given against it. The JVCL will not use hungarian notation, ever! There, that's out of the way. Identifiers in the JVCL should be named as the examples above, names which describe the purpose of the identifier not what type they happen to be of.
3.1 Unit Naming

Use InfixCaps, as described at the beginning of this section. See also the section on unit declarations



As described earlier, all JVCL units should have the "Jv" prefix.
3.2 Class/Interface Naming

Use InfixCaps, as described at the beginning of this section. Begin each type declaration with a capital T:



TJvMyType



See also the section on class/interface declarations.



All JVCL classes are prefixed with 'TJv' not just a capital T. Types which are used only internally don't have to include the 'Jv' prefix although you should be carefull with the naming when they are declared in the interface section.
3.3 Field Naming

Use InfixCaps, as described at the beginning of this section. Begin each type declaration with a capital F, and declare all data types in the private section, using properties or getters and setters to provide public or protected access. For example, use the name GetSomething to name a function returning an internal field value and use SetSomething to name a procedure setting that value.



Do not use all caps for const declarations except where required in header translations.



Delphi is created in California, so we discourage the use of hungarian notation, except where required in header translations:



CORRECT
  FMyString: string;

INCORRECT
  lpstrMyString: string;



The exception to the Hungarian notation rule is in enumerated types.



  TBitBtnKind = (bkCustom, bkOK, bkCancel, bkHelp,
    bkYes, bkNo, bkClose, bkAbort, bkRetry,
    bkIgnore, bkAll);



In this case the letters bk are inserted before each element of this enumeration. bk stands for ButtonKind.

When thinking about naming conventions, consider that one-character field names should be avoided except for temporary and looping variables.

Looping variables are by convention named I (capital i) and J. Other commonly used single character identifier names are: S (string) and R (Result). Single letter variables/field names should always be capitals but other than the ones mentioned above you should avoid them and use more meaningful names.

Avoid variable l ("el") because it is hard to distinguish it from 1 ("one") on some printers and displays.
3.4 Method Naming

Method names should use the InfixCaps style. Start with a capital letter, and capitalize the first letter of any subsequent word in the name, as well as any letters that are part of an acronym. All other characters in the name are lower case. Do not use underscores to separate words. Note that this is identical to the naming convention for non-constant fields; however it should always be easy to distinguish the two from context. Method names should be imperative verbs or verb phrases.



Examples:

      
    Good method names:

        ShowStatus
        DrawCircle
        AddLayoutComponent

    Bad method names:

        MouseButton (noun phrase; doesn't describe function)
        drawCircle (starts with lower-case letter)
        add_layout_component (underscores)
        ServerRunning (verb phrase, but not imperative)

A note about the last example (ServerRunning): The function of this method is unclear. Does it start the server running (better: StartServer), or test whether or not it is running (better: IsServerRunning)?



A method to get or set some property of the class should be called GetProperty or SetProperty respectively, where Property is the name of the property.



Examples:



GetHeight, SetHeight



A method to test some boolean property of the class should be called IsVisible, where Visible is the name of the property.



Examples:



IsResizable, IsVisible

3.5 Local Variable Naming

Local variables follow the same naming rules as field names, except you omit the initial F, since this is not a Field of an object. (see section 3.3).
3.6 Reserved Words

Reserved words and directives should be all lowercase. This can be a bit confusing at times. For instance types such as Integer are just identifiers, and appear with a first cap. Strings, however, are declared with the reserved word string, which should be all lowercase.
3.7 Type Declarations

All type declarations should begin with the letter T, and should follow the same capitalization specification laid out in the beginning of this section, or in the section on class declarations.


4.0 White Space Usage
4.1 Blank Lines

Blank lines can improve readability by grouping sections of the code that are logically related. A blank line should also be used in the following places:



    After the copyright block comment, package declaration, and import section.

    Between class declarations.

    Between method declarations.

4.2 Blank Spaces

Object Pascal is a very clean, easy to read language. In general, you don't need to add a lot of spaces in your code to break up lines. The next few sections give you some guidelines to follow when placing spaces in your code.
4.2.2 Blanks should not be used:

    Between a method name and its opening parenthesis.

    Before or after a .(dot) operator.

    Between a unary operator and its operand.

    Between a cast and the expression being cast.

    After an opening parenthesis or before a closing parenthesis.

    After an opening square bracket [ or before a closing square bracket ].

    Before a semicolon.



Examples of correct usage:



function TMyClass.MyFunc(var Value: Integer);
MyPointer := @MyRecord;
MyClass := TMyClass(MyPointer);
MyInteger := MyIntegerArray[5];

Examples of incorrect usage:

function TMyClass.MyFunc( var Value: Integer ) ;
MyPointer := @ MyRecord;
MyClass := TMyClass ( MyPointer ) ;
MyInteger := MyIntegerArray [ 5 ] ;

4.3 Indentation

You should always indent two spaces for all indentation levels. In other words, the first level of indentation is two spaces, the second level four spaces, the third level 6 spaces, etc. Never use tab characters.



There are few exceptions. The reserved words unit, uses, type, interface, implementation, initialization and finalization should always be flush with the margin. The final end statement at the end of a unit should be flush with the margin. In the project file, the word program, and the main begin and end block should all be flush with the margin. The code inside the begin..end block, should be indented at least two spaces.
4.4 Continuation Lines

Lines should be limited to 100 columns. Lines longer than 100 columns should be broken into one or more continuation lines, as needed. All the continuation lines should be aligned and indented from the first line of the statement, and indented two characters. Always place begin statements on their own line.



Examples:



// CORRECT

function CreateWindowEx(dwExStyle: DWORD;
  lpClassName: PChar; lpWindowName: PChar;
  dwStyle: DWORD; X, Y, nWidth, nHeight: Integer;
  hWndParent: HWND; hMenu: HMENU; hInstance: HINST;
  lpParam: Pointer): HWND; stdcall;


if ((X = Y) or (Y = X) or
  (Z = P) or (F = J) then
begin
  S := J;
end;



Never wrap a line between a parameter and its type, unless it is a comma separated list, then wrap at least before the last parameter so the type name follows to the next line. The colon for all variable declarations contains no whitespace between it and the variable. There should be a single space following the colon before the type name;



procedure Foo(Param1: Integer; Param2: Integer);


procedure Foo( Param :Integer; Param2:Integer );



A continuation line should never start with a binary operator. Avoid breaking a line where normally no white space appears, such as between a method name and its opening parenthesis, or between an array name and its opening square bracket. If you must break under these circumstances, then one viable place to begin is after the opening parenthesis that follows a method name. Never place a begin statement on the same line with any other code.



Examples:



// INCORRECT
while (LongExpression1 or LongExpression2) do begin
  // DoSomething
  // DoSomethingElse;
end;

// CORRECT
while (LongExpression1 or LongExpression2) do
begin
  // DoSomething
  // DoSomethingElse;
end;

if (LongExpression1)  
   or (LongExpression2)
   or (LongExpression3) then
// CORRECT
if (LongExpression1) or
  (LongExpression2) or
  (LongExpression3) then
  

5.0 Comments

The Object Pascal language supports two kinds of comments: block, and single-line comments. Some general guidelines for comment usage include:



    It is helpful to place comments near the top of unit to explain its purpose.

    It is helpful to place comments before a class declaration.

    It is helpful to place comments before some method declarations.

    Avoid making obvious comments:



i := i + 1;     // Add one to i



    Remember that misleading comments are worse than no comments at all.

    Avoid putting any information into comments that is likely to become out of date.

    Avoid enclosing comments in boxes drawn with asterisks or other special typography.

    Temporary comments that are expected to be changed or removed later should be marked with the special tag "TODO:" so that they can easily be found afterwards. Ideally, all temporary comments should have been removed by the time a program is ready to be shipped.



Example:



// TODO: Change this to call Sort when it is fixed
List.MySort;

5.1 Block Comments



Object Pascal supports two types of block comments. The most commonly used block comment is a pair of curly braces: { }. The Delphi team prefers to keep comments of this type as spare and simple as possible. For instance, you should avoid using asterisks to create patterns or lines inside your comments. Instead, make use of white space to break your comments up, much as you would in a word processing document. The words in your comments should start on the same line as the first curly brace, as shown in this excerpt from DsgnIntf.pas:



{ TPropertyEditor

  Edits a property of a component, or list of components,
  selected into the Object Inspector.  The property
  editor is created based on the type of the
  property being edited as determined by the types
  registered by...

  etc...

    GetXxxValue
      Gets the value of the first property in the
      Properties property.  Calls the appropriate
      TProperty GetXxxValue method to retrieve the
      value.

    SetXxxValue Sets the value of all the properties
      in the Properties property.  Calls the appropriate
      TProperty SetXxxxValue methods to set the value. }



A block comment is always used for the copyright/ID comment at the beginning of each source file. It is also used to "comment out" several lines of code.

Block comments used to describe a method should appear before the method declaration.



Example:



// CORRECT

{ TMyObject.MyMethod

  This routine allows you to execute code. }

procedure TMyObject.MyMethod;
begin
end;


// INCORRECT

procedure TMyObject.MyMethod;
{******************************************************
  TMyObject.MyMethod

  This routine allows you to execute code.
*******************************************************}
begin
end;



A second kind of block comment contains two characters, a parenthesis and an asterisk: (* *). This is sometimes called starparen comments. These comments are generally useful only during code development, as their primary benefit is that they allow nesting of comments, as long as the nest level is less than 2. Object Pascal doesn't support nesting comments of the same type within each other, so really there is only one level of comment nesting: curly inside of starparen, and starparen inside of curly. As long as you don't nest them, any other standard Pascal comments between comments of this type will be ignored. As a result, you can use this syntax to comment out a large chunk of code that is full of mixed code and comments:



(* procedure TForm1.Button1Click(Sender: TObject);
begin
  DoThis; // Start the process
  DoThat; // Continue iteration
  { We need a way to report errors here, perhaps using
    a try finally block ??? }
  CallMoreCode; // Finalize the process
end; *)



In this example, the entire Button1Click method is commented out, including any of the subcomments found between the procedure's begin..end pair.
5.2 Single-Line Comments

A single-line comment consists of the characters // followed by text. Include a single space between the // and the comment itself. Place single line comments at the same indentation level as the code that follows it. You can group single-line comments to form a larger comment.

A single-line comment or comment group should always be preceded by a blank line, unless it is the first line in a block. If the comment applies to a group of several statements, then the comment or comment group should also be followed by a blank line. If it applies only to the next statement (which may be a compound statement), then do not follow it with a blank line.



Example:



Table1.Open;



Single-line comments can also follow the code they reference. These comments, sometimes referred to as trailing comments, appear on the same line as the code they describe. They should have at least one space-character separating them from the code they reference. If more than one trailing comment appears in a block of code, they should all be aligned to the same column.



Example:



if (not IsVisible) then
  Exit;          // nothing to do
Inc(StrLength);  // reserve space for null terminator



Avoid commenting every line of executable code with a trailing comment. It is usually best to limit the comments inside the begin..end pair of a method or function to a bare minimum. Longer comments can appear in a block comment before the method or function declaration.


Classes
6.1 Class Body Organization

The body of a class declaration should be organized in the following order:



    Field declarations

    Method declarations

    Property declarations



The fields, properties and methods in your class should be arranged alphabetically by name.
6.1.1 Access levels

Except for code inserted by the IDE, the scoping directives for a class should be declared in the following order:



    Private declarations

    Protected declarations

    Public declarations

    Published declarations



There are four access levels for class members in Object Pascal: published, public, protected, and private -- in order of decreasing accessibility. By default, the access level is published. In general, a member should be given the lowest access level which is appropriate for the member. For example, a member which is only accessed by classes in the same unit should be set to private access. Also, declaring a lower access level will often give the compiler increased opportunities for optimization. On the other hand, use of private makes it difficult to extend the class by sub-classing. If there is reason to believe the class might be sub-classed in the future, then members that might be needed by sub-classes should be declared protected instead of private, and the properties used to access private data should be given protected status.

You should never allow public access to data. Data should always be declared in the private or protected section, and any public access should be via getter and setter methods, or properties.
6.1.8 Constructor declarations

Methods should be arranged alphabetically. It is correct either to place your constructors and destructors at the head of this list in the public section, or to arrange them in alphabetical order within the public section. If you have a large number of methods, you may want to group them by functionnality, and then order them alphabetically in the group. Always separate groups by a blank line.



If there is more than one constructor, and if you choose to give them all the same name, then sort them lexically by formal parameter list, with constructors having more parameters always coming after those with fewer parameters. This implies that a constructor with no arguments (if it exists) is always the first one. For compatibility with C++Builder, you must make the parameter lists of your constructors unique. C++ cannot call constructors by name, so the only way to distinguish between multiple constructors is by parameter list.
6.2 Method Declarations

If possible, a method declaration should appear on one line.



Examples:



procedure ImageUpdate(Image img, infoflags: Integer,
  x: Integer, y: Integer, w: Integer, h: Integer)



Interfaces

Interfaces are declared in a manner that runs parallel to the declaration for classes:



InterfaceName = interface([Inherited Interface])
InterfaceBody
end;



    An interface declaration should be indented two spaces. The body of the interface is indented by the standard indentation of four spaces. The closing end statement should also be indented two characters. There should be a semi-colon following the closing end statement.



    There are no fields in an interface declaration. Properties, however, are allowed.



    All interface methods are inherently public and abstract; do not explicitly include these keywords in the declaration of an interface method.



    Except as otherwise noted, interface declarations follow the same style guidelines as classes.

7.1 Interface Body Organization

The body of an interface declaration should be organized in the following order:



    Interface method declarations

    Interface property declarations



The declaration styles of interface properties and methods are identical to the styles for class properties and methods.


8.0 Statements

Statements are one or more lines of code followed by a semicolon. Simple statements have one semicolon, while compound statements have more than one semicolon and therefore consist of multiple simple statements.



Here is a simple statement:



A := B;



Here is a compound, or structured, statement:



begin
  B := C;
  A := B;
end;



8.0.1 Simple Statements

A simple statement contains a single semicolon. If you need to wrap the statement, indent the second line two spaces in from the previous line:



MyValue :=
  MyValue + (SomeVeryLongStatement / OtherLongStatement);



8.0.1 Compound Statements

Compound Statements always end with a semicolon, even if it is syntactically not required. For example the last statement in the following example must have a terminating semicolon for it to be acceptable in the JVCL.



begin
  MyStatement;
  MyNextStatement;
  MyLastStatement;   // semicolon optional
end;



8.1.1 Assignment and expression statements



Each line should contain at most one statement. For example:



a := b + c; Inc(Count); // INCORRECT
a := b + c;             // CORRECT
Inc(Count);             // CORRECT

8.1.2 Local variable declarations

Local variables should have Camel Caps, that is, they should start with a capital letter, and have capital letters for the beginning of each embedded word. Do not preface variable names with an F, as that convention is reserved for Fields in a class declaration:



var
  MyData: Integer;
  MyString: string;

You may declare multiple identifiers of the same type on a single line:



var
  ArraySize, ArrayCount: Integer;



This practice is discouraged in class declarations. There you should place each field on a separate line, along with its type.

You should only declare identifiers on a single line if they are logically related.
8.1.3 Array declarations

There should always be a space before the opening bracket "[" and after the closing bracket.



type
  TMyArray = array [0..100] of Char;

8.2.3 if statement

If statements should always appear on at least two lines.



Example:



if A < B then DoSomething;

if A < B then
  DoSomething;



In the JVCL the first example is allowed but discouraged. Use it only in "obvious" situations such as "if ParameterIncorrect then Exit;"

In compound if statements, put each element separating statements on a new line:



Example:



  // INCORRECT
  if A < B then begin
    DoSomething;
    DoSomethingElse;
  end else begin
    DoThis;
    DoThat;
  end;

  // CORRECT
  if A < B then
  begin
    DoSomething;
    DoSomethingElse;
  end
  else
  begin
    DoThis;
    DoThat;
  end;
  

Here are a few more variations that are considered valid (except for the second one, all these variations are discouraged):



  // CORRECT
  if Condition then
  begin
    DoThis;
  end else
  begin
    DoThat;
  end;

  // CORRECT
  if Condition then
  begin
    DoThis;
  end
  else
    DoSomething;

  // CORRECT
  if Condition then
  begin
    DoThis;
  end else
    DoSomething;

  if Condition then
  begin
    DoThis;
  end
  else DoSomething;
  

One that has fallen out of favor but deserves honorable mention:



  if Condition then
    DoThis
  else DoThat;
  

Avoid extraneous parentheses when formulating the conditional in an if statement. In other words, don't encapsulate the conditional statement in parenthesis if it's not syntactically required and doesn't provide additional readability. An obvious example:



  // CORRECT
  if I > 0 then
    DoSomething;
   
  // INCORRECT
  if (I > 0) then
    DoSomething;

8.2.4 for statement

Example:



  // INCORRECT
  for i := 0 to 10 do begin
    DoSomething;
    DoSomethingElse;
  end;

  // CORRECT
  for i := 0 to 10 do
  begin
    DoSomething;
    DoSomethingElse;
  end;



If the body of the for loop consist of a single statement then both of the examples below are allowed. As with if statements, the first one is discouraged though.



  for I := 0 to 10 do DoSomething;

  for I := 0 to 10 do
    DoSomething;
  

8.2.5 while statement

Example:



  // INCORRECT
  while x < j  do begin
    DoSomething;
    DoSomethingElse;
  end;

  // CORRECT
  while x < j do
  begin
    DoSomething;
    DoSomethingElse;
  end;
  

The same as with for loops applies here. Both of the following examples are allowed but the first one is discouraged.



  while x < j do Something;

  while x < j do
    Something;

8.2.6 repeat until statement

Example:



  // CORRECT
  repeat
    x := j;
    j := UpdateValue;
  until j > 25;

8.2.7 case statement

Example:



  // CORRECT
  case Control.Align of
    alLeft, alNone: NewRange := Max(NewRange, Position);
    alRight: Inc(AlignMargin, Control.Width);
  end;


  // CORRECT
  case ScrollCode of
    SB_LINEUP, SB_LINEDOWN:
      begin
        Incr := FIncrement div FLineDiv;
        FinalIncr := FIncrement mod FLineDiv;
        Count := FLineDiv;
      end;
    SB_PAGEUP, SB_PAGEDOWN:
      begin
        Incr := FPageIncrement;
        FinalIncr := Incr mod FPageDiv;
        Incr := Incr div FPageDiv;
        Count := FPageDiv;
      end;
  else
    Count := 0;
    Incr := 0;
    FinalIncr := 0;
  end;



Except the obvious situations in which the first variation is used for readability, the JVCL only uses the second example. That is, each case label starts on a separate line with a two column indent relative to the case statement and the body of that particular case follows on the next line with an additional two column indent. If the body on a particular case consists of only a single statement then the begin...end pair can be omitted. In this case the statement is aligned using a two column indent relative to the case statement (identical to the begin reserved word in the example above).
8.2.8 try statement

Example:



  // Correct
  try
    try
      EnumThreadWindows(CurrentThreadID, @Disable, 0);
      Result := TaskWindowList;
    except
      EnableTaskWindows(TaskWindowList);
      raise;
    end;
  finally
    TaskWindowList := SaveWindowList;
    TaskActiveWindow := SaveActiveWindow;
  end;

9.0 Miscellanous
9.1 Const, Var and Type

The reserved words var, const and type always appear alone on a line. These example are correct:



type
  TMyType = Integer;

const
  MyConstant = 100;

var
  MyVar: Integer;



But these are not:



type TMyType = Integer;

const MyConstant = 100;

var MyVar: Integer;



Additionally a procedure should only have a single type, const and var section and, if possible, in that order. For example:



procedure SomeProcedure;
type
  TMyType = Integer;
const
  ArraySize = 100;
var
  MyArray: array [1..ArraySize] of TMyType;
begin
  ...
end;

9.2 Conditional compilation

All JVCL units must include the JVCL.INC file. This file defines a number of global directives. The include statement should be placed between the unit and interface keywords. When using any of the directives from this file, or others for that matter, you should always repeat the conditional in the ENDIF directive. For example:

{$IFDEF MSWINDOWS}
  // conditionally compiled code
{$ENDIF}

This may seem overkill if the conditionally compiled code only extends a few lines but is a tremendous visual aid when the code actually spans multiple pages of code (as is often the case for platform dependent code). Platform dependent code should use the conditionals MSWINDOWS, for code intended to run on MS Windows platforms, WIN32 for windows 32 only code, UNIX for unix compatible code and LINUX for Linux only code.
9.3 Resource strings

All resourcestrings should be of the format 'Rs'[Category][Name]. [Category] should be (an abbreviation of) the category in which the code resides, [Name] is a descriptive name for the string itself. For example, the TJvCriticalSectionEx CreateEx constructor raises an exception on initialization failure. The exception message is declared as a resourcestring with the name RsSynchInitCriticalSection.

All resourcestrings must be declared in the global JvResources.pas file which is included in each JVCL unit. This is to ease translation. Literal strings should be avoided where possible (use a constant whenever you can).
9.4 Exceptions

Exceptions are prefixed with 'EJv' instead of 'TJv'. All JVCL exceptions should be ultimately derived from EJvError which is declared in JvBase.

When raising an exception you should prefer the CreateRes(ResStringRec: PResStringRec) constructor for efficiency. Thus, an exception is raised like this:

raise EJvSomeException.CreateRes(@RsSomeResourceString);

9.5 Categories and routine separation

Typically each JVCL unit is a single category. For example, JvSynch contains all kinds of synchronization classes and subroutines. Within a unit there is usually a further categorization, for example JvSynch has a number of 'Locked Integer Manipulation' routines which form a subcategory within this unit. In the interface section each subcategory is divided using two 100 column width lines in between which there is a one line description of the subcategory. For example:



function LockedAdd(var Target: Integer; Value: Integer): Integer;
function LockedCompareExchange(var Target: Integer; Exch, Comp: Integer): Integer;



In the implementation section this separation is identical except that the lines are composed using the equals character (=).



function LockedAdd(var Target: Integer; Value: Integer): Integer;
asm
  MOV     ECX, EAX



In the implementation section each routine or method is separated from its predecessor using a 80 column width line composed of minus characters (-).



  if (L > 0) and (Path[L] <> PathSeparator) then Result := Path + PathSeparator;
end;


function PathAddExtension(const Path, Extension: string): string;
begin
  Result := Path;
  if (Path <> '') and (ExtractFileExt(Path) = '') and (Extension <> '') then
  begin
    if Extension[1] = '.' then
      Result := Result + Extension
    else
      Result := Result + '.' + Extension;
  end;
end;


function PathAppend(const Path, Append: string): string;
var
  PathLength: Integer;
  B1, B2: Boolean;
begin
  if Append = '' then
    Result := Path

9.6 Assembler

While use of Assembler is discouraged, there are cases when it cannot be avoided. In that case, assembler is to be formatted like this:



        REP     MOVSW
        JMP     @@2
@@1:
        LEA     ESI, [ESI + 2 * ECX - 2]
        LEA     EDI, [EDI + 2 * ECX - 2]



That is, the opcode is indented 8 spaces and the operands are aligned on the 16th column. Labels should be indented with two spaces or aligned on the left side and be camel case. All opcodes and registers should be written fully in uppercase. Numeric labels are acceptable but a more descriptive name is preferred. General punctuation formatting still applies, e.g. a single space after each comma and a space on both sides of an operator (such as the addition operator +). Additionally, never put labels and commands on the same line and always prefix labels with the @ character to make the label scope local. Generically speaking, assembler should be avoided but if it is used it should be heavily commented.
9.7 Local routines

Local functions should be indented two spaces in their entirety and separated from the procedure declaration and begin statement by a single line. If the 'outer' procedure (SomeProcedure in the example) has local variables these should be declared before the local procedure, regardless of whether the local procedure needs access to them. However, local routines should be avoided. Whenever it seems reasonable to extract code as a local subroutine, think carefully whether the routine can be made more generic and extracted as a normal, global routine (of course should be moved to the appropriate unit as well). This will make the life of C++ Builder users far easier. For example:



procedure SomeProcedure;
var
  I: Integer;

  procedure LocalProcedure;
  begin
    ...
  end;

begin
  ...
  LocalProcedure;
  ...
end;

9.8 Parameter declarations

When declaring the parameters list of a procedure, function or method, observe the following recommendations:



    Combine formal parameters of same type into one statement

    Usage of A in parameter names is discouraged unless it concerns a method of a class which has as a property which is named identical.





Although technically these are not formatting issues, I'd like to methion them here anyway:



    Ordering of parameters should be: input, input./output and output parameters and within that ordering: most used, least used. Parameters with default values are, as required by the Object Pascal language rules, always placed at the back of the list.

    Use of const for parameter types is recommended even if it's use does not increases the efficiency of the parameter passing. For example, a parameter of type Integer should be declared with the const modifier if that's the semantic meaning of the parameter in question. However, you MUST not use the const specifier for parameters that are objects (derived from TObject). This is because of a bug in C++ Builder which results in unusable header files.

9.9 Initialization of global variables

Global variables are, like class members, automatically initialized to 0. This has different meaning for different types. For example, and Integer is initialized to 0 while a pointer is initialized to nil. Because of this, if the global variable needs to be 0 initialized, which is often the case, this should not be done explicitly. Instead you should rely on Delphi to do this for you. This is for efficiency reasons because it influences how the variable ends up in the executable and later in memory. If you desire you can add a comment to indicate reliance on 0 initialization like so:



var
  MyGlobalVariable: Pointer{ = nil};
Top
 




All times are GMT++8, the time now is 2024-11-22 01:46

    本论坛支付平台由支付宝提供
携手打造安全诚信的交易社区 Powered by Discuz! 5.0.0  © 2001-2006 Comsenz Inc.
Processed in 0.031040 second(s), 9 queries , Gzip enabled

Clear Cookies - Contact Us - CnPack Website - Archiver - WAP