Project Analyzer Help - Comment directives

Problem detection finds coding problems. Comment directives hide them. When you wish to ignore problems in a certain section of code, you write a comment directive in your code. Comment directives are specially formatted comments. Example: '$PROBHIDE ALL hides all problems on a single line.

Comment directives: Table of Contents

Comment directive syntax: 1. PROBHIDE — Hide problems; 2. PROBSHOW – Show problems hidden by PROBHIDE; 3. END – Close a BEGIN..END block
Scope: On which lines should I put the directives?
Precedence: In the case of conflicting directives, which one has precedence?
Turning dead procedures alive: You can flag a procedure as live code.
Ignore comment directives: Command line option /NOCDIRECTIVES ignores directives.
PROJECT_ANALYZER constant: PROJECT_ANALYZER is a special constant that allows further fine-tuning of code analysis.

Comment directive syntax

That's it! This will hide a "Goto statement found" problem for a line. Now let's look at the syntax. Comment directives always start with '$. The '$ prefix marks the start of a comment directive line.

The available directives are: PROBHIDE to hide problems, PROBSHOW to show them again and END to finish a BEGIN..END block.

You can use spaces the way you like. '$PROBHIDE, ' $PROBHIDE and '$ PROBHIDE are the same. Rem comments can't contain any directives, though. Directives can appear as full-line comments or as end-of-line comments. A full-line directive affects the next line while an end-of-line directive affects the current line (see Scope). Directives can't appear inside other comments. Thus 'Hidden with '$PROBHIDE is not an acceptable directive.

You can put multiple directives on a line by separating them with a colon ( : ). Usually you would just put a single directive on a line, though.

1. PROBHIDE – Hide problems

The PROBHIDE directive hides problems of one or more types. Use it to disable problem detection at locations you don't want to change. Syntax:

'$PROBHIDE problemtype [problemtype...] [EXCEPT problemtype [problemtype...] ] [BEGIN]

PROBHIDE examples

A procedure does not require an error handler. Hide the "Error handler missing" problem for that procedure but show it everywhere else.

  '$PROBHIDE NO_ERROR_HANDLER

 Sub ProcedureWithoutErrors()

  ' This procedure needs no error handling



  End Sub

A procedure is dead at the moment but you don't want to remove it. You want to flag the procedure live without hiding any other dead code related problems. Notice how the syntax affects just the next line (the line with the dead procedure problem). Any other lines are not affected and the dead variable problem will thus be shown. Notice that hiding a dead procedure problem propagates down along the call tree.

  '$PROBHIDE DEAD

 Sub DeadProcedure()

  ' This procedure should be kept



 Dim DeadVariable As Integer
  ' But this variable should be removed



  Call ChildProcedure



  End Sub



 Sub ChildProcedure()

  ' This procedure shows up as live

  ' because it's called by DeadProcedure

  ' which was flagged live

  End Sub

Hide all problems in a range of code between BEGIN and END. Keep showing dead code related problems at all locations.

  '$PROBHIDE ALL EXCEPT DEAD BEGIN

 code with problems

 to hide

  '$END

 show all problems after this point

A Goto statement is acceptable at a certain location. Hide the problem on that specific line. There are two alternative ways to write the same directive: write the comment either on the line with the problem or add it on a new line immediately above it.

2. PROBSHOW – Show problems hidden by PROBHIDE

'$PROBSHOW <problem type> [<problem type> ...] [EXCEPT <problem type> [<problem type> ...] ] [BEGIN]

The syntax is similar to that of PROBHIDE. You should use PROBSHOW only inside a PROBHIDE BEGIN...END block.

PROBSHOW example

Hide all problems in a class, but enable dead code problems after class declarations to remove any dead procedures and local variables/constants.

'$PROBHIDE ALL BEGIN

Dim x, y

Const PI = 3.14

Declare Function ChooseFont Lib "comdlg32.dll" Alias "ChooseFontA" (pChoosefont As CHOOSEFONT) As Long



'$PROBSHOW DEAD BEGIN

Sub MyProc ( x )



End Sub

'$END

...

3. END – Close a BEGIN..END block

4. USED (obsolete)

This obsolete directive marks things as being in use. It is provided for only backward compatibility. Older Project Analyzer versions used to support it until 1999. You should not use this directive any more since future versions may not support it. Project Analyzer replaces it internally as follows:

USED => PROBHIDE DEAD

USED PROC => PROBHIDE DEAD_PROC

USED PARAM => PROBHIDE DEAD_PARAM

Scope

How do you work out where to put the directives and what code they will then affect? First of all, look at the problem icons in the hypertext view of Project Analyzer. Your comment directives should be placed to affect the lines with the icons. If your directives don't seem to work, pay attention to the exact line and the scope you're using. There are several alternative ways to define the scope of a comment directive.

A. Immediately above the problem icon

A directive on an otherwise empty line affects the immediately following line. In this case, MyVariable is not reported dead.

B. On the line with the problem icon

A directive after the code on a given line affects that line. In this case, MyVariable is not reported dead.

C. BEGIN..END block

The directive affects all code starting with BEGIN up to the next '$END. In this case, MyVariable and Text are not reported dead.

You may nest multiple BEGIN...END blocks. You may also omit the END directive. In that case, the block continues up to end-of-file.

D. WHERE: Lines containing given text

To hide problems on specific lines use the WHERE scope. The directive affects all lines containing expression. Examples:

The expression is case insensitive. It can be any string, but a colon (:) cannot be included. If expression contains multiple words, spaces between the words are significant. Spaces before and after expression don't have an effect. The WHERE directive can exist in any file. It affects all files in the analysis (similar to ANYWHERE).

E. IN_THIS_FILE: Entire file

To hide problems in an entire file, regardless of which line they affect, use the IN_THIS_FILE scope. You can put this keyword anywhere in the file to make a directive apply to all lines in the file, also lines before the directive. IN_THIS_FILE is useful when you need to hide problems at the start of a file, but that area is reserved for comment block and a directive would not look good there.

F. IN_THIS_PROJECT: Entire project

To hide problems in the whole project, regardless of which file they appear in, use the IN_THIS_PROJECT scope. You can put this keyword anywhere to make a directive apply to all files of the project. This directive may be overridden in each individual file by scopes A–D. On the other hand, it will override ANYWHERE directives.

IN_THIS_PROJECT is useful when your regular coding standards don't apply to a certain project. As an example, '$PROBHIDE STYLE IN_THIS_PROJECT ignores style rules in the project while they are still enabled for other projects.

G. ANYWHERE: All files

To hide problems in all files, regardless of which project the file may be in, use the ANYWHERE scope. You can put this keyword wherever you like to make a directive apply to all files of the analysis. Note that this directive may be overridden in each individual file by scopes A-D and each individual project by scope E.

Use ANYWHERE instead of IN_THIS_PROJECT when you want to control an entire analysis instead of an individual project. It will cover all files when analyzing one or more solutions, project groups or a directory tree.

Precedence

Scope ANYWHERE has the lowest precendence, IN_THIS_PROJECT is in the middle. The other scopes are on the highest precedence level.

To avoid any conflicts put all ANYWHERE directives in a single file. The precedence of conflicting ANYWHERE directives mixed in different files is undefined. The same goes for conflicting IN_THIS_PROJECT directives. You may want to put them in a global declarations file to make them easier to manage.

Directives coming later override directives on preceding lines. Thus, you can always override a previous directive by writing a new directive after it. As an example, you can have a '$PROBHIDE ALL early in the file and then override it with a '$PROBSHOW DEAD IN_THIS_FILE later. In this case, the later PROBSHOW has higher precedence and will override the PROBHIDE for dead code problems.

If you don't want a directive to override others, use the EXCEPT keyword. As an example, '$PROBSHOW ALL EXCEPT STYLE IN_THIS_FILE would show all other problems but it wouldn't have any effect on style related problems. Style problems would be entirely controlled by other directives, should any exist.

Examples

Precedence examples

general.bas:
'$PROBHIDE STYLE IN_THIS_PROJECT causes style related problems to be ignored in a project where style doesn't matter.

important.bas:
'$PROBSHOW ALL IN_THIS_FILE enables all problems in important.bas regardless of any setting in general.bas.

WHERE examples

This example disables the rules "Function/constant/variable with type character" on lines containing an ampersand (&). The line exceptionally allows declarations such as Dim Number& instead of Dim Number As Long, but only with & and no other type characters.

This example allows the use of the Goto keyword to jump to Quit. In addition, it also allows a single-line If..Then to exist on the same line. Code that would be allowed: If somecondition Then Goto Quit

This example allows a single line If..Then but not a single-line If..Then..Else.

Turning dead procedures alive

You can flag a procedure as live code in several ways. The recommended way is by '$PROBHIDE DEAD or '$PROBHIDE DEAD_PROC.

Hiding a dead procedure problem flags the affected procedure as live code. What is more, also the procedures it calls are flagged live, down along the entire call tree. Consider a call tree such as A → B → C. If you flag A as a live procedure, also B and C will show up as live. Thus, using '$PROBHIDE to hide one dead procedure also propagates downwards in the call tree. You can mark just the root procedure as live and Project Analyzer will take care of the rest. The call tree propagation only affects procedures. Variables, classes or any other code is not affected.

Ignore comment directives

There's a command line option that tells Project Analyzer to ignore all comment directives. Start Project Analyzer like this:

Use this option to detect all problems even if $PROBHIDE has been used to hide some of them. You can also use it if your code happens to have existing lines that start with '$PROBHIDE, '$PROBSHOW or '$USED.

PROJECT_ANALYZER constant

PROJECT_ANALYZER is a special built-in "compiler constant" in Project Analyzer. It allows you to tweak the code analysis. You can add missing definitions or exclude sections from being analyzed. During code analysis, PROJECT_ANALYZER = True. In Visual Basic, PROJECT_ANALYZER is not defined and thus evaluates to False.

PROJECT_ANALYZER is not a comment directive. It's a constant for #If..#ElseIf..#Else..#End If.

The following trick lets you write code that won't be analyzed at all. You can also add code for the purpose of code analysis only.

The next trick lets you run your code with late binding, but analyze it as if it were early-bound.

Versions. Suppose your code compiles two versions of the same program. To compile version 1, you set VERSION=1 in your project settings. To compile the other version, you set VERSION=2. Now, you exclude and include sections of code with #If to compile the 2 versions. This causes problems in the analysis, because Project Analyzer should read your code in its entirety and not just one of the versions.

The following trick lets you compile several versions of your program while keeping Project Analyzer analyzing the full code.

Tweak metrics. Suppose you have a large section of generated code that shouldn't be counted in the logical lines of code (LLOC) metric, because it exaggerates the size of your program. By adding the generated code in a False block you exclude it from LLOC, even when it still compiles and runs just fine.