The dangling else ambiguity

Ambiguity in programming language constructs
- A construct in a programming language is ambiguity, if the construct can be interpreted in more than 1 way
- Some programming languages (notably: C, C++, Java) have an if statement that is ambiguous - i.e., the statement can be interpreted in MORE THAN 1 WAY
- This dangling-else ambiguity problem was first discovered around 1960 in the ALGOL60 programming language

The Dangling-Else ambiguity

Suppose we want to write an FONT color="blue"> conditional statement to determine the shipping rate to different countries of the world:

To ship a 1 lb packet within the US, the cost is $5 , except for Hawaii
To ship to Hawaii (which is very remote), the cost $10
To ship a 1 lb packet outside the US, the cost is $20

The following is an if-statement that computes the "shippingCharge":

double shippingCharge; shippingCharge = 5.00; // Assume the cost is $5 if ( country is equal to "USA") ) if ( state is equal to "HI" ) shippingCharge = 10.00; // Hawaii will cost $10 else shippingCharge = 20.00; // Outside USA is $20

There are two ways to read the above program syntactically correctly:

First Way	Second Way
double shippingCharge; shippingCharge = 5.00; if ( country equals "USA" ) if ( state equals "HI" ) shippingCharge = 10.00; else shippingCharge = 20.00;	double shippingCharge; shippingCharge = 5.00; if ( country equals "USA" ) if ( state equals "HI" ) shippingCharge = 10.00; else shippingCharge = 20.00;

First Way

Second Way

double shippingCharge; shippingCharge = 5.00; if ( country equals "USA" ) if ( state equals "HI" ) shippingCharge = 10.00; else shippingCharge = 20.00;

You can see the if-else association better after indenting it properly:

First Way	Second Way
double shippingCharge; shippingCharge = 5.00; if ( country equals "USA" ) if ( state equals "HI" ) shippingCharge = 10.00; else shippingCharge = 20.00;	double shippingCharge; shippingCharge = 5.00; if ( country equals "USA" ) if ( state equals "HI" ) shippingCharge = 10.00; else shippingCharge = 20.00;

First Way

Second Way

double shippingCharge; shippingCharge = 5.00; if ( country equals "USA" ) if ( state equals "HI" ) shippingCharge = 10.00; else shippingCharge = 20.00;

Clearly, the two interpretations are syntactically correct
Futhermore the two interpretations result in different program execution:
1. The First Way interpretation result in computing the cost in the way that we wanted
2. The Second Way interpretation computes a shipping charge of $5 for international shipments, $10 for Hawaii, and $20 for other states in the US !!!
Clearly, ambiguity in computer programs is unaceptable (the computer could pick the wrong interpretation )

Resolving the Dangling-Else ambiguity

There are a number of techniques to resolve ambiguity in a language:
1. The elegant way is to design languages that are NOT ambiguous
2. The simple-minded way is to design rules to resolve ambiguity
(The latter method is used when the resolve an ambiguity when an ambiguity is discovered after a language has been designed and in use)

How to design languages that avoid the dangling-else ambiguity

A slightly more complex syntax rule is used for the conditional statements that includes a terminal "fi" symbol :

if ( boolean expression ) one-or-more-statement fi
if ( boolean expression ) one-or-morestatement else one-or-more-statement fi

Using the new syntax rule the above programs will be written as:

First Way in new syntax Second Way in new syntax

double shippingCharge; shippingCharge = 5.00; if ( country equals "USA" ) if ( state equals "HI" ) shippingCharge = 10.00; fi else shippingCharge = 20.00; fi

double shippingCharge; shippingCharge = 5.00; if ( country equals "USA" ) if ( state equals "HI" ) shippingCharge = 10.00; else shippingCharge = 20.00; fi fi

There is ONLY ONE WAY to read each of the program segment above !! (try it !!!)

Java, C and C++ have chosen to resolve the Dangling-Else ambiguity uses the following rule:

An else keyword always associates with the nearest preceeding if keyword that does NOT cause a syntax error

First Way in new syntax	Second Way in new syntax
double shippingCharge; shippingCharge = 5.00; if ( country equals "USA" ) if ( state equals "HI" ) shippingCharge = 10.00; fi else shippingCharge = 20.00; fi	double shippingCharge; shippingCharge = 5.00; if ( country equals "USA" ) if ( state equals "HI" ) shippingCharge = 10.00; else shippingCharge = 20.00; fi fi

Therefore, the C/C++ compiler will read this program as follows:

double shippingCharge; shippingCharge = 5.00; if ( country.equals("USA") ) if ( state.equals("HI") ) shippingCharge = 10.00; else shippingCharge = 20.00;

It will associate the else keyword with the second if keyword

$64,000 Question:

Then how do you write the following C++ program:

double shippingCharge; shippingCharge = 5.00; if ( country equals "USA" ) if ( state equals "HI" ) shippingCharge = 10.00; else shippingCharge = 20.00;

You need to use { } to introduce syntax errors when C++ tries to associate the else with the second if:

So, the proper way to write the program is:
double shippingCharge; shippingCharge = 5.00; if ( country.equals("USA") ) { if ( state.equals("HI") ) shippingCharge = 10.00; // Shipping to Hawaii is $10 } else shippingCharge = 20.00; // Shipping to outside USA is $20

Associating the else with the second if will now result in a syntax error !!!