boost::shared_ptr and NULL
In this function declaration, is there a simpler way to express a null shared pointer as the default argument value? void foo( const boost::shared_ptr<T> & p = boost::shared_ptr<T>( ) ); In this function call, is there a simpler way to express a null shared pointer as the actual argument? foo( boost::shared_ptr<T>( ) ); ----------------------------------------------------------------------- Andrew R. Thomas-Cramer Work: artc@prism-cs.com http://www.prism-cs.com/ Home: artc@execpc.com http://www.execpc.com/~artc/ ----------------------------------------------------------------------- [Non-text portions of this message have been removed]
From: "Andrew R. Thomas-Cramer"
In this function declaration, is there a simpler way to express a null
void foo( const boost::shared_ptr<T> & p = boost::shared_ptr<T>( ) );
In this function call, is there a simpler way to express a null shared
shared pointer as the default argument value? pointer as the actual argument?
foo( boost::shared_ptr<T>( ) );
No to both. In 1.27 boost::shared_ptr can be constructed from the literal 0, but the constructor is explicit. In 1.28, the literal 0 is no longer a valid argument to the constructor.
----- Original Message -----
From: "Peter Dimov"
From: "Andrew R. Thomas-Cramer"
In this function declaration, is there a simpler way to express a null
void foo( const boost::shared_ptr<T> & p = boost::shared_ptr<T>( ) );
In this function call, is there a simpler way to express a null shared
shared pointer as the default argument value? pointer as the actual argument?
foo( boost::shared_ptr<T>( ) );
No to both. In 1.27 boost::shared_ptr can be constructed from the literal 0, but the constructor is explicit. In 1.28, the literal 0 is no longer a valid argument to the constructor.
Has there been any discussion in the past about supporting a more concise/readable value? E.g., the second call below appears more readable to me. (This particular example relies on "boost::null" being a class with a default constructor, and boost::shared_ptr providing a non-explicit constructor accepting a boost::null instance as an argument; this may be flawed.) foo( boost::shared_ptr<MyVeryLongClassName>() ); foo( boost::null() );
From: "Andrew R. Thomas-Cramer"
Has there been any discussion in the past about supporting a more concise/readable value? E.g., the second call below appears more readable
to me.
(This particular example relies on "boost::null" being a class with a default constructor, and boost::shared_ptr providing a non-explicit constructor accepting a boost::null instance as an argument; this may be flawed.)
foo( boost::shared_ptr<MyVeryLongClassName>() ); foo( boost::null() );
No, I don't recall any discussion about it. I haven't encountered this problem myself; I typically overload foo so that the call would be simply foo(); This doesn't extend well to multiple optional shared_ptr arguments, of course.
As an alternative, I usually find that the "Introduce Null Object" refactoring results in simpler and more readable code overall. Perhaps not as concise though.
----- Original Message -----
From: Andrew R. Thomas-Cramer
To: Boost-Users@yahoogroups.com
Sent: Monday, May 13, 2002 12:50 PM
Subject: Re: [Boost-Users] boost::shared_ptr and NULL
----- Original Message -----
From: "Peter Dimov"
From: "Andrew R. Thomas-Cramer"
In this function declaration, is there a simpler way to express a null
void foo( const boost::shared_ptr<T> & p = boost::shared_ptr<T>( ) );
In this function call, is there a simpler way to express a null shared
shared pointer as the default argument value? pointer as the actual argument?
foo( boost::shared_ptr<T>( ) );
No to both. In 1.27 boost::shared_ptr can be constructed from the literal 0, but the constructor is explicit. In 1.28, the literal 0 is no longer a valid argument to the constructor.
Has there been any discussion in the past about supporting a more concise/readable value? E.g., the second call below appears more readable to me. (This particular example relies on "boost::null" being a class with a default constructor, and boost::shared_ptr providing a non-explicit constructor accepting a boost::null instance as an argument; this may be flawed.) foo( boost::shared_ptr<MyVeryLongClassName>() ); foo( boost::null() ); Yahoo! Groups Sponsor ADVERTISEMENT Info: http://www.boost.org Wiki: http://www.crystalclearsoftware.com/cgi-bin/boost_wiki/wiki.pl Unsubscribe: mailto:boost-users-unsubscribe@yahoogroups.com Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service. [Non-text portions of this message have been removed]
As an alternative, I usually find that the "Introduce Null Object" refactoring results in simpler and more readable code overall. Perhaps not as concise
----- Original Message -----
From: "Sean Shubin"
From the book "Refactoring"
As an alternative, I usually find that the "Introduce Null Object" refactoring results in simpler and more readable code overall. Perhaps not as concise
Sometimes you can simplify a conditional expression such as this:
class Foo {
void foo() {
...
}
};
...
x = 0;
...
if(x == 0) handleNullCondition(); //conditional expression
else x->foo();
with something like this:
class Foo {
void foo() {
...
}
};
class NullFoo {
void foo() {
handleNullCondition();
}
};
...
x = new NullFoo(); //instead of x = 0
...
x->foo(); //conditional expression goes away
//polymorphism handles it instead
So if I only have one check for class Foo being null, the code becomes bigger (I create a new class). But if I had many checks all over the place, it actually simplifies the code (a lot of if statements go away).
----- Original Message -----
From: Andrew R. Thomas-Cramer
To: Boost-Users@yahoogroups.com
Sent: Tuesday, May 14, 2002 7:11 AM
Subject: Re: [Boost-Users] boost::shared_ptr and NULL
----- Original Message -----
From: "Sean Shubin"
participants (3)
-
Andrew R. Thomas-Cramer -
Peter Dimov -
Sean Shubin