David's Blog: May 2009 Archives

I am reading this document about the C# language.

Here are a list of features in C# that improve on how C++ works…

C++	C#
Reference to a class in another file of a project requires an include file	No include file needed
Reference to a class in an external DLL requires an include file and the DLL’s .LIB file passed to the linker	No .LIB file needed. Add a reference to the DLL in the build(example: csc /r:external.dll helloworld.cs)
Exported symbols must use declspec or .DEF file	No declspec or .DEF file. Public symbols are exported, private/protected are not. ”internal” symbols are only available within module.
Referencing a class/structure before it is declared requires forward declaration	Forward declaration not needed. Declaration order is insignificant.
Switching to Unicode/16-bit text requires: Different entry point (main/wmain) Prefixing text with “L” or “_T” New string manipulation functions	Unicode is the native string format
NULL is a concept, not part of the language. It is typically defined as 0 and thus is the equivalent of 0.	null is a keyword. It is not equivalent to 0.
Basic data types (int, float, char, etc.) cannot accept a “null” value. Only pointers can have a “null” value.	All C# types can accept null as a value For type T with null support, use type T? For example, int with null support is int?
No support for “foreach”	foreach keyword for iterating over a collection
Support for try and catch. No support for "finally”	try-catch-finally support. finally is always called, regardless of exception or not
No notification of overflow operations. For example: unsigned x = UINT_MAX; x++; // x == 0 now, no exception thrown	Using checked/unchecked you can get overflow or exception thrown behavior
No native support for critical sections	lock keyword for critical sections
For switch statements, break is optional.	break is required (to prevent bugs from accidentally forgetting to add break)
No built-in documentation	XML documentation via “///”- prefixed comment. Used by IntelliSense in Visual Studio
Cannot split the definition of a class across multiple files	Partial classes allow breaking up a source file (useful when multiple people are working on a class or when part of a class is auto-generated)
No support for mixing versions of libraries	Can mix library versions without breaking existing code.
Developer must manage deletion of memory and potential issues from reading/writing to invalid memory	Garbage collection: No memory leaks Can’t read uninitialized variables Can’t index past array boundaries
Types do not share a common root	All types derived from base “object”