http://www.roguewave.com/blog/c-in-2008/ Rogue Wave's C++ blog community shares thoughts on enterprise C++ development, Rogue Wave's products, and application development challenges. Wed, 17 Mar 2010 05:39:34 +0000 http://wordpress.org/?v=2.3.1 http://www.roguewave.com/blog/c-in-2008/#comment-1275 Nathaniel Christen Thu, 07 May 2009 16:12:43 +0000 http://www.roguewave.com/blog/c-in-2008/#comment-1275 One underappreciated C++ strength is the detail and flexibility of its type system. Even with the new generics in Java and C#, C++ templates are more powerful. With or without templates, C++ forces you to think about the structure and semantics of types, which is I think a healthy discipline in an era when data originating from applications written in many different programming languages and environments must interoperate. While XML, SQL, JSON, etc., have their uses, the mere existence of standardized structures for transmitting data is not akin to true data transparency, since algorithms need to be rolled out to to encode and decode data into and out of XML et. al. (and LINQ or any platform-specific object-transfer system are not really "transparent" for obvious reasons). The fact that fully automatic algorithms for translating C++ objects to or from an XML stream -- and reconstituting them in another language, say -- do not exist, demonstrates that XML is not a "real" data transmission language. As far as I can tell, interesting theoretical work on generic programming, "type theory", etc., is almost exclusively done in or about C++ relative to other mainstream languages. Good research is done in or about Haskell, Lisp (specifically Common Lisp Object System), etc., but C++ is really the only place where sophisticated generic programming research meets real-world, robust applications, IMO. One underappreciated C++ strength is the detail and flexibility of its type system. Even with the new generics in Java and C#, C++ templates are more powerful. With or without templates, C++ forces you to think about the structure and semantics of types, which is I think a healthy discipline in an era when data originating from applications written in many different programming languages and environments must interoperate. While XML, SQL, JSON, etc., have their uses, the mere existence of standardized structures for transmitting data is not akin to true data transparency, since algorithms need to be rolled out to to encode and decode data into and out of XML et. al. (and LINQ or any platform-specific object-transfer system are not really “transparent” for obvious reasons). The fact that fully automatic algorithms for translating C++ objects to or from an XML stream — and reconstituting them in another language, say — do not exist, demonstrates that XML is not a “real” data transmission language. As far as I can tell, interesting theoretical work on generic programming, “type theory”, etc., is almost exclusively done in or about C++ relative to other mainstream languages. Good research is done in or about Haskell, Lisp (specifically Common Lisp Object System), etc., but C++ is really the only place where sophisticated generic programming research meets real-world, robust applications, IMO.

]]>