- I liked the "hacking" diagram on second slide. This comes up a lot. - It's interesting that you map classes to C++ concepts. Jeremy Siek's work on C++ concepts and on G++ actually argued that concepts should be inspired by the Haskell class system. - Martin questions your statement that "the users like C++", and noted that we should work on encouraging them to raise their level of abstraction. My response to this is that some users insist on having the product in a particular language, for a variety of reasons. These reasons can relate to integration with existing platforms, certification, maintainability, and avoiding dependence on a new technology. - You can say that your approach partly involves a refinement steps where more information is added by the user to the Haskell to C++.
A thorough comparison of C++ concepts with Haskell type classes can be found in the WGP 2008 paper (BibTeX) by Bernardy, Jansson et al.
ReplyDeleteDaniel's talk:
ReplyDelete- I liked the "hacking" diagram on second slide. This comes up a lot.
- It's interesting that you map classes to C++ concepts. Jeremy Siek's work on C++ concepts and on G++ actually argued that concepts should be inspired by the Haskell class system.
- Martin questions your statement that "the users like C++", and noted that we should work on encouraging them to raise their level of abstraction. My response to this is that some users insist on having the product in a particular language, for a variety of reasons. These reasons can relate to integration with existing platforms, certification, maintainability, and avoiding dependence on a new technology.
- You can say that your approach partly involves a refinement steps where more information is added by the user to the Haskell to C++.
Unfortunately concepts were dropped from C++ (for now).
ReplyDeleteAnyway, another interesting paper on Haskell/C++ from BoostCon'09:
An experimental domain specific language for template metaprogramming - pdf