Comments on: Dion & Steve’s RoR Chat http://www.intertent.net/2008/01/30/dion-steves-ror-chat/ Technology in a pita pocket - fur reelz! Tue, 06 Jan 2009 01:03:00 +0000 http://wordpress.org/?v=2.3.3 By: Yossi Kreinin http://www.intertent.net/2008/01/30/dion-steves-ror-chat/#comment-11 Yossi Kreinin Tue, 05 Feb 2008 21:27:27 +0000 http://www.intertent.net/2008/01/30/dion-steves-ror-chat/#comment-11 It really depends on what the popular libraries are doing, or what the libraries used in a company X are doing. I don't know; I think that a template or six tend to sneak in these days, and that you absolutely have to support the STL container types (which is probably on the border of "possible without having a full C++ front-end"). When I think about it, the strength of C++ lock-in is amazing. I mean, you have Wine on Linux, but you don't have that easy a way to interface with C++ libraries from Java. Captures one's imagination. It really depends on what the popular libraries are doing, or what the libraries used in a company X are doing. I don’t know; I think that a template or six tend to sneak in these days, and that you absolutely have to support the STL container types (which is probably on the border of “possible without having a full C++ front-end”).

When I think about it, the strength of C++ lock-in is amazing. I mean, you have Wine on Linux, but you don’t have that easy a way to interface with C++ libraries from Java. Captures one’s imagination.

]]> By: Jeremy http://www.intertent.net/2008/01/30/dion-steves-ror-chat/#comment-10 Jeremy Sat, 02 Feb 2008 08:37:58 +0000 http://www.intertent.net/2008/01/30/dion-steves-ror-chat/#comment-10 The fact it's impossible to define a portable interface in C++ is really sad. But what if we limited ourselves to linking with simple classes that only accepted C types? Or some other similar subset of C++? But if some subset was compatible, would that help? Would it even be interesting? The fact it’s impossible to define a portable interface in C++ is really sad.

But what if we limited ourselves to linking with simple classes that only accepted C types? Or some other similar subset of C++?

But if some subset was compatible, would that help? Would it even be interesting?

]]> By: Yossi Kreinin http://www.intertent.net/2008/01/30/dion-steves-ror-chat/#comment-9 Yossi Kreinin Fri, 01 Feb 2008 22:07:28 +0000 http://www.intertent.net/2008/01/30/dion-steves-ror-chat/#comment-9 This link-compatibility-with-C++-thing, you just can't do it, I mean it's amazing that Steve Y suggests it. I've read most of what he's published and he said he wasn't using STL somewhere; well, it shows in this case, 'cause in STL, most of your problems are compile-compatibility - to call it, your compiler has to parse the templates and then implement the C++ template instantiation, specialization selection and overload resolution rules, all before you get to link time stuff. And STL is a tame example by modern standards. To interact with C++, you ought to be C+=2 - a full superset of C++. D rules. C++ is good for fertilizer. D programmers around the world, unite! :) This link-compatibility-with-C++-thing, you just can’t do it, I mean it’s amazing that Steve Y suggests it. I’ve read most of what he’s published and he said he wasn’t using STL somewhere; well, it shows in this case, ’cause in STL, most of your problems are compile-compatibility - to call it, your compiler has to parse the templates and then implement the C++ template instantiation, specialization selection and overload resolution rules, all before you get to link time stuff. And STL is a tame example by modern standards. To interact with C++, you ought to be C+=2 - a full superset of C++.

D rules. C++ is good for fertilizer. D programmers around the world, unite! :)

]]>