Comments on: Stupid Reference Tricks, Part 2 http://www.donniepinkston.net/2007-01-19/stupid-reference-tricks-part-2.html (there are many more like it, but this one is mine) Tue, 06 Jan 2009 09:39:14 +0000 http://wordpress.org/?v=2.5.1 By: donnie http://www.donniepinkston.net/2007-01-19/stupid-reference-tricks-part-2.html#comment-2512 donnie Thu, 25 Jan 2007 18:00:53 +0000 http://www.donniepinkston.net/2007-01-19/stupid-reference-tricks-part-2.html#comment-2512 As you say, "no one who's anyone" says that this is a good pattern. It's clearly an anti-pattern. Nonetheless, I have come across people using it on large-scale C++ coding projects <em>twice</em> already here at Caltech, and I've only been here half as long as you were. There are standard C++ APIs that do this kind of thing, too. For example, if you look at the <code>std::string</code> <code>at()</code> function, it can also be used as the target of assignment. I think that what may happen is that people see this, and they think it's a good approach in general. They don't realize that providing a non-const reference to an internal data member is to be used <em>very</em> infrequently, and only in very particular situations. Anyway, by virtue of the simple fact that people are using it, somebody somewhere clearly thinks it's a good idea. They just don't realize that it completely violates encapsulation. If they did, hopefully they would stop using it. All I can do is push back the boundaries of ignorance one class at a time... ;-) As you say, “no one who’s anyone” says that this is a good pattern. It’s clearly an anti-pattern. Nonetheless, I have come across people using it on large-scale C++ coding projects twice already here at Caltech, and I’ve only been here half as long as you were.

There are standard C++ APIs that do this kind of thing, too. For example, if you look at the std::string at() function, it can also be used as the target of assignment. I think that what may happen is that people see this, and they think it’s a good approach in general. They don’t realize that providing a non-const reference to an internal data member is to be used very infrequently, and only in very particular situations.

Anyway, by virtue of the simple fact that people are using it, somebody somewhere clearly thinks it’s a good idea. They just don’t realize that it completely violates encapsulation. If they did, hopefully they would stop using it.

All I can do is push back the boundaries of ignorance one class at a time… ;-)

]]> By: Ben http://www.donniepinkston.net/2007-01-19/stupid-reference-tricks-part-2.html#comment-2511 Ben Thu, 25 Jan 2007 07:05:39 +0000 http://www.donniepinkston.net/2007-01-19/stupid-reference-tricks-part-2.html#comment-2511 Where do people say this is a good pattern? I'm confused about the whole example. This doesn't look like modern C++ code that anyone would be writing these days... well, at least, anyone who's anyone. :) Where do people say this is a good pattern?

I’m confused about the whole example. This doesn’t look like modern C++ code that anyone would be writing these days… well, at least, anyone who’s anyone. :)

]]>