Heh, but it's not entirely made up. Mentor Graphics really did epically fail when they tried that rewrite and it was a major event in C++ history, seeing as how they had (informal?) support from Bell Labs etc. etc.
The number one lesson learned from that experience according to the author and my memory was "Make your class hierarchies directed acyclic graphs" ... which of course wasn't a C++ issue at all. As the first widely adopted OO language C++ gets a lot of mostly undeserved flack for being the medium in which a lot of us learned the OO paradigm.
A bit like how all those companies that oversold expert systems in the '80s blamed Lisp for their problems, causing both AI and Lisp winters.
A question for C++ coders (I'm not one myself): how much of what he says about the hidden costs of C++ ring true? (I know this is a parody, but successful humor is often rooted in some element of truth.)
Well, there's certainly a lot of "bloated" code. I'm using quotation marks because I'm coming from a point of view where 'lean and mean' is golden - I like beautiful code iff the implementation of the pretty abstractions is equally beautiful.
Also, the language is so huge, that it's practically impossible to learn all of it in any reasonably short amount of time. Most people learn to use the features they need, and skip the rest entirely. That has led to a lot of controversy about the language (see the weekly rants about C++, and some that popped up last week, they're pretty funny).
I've also found that proper design is indeed a more demanding task in C++ than C. I think that's reasonable, with all the features at work, you need to be sure you're not using a cannon to kill mosquitoes. Also, I got this colleague who has said the same thing written in the article about references and pointers. That didn't feel like a joke to me. :(
But the idea of Unix++? Man, I blew orange juice all over my desk. I can't help but think it might be true. :-D
8 comments
[ 4.2 ms ] story [ 30.1 ms ] threadThis very useful book covers it a bit in the intro: http://www.amazon.com/Taming-Pattern-Classes-Persistence-Pro...
The number one lesson learned from that experience according to the author and my memory was "Make your class hierarchies directed acyclic graphs" ... which of course wasn't a C++ issue at all. As the first widely adopted OO language C++ gets a lot of mostly undeserved flack for being the medium in which a lot of us learned the OO paradigm.
A bit like how all those companies that oversold expert systems in the '80s blamed Lisp for their problems, causing both AI and Lisp winters.
Also, the language is so huge, that it's practically impossible to learn all of it in any reasonably short amount of time. Most people learn to use the features they need, and skip the rest entirely. That has led to a lot of controversy about the language (see the weekly rants about C++, and some that popped up last week, they're pretty funny).
I've also found that proper design is indeed a more demanding task in C++ than C. I think that's reasonable, with all the features at work, you need to be sure you're not using a cannon to kill mosquitoes. Also, I got this colleague who has said the same thing written in the article about references and pointers. That didn't feel like a joke to me. :(
But the idea of Unix++? Man, I blew orange juice all over my desk. I can't help but think it might be true. :-D