[boost] Re: TR2 and C++0x heads up

From: Thorsten Ottosen (nesotto_at_[hidden])
Date: 2005-04-27 16:33:19

"Andrei Alexandrescu (See Website For Email)" 
<SeeWebsiteForEmail_at_[hidden]> wrote in message 
news:d4ov22$86i$1_at_sea.gmane.org...
| David Abrahams wrote:

| > Maybe, not but it's not a long way from what you *actually wrote* to
| > "evil spirits."  Starting from "conflict of interest" connected to
| > lobbying and voting and then proceeding to a suggestion that people
| > are claiming something they don't "really really believe" and
| > "convenient" forgetting of crucial facts, there's a lot in there to
| > take as disparaging.
|
| Yeah, "conflict of interest" isn't the word because there's not a real
| "interest" there, but rather a mere subjective opinion. But then let me
| ask this again: if binary compatibility is assigned such a weight on
| shared_ptr, then why isn't it a concern for other obvious candidates in
| the standard library, such as strings and vectors? Shouldn't we measure
| those guys by the same measure? Again, binary compatibility is a very
| useful feature, but it should only be assigned so much importance.

Let me wrap up my reasons for not showing that much interest in the proposal:

1. is it worth the effort to standardize it?
2. does it help many users?
3. how does it compare to adding GC instead to the language?
4. won't experts just go to boost to pick up this thing?
5. can many, many smart pointers hurt C++, because it makes interoperability 
harder?
6. can all these different pointers be hard to use in exception-safe code 
because they all behave
    slightly different; some might throw on assignment, some might not etc.

-Thorsten