From: Greg Colvin (Gregory.Colvin_at_[hidden])
Date: 2002-08-29 10:19:45
At 08:39 AM 8/29/2002, Björn Karlsson wrote:
>> From: Philippe A. Bouchard [mailto:philippeb_at_[hidden]]
>> C# will fit better your needs then.
>This (unnecessary) remark implies that you haven't understood the answer to
>the question. See below.
>> "Greg Colvin" <Gregory.Colvin_at_[hidden]> wrote:
>> > I prefer new T
>This sentence represents a (remarkable) compression of experiences gained by
>thousands, over decades. Had my blade of eastern philosophy been sharper, I
>would make this my mantra. Thus, there is no need for high-pitched notes
>(just this once, pun intended).
>That you are seeking different answers, and smarter/smaller/faster pointers,
>is fine. There should still be room for a collegial tone.
Sorry, no offense intended, just humor. And yes, there are good and well
known reasons to avoid requiring placement new if possible. So ...
1) I have yet to see an explanation as to why this kind of pointer, as
opposed to this particular implementation, *requires* a placement new.
2) If it does require a placement new, I have already pointed out that
a smaller, faster, simpler, and more portable implementation might be
had by taking advantage of the size-segregated storage techniques in
boost::pool and other places.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk