[boost] Re: new approach to smart ptrs -- no new()s is good news

From: Beman Dawes (beman_at_[hidden])
Date: 2000-01-30 20:33:08

At 06:58 PM 1/30/00 -0500, Dave Abrahams wrote:

>on 1/30/00 3:18 PM, Gavin Collings at gcollings_at_[hidden]
wrote:
>
>> OK, I have the test set up anyway, so if you can supply me with
your
>> shared_in_ptr (to gcollings_at_[hidden], or here) I'll run
the
>> above test on all four for comparison (shared, linked, shared_in
and
>> cyclic).  Give me a couple of days though.  I only have access to
>> Visual C++ 6.0 SP 3 and gcc 2.95 on NT.
>
>Someone had better try another compiler. Both of these have
essentially a
>setjump/longjump exception-handling implementation and so will
undoubtedly
>display steep overheads in places where they could be avoided. Is
the
>current "best/latest" test code available in the vault? Someone
should try
>this with CodeWarrior.

Gavin,

After you run your test, it certainly would be nice if you would
place a copy of all files in the Boost eGroups vault.  I expect
people will be interested in running the tests on several compilers
and/or hardware platforms.

I always wonder about older and embedded systems.  Most of the people
who report timings on boost seem to have fairly fast hardware.  I
wish I had hung onto an old 486 machine just to run timings.

--Beman