Re: [boost] operator++

From: David Abrahams (abrahams_at_[hidden])
Date: 2000-10-12 17:05:10

But using 2.95 when 2.95.2 is known to have fewer bugs is just asking for
trouble.

why-would-anyone-do-that-ly y'rs,
dave

----- Original Message -----
From: "Gary Powell" <Gary.Powell_at_[hidden]>
To: <boost_at_[hidden]>
Sent: Thursday, October 12, 2000 5:30 PM
Subject: RE: [boost] operator++

> Hi,
>   Let me first apologize for sending raw code around. I hit reply when I
> should have cut and pasted his address. So if anyone other then Andras is
> interested in pursuing these proxies let me know and I'll send the updates
> directly to you.
>
> Second. In my experimenting I have uncovered several annoying bugs in gcc
> 2.95 which leads me to suspect using proxies will not gain any efficiency
> unless your compiler is very good.
>
> bug 1 was duplicate construction,destruction, which in most cases can be
> ignored in this case it did a double increment. Yikes! I made the copy
> constructor private and it went away.
>
> bug2 was in the PointerProxy forwarding of operator*() for doing **(op++),
> when I used the flag -03 the code works fine!.
>
> Anyway after more playing around with this, I think it works well enough
for
> those who don't care whether std::copy and their ilk work without a cast.
>
> your-real-not-a-proxy-template-happy-generic-programmer
> wishing-gcc3.0-would-come-out-soon
>
>   -gary-
>
> gary.powell_at_[hidden]
>
>
>
>
>