Re: [boost] [xint] Third release is ready, requesting preliminary review

Subject: Re: [boost] [xint] Third release is ready, requesting preliminary review
From: Chad Nelson (chad.thecomfychair_at_[hidden])
Date: 2010-05-03 21:43:02

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 05/03/2010 09:22 PM, Scott McMurray wrote:

>> Would your average developer know about that sort of thing? I've
>> been using C++ since 1987, and I'd never seen it until a couple
>> months ago. And then only by digging into the Boost.TypeTraits
>> library's source code to figure out how it worked.
> 
> Does "your average developer" even write templates that use the extra
> power than templates offer over generics?

Not sure what you mean, since in C++ "generics" means templates to me.
And I don't know how I stack up to the average developer, but I've been
writing template code for years and never knew about that.

> I'm not convinced that Boost should be all that concerned about
> developers attempting to write generic code while ignorant of
> fundamental template building blocks.  The trick behind enable_if has
> been known by the acronym SFINAE since at least 2002, thanks to
> Vandevoorde and Josuttis's book [1].  Alexan­dres­cu's _Modern C++
> Design_ [2] -- containing templates far more complicated than a simple
> specialization -- was published more than 20 months earlier than that.
> These are not bleeding-edge or secret techniques.

Ah, that explains it -- most programmers don't buy or read books since
the Internet became widely available. I do, but I only go looking for
one when I run into a problem, which is likely why I missed those two...
I've never run into a problem that only they would have a solution for.

> That said, I'm not sure whether it's better to not provide min and 
> max, and thus catch code that uses them without checking is_bounded, 
> or to possibly make it easier to correctly write code that does
> check is_bounded.

I figured that compiler writers probably knew more about it than I did,
and followed their example.

> And who knows -- maybe making it easy with SFINAE would mean that 
> those developers that don't know will go looking for the clean way
> and learn a new tool.

Stranger things have happened.
- -- 
Chad Nelson
Oak Circle Software, Inc.
*
*
*
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAkvfe6QACgkQp9x9jeZ9/wRCCQCeLM0JPubKh9DMWPAutCLiHIZe
toMAoOxS4jIcIOQeyINhGc6knKEx8aNx
=6CL8
-----END PGP SIGNATURE-----