Re: About dlw_int.zip 7 and dlw_gcd.zip 7 (was: Re: About iostate.zip 1)

From: Daryle Walker (darylew_at_[hidden])
Date: 2001-10-21 22:00:15

on 10/21/01 1:53 PM, Helmut at helmut.zeisel_at_[hidden] wrote:

> --- In boost_at_y..., Daryle Walker <darylew_at_m...> wrote:
>> on 10/20/01 5:59 AM, David Abrahams at david.abrahams_at_r... wrote:
>> 
>>> I think it makes sense that you resolve the outstanding issues with your
>>> currently-accepted submissions (int/gcd) first, so that they can be checked
>>> in to the boost distribution. Does that make sense to you?
>> 
>> I uploaded updated versions (both 7) of dlw_int.zip and dlw_gcd.zip a little
>> over a week ago, but no one responded.
>> 
> 
> I did:
> 
> http://groups.yahoo.com/group/boost/message/18502

OK.  But I was looking more for the opinions that thought that the previous
version was too complex, instead of others (like you) that thought it didn't
go far enough.

Actually, the design isn't that complex.  There's a little bit of complexity
from me using the "function template that works by calling a class template"
idiom.  Besides that, the header looks big because:

1.  I use a verbose coding style
2.  One header holds the compile-time and run-time GCD & LCM code.
3.  I do a lot of _inline_ function forwarding to get to the
    actual routines.

> Anyway, AFAIS we now have the following situation:
> A major design goal of dlw_gcd was that it should unify
> the GCD code existing in several boost libraries.

Yes, I want to unify all the GCD code that is currently spread around the
Boost libraries.  That way, people can use the GCD code without having to
include rational stuff or dig deep into pool details.

Has anyone tried out the newest versions for MSVC compatibility?

> From the reactions of the owner of these libraries, however, I get the
> impression that they are not willing to use your code instead of their
> original one.
> 
> So a major effort should be to convince the owners of these libraries that
> they indeed include your code. AFAIK, there are still open questions:
> 
> http://groups.yahoo.com/group/boost/message/18367
> 
> http://groups.yahoo.com/group/boost/message/18374

I haven't seen a response from these posters on the latest code.

> An alternative solution could be that you convince the boost community that
> there is need to have now three GCD instead of two. You could justify your
> additional GCD by additional features such as that your GCD now works for all
> Euclidean rings. This is indeed convincing for me, but I fear that it is
> convincing for me only.

I wouldn't like that alternative solution.

-- 
Daryle Walker
Mac, Internet, and Video Game Junkie
darylew AT mac DOT com