Re: [boost] [BGL] Maximum Weight Matching

Subject: Re: [boost] [BGL] Maximum Weight Matching
From: Jeremiah Willcock (jewillco_at_[hidden])
Date: 2013-10-10 13:16:26

On Wed, 9 Oct 2013, Piotr Wygocki wrote:

> Hi, I'd be very glad to see this algorithm in boost bgl. I don't want to
> interfere with the library maintainer, but there are some issues that
> bother me
> - The code is not header-only what is one of the main assumptions of BGL.
>
> - The big issue is to adjust the algorithm to standard bgl design pattern
> (for example algorithms use named_parameters). I think it is a good idea to
> read carefully implementation of some simple algorithm from bgl and copy
> the patterns. This includes also trivial things like missing licenses,
> namespaces etc.
>
> - Functions should be templated to achieve appropriate level of genericity.
>
> - There are many big macros, many of them looks just like a function for
> example GET_PENULTIMATE_BLOSSOM(j)
>
> - There are many not-necessary heap allocation. For example there is often
> a pattern
>        void f()  {
>              A * a = new A;
>              ...
>              delete a;
>        }
>        instead of just
>        void f()  {
>               A a;
>        }
>
> - There are c arrays instead of std::vector (generally the code looks like
> c-style c++)
>
>
> I'm hoping to see this algorithm in the next boost release!

I agree with your criticisms.  Please follow the guidelines at 
<URL:http://www.boost.org/development/requirements.html> and its linked 
pages in terms of naming conventions, directory structure, license, etc. 
The license, in particular, is a show-stopper for inclusion in Boost.

-- Jeremiah Willcock