Re: [boost] the minimum degree ordering algorithm

From: David Abrahams (david.abrahams_at_[hidden])
Date: 2001-06-18 08:34:37

Rich,

Thanks for posting this. Overall, I am impressed with the readability of the
code. There are, however, a few blemishes on that impression. These are
really nitpicks as I haven't had a chance to look at the substance, so take
them with a grain of salt:

1. Use of TAB characters (these are banned in boost code). I took the
liberty of untabifying the file in CVS.

2. Ordering of class members: it really helps readability to have public
members declared first, whenever possible.  If you don't like starting your
class with an access specifier, you can always use "struct" ;-)

3. Mixing of interface and implementation: it can really help readability to
separate member function implementations from their declarations whenever
possible

4. long functions. Functions that don't fit on my 19" monitor with 10-point
type could benefit from being broken into several smaller ones.
mmd_impl::update() is a good candidate.

5. Vertical density. Some of the code blocks are all jammed together with no
"paragraph breaks". It can be useful to add a little vertical space between
things that group logically:

            vertex_t element = index_vertex_map[e_id];
e.g. here=> adj_iter i, i_end;
            for (tie(i, i_end) = adjacent_vertices(element, G); i != i_end;
++i){

I realize it is a matter of taste and a religious issue, but I think the use
of K&R bracing hurts you in this department. I notice that you frequently do
add a blank line after an open brace, however:

          list_min_degree = degreelists[min_degree];
        }

        // check if the whole eliminating process is done
        while (!numbering.all_done()) {

          size_type min_degree_limit = min_degree + delta; // WARNING

In cases like this one where there's space above and below, using aligned
braces helps to connect the control clause visually with the rest of the
loop.

Anyway, we're excited to have your code and are thinking of submitting a few
enhancements.

Best Regards,
Dave
----- Original Message -----
From: "Rich Lee" <llee1_at_[hidden]>
To: <boost_at_[hidden]>
Sent: Friday, June 15, 2001 7:35 PM
Subject: [boost] the minimum degree ordering algorithm

>
> For reference purpose, the email is to inform that we have added the
> minimum degree ordering algorithm into boost BGL (cvs).
>
> --
> Lie-Quan Lee
> University of Notre Dame
>
>
> To unsubscribe, send email to: <mailto:boost-unsubscribe_at_[hidden]>
>
>
> Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
>
>