[boost] Dynamic Bitset formal review

From: David Abrahams (david.abrahams_at_[hidden])
Date: 2002-06-14 14:08:32

While I think the world needs a component like this one, on the basis of
the current state of the documentation I would be forced to vote no.

The reference documentation is not sufficiently carefully constructed,
especially for a component of this simplicity. All of the functions which
operate on dyn_bitset are listed in a section headed "Members", though some
are not in fact members -- having a note in each non-member's documentation
doesn't make it OK. The documentation for many of the operators is pretty
useless. An assignment operator shouldn't be documented as "Assignment
operator". The &= operator docuementation just says "Bitwise and". What
kind of an ordering is imposed by operator<()? How will I predict whether
one bitset will be less than another?

Very important: what happens when bitsets of different sizes are combined?

The links to the LessThanComparable page at SGI STL are broken.

A minor problem is that the introduction says that it supports "all of the
bitwise operators that one can apply to builtin integers, such as
operator&&", but operator&& is neither supported nor a bitwise operator.

If these things were fixed before the review was over I could be persuaded
to vote yes.

I would prefer if the documentation took advantage of some of the fine work
that Bill Kempf has done in defining a documentation format for our
libraries, but I wouln't insist on it.

Provisional NO vote, subject to documentation updates.

-Dave

----- Original Message -----
From: "Mat Marcus" <mmarcus_at_[hidden]>

> Reminder - We have just past the half way point for the formal
> review of dynamic bitset. So far we have:
>
> 1 formal yes vote
> 1 formal no vote
>
> Discussion regarding iterators for dynamic bitset and
> comparison to vector<bool>.
> Some minor corrections and updates.
>
> I would like to see more formal reviews (including a yes or no
> vote) before the review period ends on the 17th.