Re: [boost] Dynamic Bitset Formal Review June 8-17

From: Ronald Garcia (garcia_at_[hidden])
Date: 2002-06-17 17:50:15

I vote "yes" to accept dynamic bitset into boost:

To me, it seems that dynamic bitset is meant more to represent sets than 
bits.  Based on this assumption, I would like to see in the interface a 
means for easily querying which elements are in the  set, possibly using 
indices.  Here's somewhat of a sketch of what I'm thinking of:

boost::array<std::string,3> domain = {{ "red","blue","yellow" }};
boost::dynamic_bitset<> bs(3);

// stuff happens here...

boost::dynamic_bitset<>::size_type i = bs.find_set_index(0);
while (i != boost::dynamic_bitset<>::npos ) {
  std::cout << domain[i] << ',';
  i = bs.find_set_index(i+1);
}

- I'm not convinced that dynamic bitset should have iterators.  I don't 
see it as a container but rather another means of representing sets. 
 Should one class (template) implement the roles of bit vectors and sets 
of bits?  IMHO, no.

- The review process appears to have revealed some confusion regarding 
the dynamic bitset's intent.  I think that a section discussing in 
greater detail its relation to vector<bool> and bitset might limit 
further confusion on the part of library users.  

- I would like to see the addition of an "all()" member function.  I am 
aware that this can already be semantically achieved using the current 
interface, but I like to spell all "a-l-l" and I believe the addition 
would not overgrow the interface.  

Previously discussed items I agree with:

- I prefer the name "dynamic_bitset" to "dyn_bitset".

- The logic tables in the documentation can go.

Little nit-pick-isms:

- The example files (example?.cpp) should include <cstdlib> in order to 
get EXIT_SUCCESS.

- In the examples, prefer '\n' to std::endl to prevent excessive buffer 
flushing.

- Is CHAR_BIT part of the dynamic bitset interface or need one include 
<climits>?  

- example1.cpp and example3.cpp have out-of-sync sample output comments. 
 In addition, the third example in the documentation seems to show 
output that is out of sync with the sample code.

- In the documentation of operator>>(basic_istream<...>&, 
dyn_bitset<...>&) (edits in sed notation):
s/extracted x.size() character/extracted x.size() characters/
s/other than 1/other than 0 or 1/

ron