Re: [Boost-users] [review] Review of Unordered library begins today December 7

From: Cromwell Enage (sponage_at_[hidden])
Date: 2007-12-14 23:15:19

--- Ion Gaztañaga wrote:
> Hi all,

Hello!

> What to include in Review Comments
> ==================================
> 
> Your comments may be brief or lengthy, but
> basicallythe Review Manager needs your evaluation
> of the library.  If you identify problems along
> the way, please note if they are minor, serious,
> or showstoppers.
> 
> Here are some questions you might want to answer
> in your review:
> 
> * What is your evaluation of the design?

The design is excellent, as it should be.  I can see
how everything works at a glance.

> * What is your evaluation of the implementation?

The implementation rationale is detailed enough;
however, I am curious as to the purpose of
BOOST_UNORDERED_LOCAL_CBEGIN.

> * What is your evaluation of the documentation?

It's pretty thorough.  A few nits and issues:

* Since the documentation assumes prior knowledge of
STL containers, it would be nice for the Introduction
to provide a link to SGI's website (or whichever STL
reference documentation you guys prefer).
* It should be stressed that either writing your own
hash function or extending/specializing boost::hash is
*essential* to get Boost.Unordered containers to work
with UDTs (not just *pointers* to UDTs).
* Sometimes I see "T" where "Value" was meant to be. 
This appears in the reference documentation for all
containers.
* I read the return types of equal_range() and
equal_range() const as the same, when the return type
of the const variant should be
std::pair<const_iterator,const_iterator>.  This
applies to the reference documentation for all
containers.
* s/elment/element

> * What is your evaluation of the potential
> usefulness of the library?

At work, we use std::map to hold nodes in our tile
search algorithms simply because std::hash_map is not
a standard container.  As soon as compiler and library
vendors start incorporating Boost.Unordered into their
TR1 implementations (or even sooner, if I can convince
my boss that Boost is worth the download), we will
definitely switch over.

> * Did you try to use the library?

I haven't had the time to run the test programs, but
I've replaced std::map with boost::unordered_map in a
few of my own examples.

> * With what compiler?

MinGW (GCC 3.4.5)

> * Did you have any problems?

None so far.

> * How much effort did you put into your evaluation?

Four hours of in-depth reading, one minute of coding
(search-and-replace works because I try to keep things
generic).

> * Are you knowledgeable about the problem domain?

I'm proficient with data structures in general, not so
much with particular hash functions.

> And finally, every review should answer this
> question:
> 
> * Do you think the library should be accepted as a
> Boost library?

Oh, yeah.

                              Cromwell D. Enage

      ____________________________________________________________________________________
Be a better friend, newshound, and 
know-it-all with Yahoo! Mobile.  Try it now.  http://mobile.yahoo.com/;_ylt=Ahu06i62sR8HDtDypao8Wcj9tAcJ