[boost] [intrusive] Review

From: Paul Rose (prose_at_[hidden])
Date: 2007-03-16 12:53:28

This is my first time submitting a review, but I find this library
exiting enough that I didn't want to risk it not getting accepted due to
lack of reviews.
 
- What is your evaluation of the design? 
    From my reading of the documents I like the design very much.  
    I'm not very fond of the i prefix given that namespaces are already
used, but to me that is a minor issue.
    I like the fact that everything (including size type and pointer
type) is customizable.
    I like that you can choose between member hooks and base hooks.
    I and that you can chose safe mode and constant time size.  
    The auto-unlink hooks are also very nice.
    I like the exposed red-black tree node functions, these will be
useful in building other advanced containers.

- What is your evaluation of the implementation? 
    I haven't dug into the implementation code very much.  
    The public interface implies a certain implementation. 
    If the implementation indeed matches what what is implied by the
public interface, then I would be happy.

- What is your evaluation of the documentation? 
    Very good.  I didn't need much motiviation, because I understood at
a glace exactly when I would want to 
        use this instead of std conatiners or Boost.Mulltindex.  But
perhaps more motivating examples would 
        be good for other potential users.
    
- What is your evaluation of the potential usefulness of the library? 
    Very useful.  I will use this library regardless of whether or not
it is accepted, even if I have to maintain my own internal copy.
    I am a very big user of (and big fan of) Boost.Multiindex.  One
feature I have wished for there were "sparse" indexes, where 
        not every node need be referenced be every index.  I understand
this would be a mess to add to the B.MI interface, but now 
        I can do it myself with Boost.Intrusive.

- Did you try to use the library? With what compiler? Did you have any
problems? 
    Unfortunately no.  I would prefer to use it before reviewing, but I
didn't want the review window to close without having submitted a
review.
    I will soon try it out with g++ 4.1.0 on linux and MSVC 7.1 on
Windows.

- How much effort did you put into your evaluation? A glance?   A quick
reading? In-depth study? 
    In depth study of the documentation.

- Are you knowledgeable about the problem domain? 
    I am a heavy and long-time user of Boost.Multiindex (which I love,
BTW), but I am no true expert in container implementation.

- Do you think the library should be accepted as a Boost library? 
    Yes!
 
Paul Rose