Re: [Boost-users] Getting started with Boost: what did you need to know?

From: David Abrahams (dave_at_[hidden])
Date: 2006-02-03 11:27:29

Jim Douglas <jim_at_[hidden]> writes:

> David Abrahams wrote:
>> What do you wish someone had
>> explained to you?  With what did you need hand-holding?  Based on this
>> feedback I expect to make some improvements to the Boost website,
>> also.  Any information you can give me will be much appreciated.
>
>  From my POV as a user and recent contributor my comments are:
>
> 1. It took me a long time for me to realise that a large proportion of 
> Boost is header-only and can be used without pre-building any libraries 
> (archives & shared objects).
>
> 2. It is also not immediately clear that there is a large amount of 
> interdependency between Boost libraries. This information would be 
> particularly useful when trying to resolve compilation problems e.g if 
> Boost.b & Boost.c both use Boost.a, if you can't get 'a' to compile then 
> you don't stand a cat in hell's chance with b and/or c.

We have to be careful about qualitative statements like that one.
Compared to many large-scale libraries, Boost is highly decoupled.

> IMHO it is high time that someone produced a dependency graph, or each 
> library document had a "uses" section.

Do you know about http://www.boost.org/tools/bcp/bcp.html and its
--report option?

That's something I can cover in my talk.

> 3. Boost's sheer size makes it difficult to document. From the responses 
> so far it's obvious that different users have their own personal 
> favourite sections. I find that I discover something new every day, 
> usually thinking "If only I had discovered that six months ago I could 
> have saved so much time!"
>
> The ultimate solution would be to have a problem oriented section in the 
> docs, i.e. if you are trying to do this, then we suggest you try these 
> libraries. If we assume that each library was constructed in order to 
> fill a need and/or solve a problem (and not just for intellectual 
> amusement) then these needs/problems could be collated into a single 
> section.

Yeah... some libraries are so general-purpose (e.g. lambda, mpl,
preprocessor) that it's hard for me to say anything that most people
will identify as their use case.

But if you could be more specific -- i.e. write some of it -- then
maybe the rest of us could pick up on what you're doing and see how it
would work.

> 4. The single word naming of the libraries can lead to ambiguities and 
> misunderstanding e.g. "serialization" means different things to 
> different people and requires a full paragraph to explain, and IMHO 
> "thread" is somewhat misleading. Other names mean nothing to me so I 
> have to go and look them up.

What would be better, "the boost serialization of a style espoused by
Robert Ramey library?" ;-)

-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com