Re: [boost] filesystem library const usage

From: Beman Dawes (bdawes_at_[hidden])
Date: 2002-08-01 16:58:09

At 12:00 PM 8/1/2002, Mattias Flodin wrote:

 >On Thu, Aug 01, 2002 at 08:33:24AM -0400, Beman Dawes wrote:
 >> >What I've been meaning to ask, after this somewhat wordy post, are two
 >> >things. First, path::begin() returns a mutable iterator. Shouldn't it 
be
 >> >returning a const iterator, for the same reasons as above? Same thing
 >> >goes for path::end().
 >>
 >> I'm not quite sure what you mean here.
 >
 >I mean that code such as
 >
 >path p2("foo/bar");
 >p2.begin() = p2.end();
 >path::iterator it = ++p2.begin();
 >
 >is legal.

Ah, yes, I see what you mean.  But that is a general problem with the 
design of iterators, isn't it, rather than a problem with 
path?  path::begin() can't return a const iterator; iteration (operator++, 
operator--) wouldn't work!

 > In other words the return value from path::begin() is mutable
 >and the one returned from path::leaf() is not. This is inconsistent. Or
 >am I mistaken?

leaf() has the same semantics as the standard's optional sequence operation 
back(), although it returns by value rather than reference.  That design 
decision reflects my opinion that a path shouldn't be viewed as a 
full-blown container, but rather a specialized component which happens to 
have a few characteristics in common with containers.

 >Speaking of the docs, a description of what many of the functions
 >(e.g. path::leaf()) do seems to be missing. But perhaps that's on the
 >todo list.

Are you looking at the latest version?  Or perhaps it is confusing that 
path::leaf() and a number of other functions are fully specified simply by 
their "Returns" paragraph?  But that's all there is to it; you don't need 
an involved description.

Thanks for the comments!

I'd also like to hear usage reports, if anyone is trying the library out on 
real work.

--Beman