Re: [boost] boost/filesystem/directory.hpp attached for comment

From: Beman Dawes (bdawes_at_[hidden])
Date: 2002-03-20 16:35:00

At 12:58 PM 3/20/2002, Peter Dimov wrote:

 >> int main( int argc, char * argv[]
 >> {
 >>    initial_directory();  // lock down before calling other functions
 >>    string dir_path;
 >>    if ( user provided a relative path )
 >>      dir_path = compose( initial_directory(), user provided path );
 >>    else
 >>      dir_path = user provided path;
 >
 >IOW a make_absolute() function, if provided, would treat relative paths 
as
 >relative to the initial_directory(). Yes, I see.
 >
 >How is this superior to simply providing a getcwd() function and letting
 >the user do
 >
 >int main()
 >{
 >  std::string initial = getcwd();
 >// rest unchanged
 >}

There would be no difference at all at that point in the code.  But if 
initial_directory() is called again later by some library routine, it still 
will give the correct answer.  Eventually initial_directory() might become 
part of the standard, and then would be set by the runtime library before 
main() runs.

If anyone comes up with a killer argument for get/set cwd, we can include 
them.  But that hasn't happened so far.

 >>  >basic_directory_iterator probably doesn't need a destructor.
 >>
 >> Depends on how the internals are implemented, doesn't it?  If the rep
 >type
 >> is opaque and non-trivial (as in the current imp) then a destructor is
 >> needed, for example.
 >
 >Well, it uses shared_ptr, and its destructor does the right thing for
 >incomplete types. :-)

Yes, but in general that is just an implementation detail AFAICS.

--Beman