Re: strings

[Andrew Johnson <anj@aps.anl.gov>]

Kay-Uwe Kasemir wrote:
> On Sep 27, 2005, at 13:04 , Andrew Johnson wrote:
>
> I agree that it's initially more convenient to be able to
> use the operators.
> But in the long run, it causes the C++ and e.g. Java sources
> differ, and in C++ I also find it difficult to understand what's
> called when everything has cast and assignment operators.

I am not providing cast operators, and to some extent I agree about the 
difficulty reading code that uses them.  However the assignment 
operator= is an integral part of C++ that the compiler will create for 
you (usually wrongly) if you don't explicitly tell it not to.  I believe 
it makes more sense to make use of the facilities provided by the 
language than to try and suppress them, so I'm providing the convenience 
operators.  If you want your code to be easily translated into Java 
you're not forced to use them to manipulate strings.

> > operator[] is not currently implemented in that fashion, but I am  
> > willing to do that too (although I don't like your name so I will  
> > probably change it).

class StringReader {
public:
...
    // Individual byte access
    virtual char at(size_t pos) const = 0;
...
    // Overloaded operator, not virtual
    char operator[] (size_t pos) const;
};

I'm now using the std::string name at() which will throw some kind of 
out_of_range exception if pos points beyond the end of the string.  The 
inline overloaded operator[] is now just an inline synonym for at(), 
which is different to std::string but safer - std::string::operator[] 
does no checking of pos.

- Andrew
--
English probably arose from Normans trying to pick up Saxon girls.