Hacker News Reader

Comment by FooBarWidget | Hacker News Reader

Sigh, another string library. And this is the reason why I prefer C++ over C. You don't end up writing another string library for the 300th time. There aren't that many differences between string libraries anyway. Most of them are just structs with a pointer to a memory block, plus a length field.

View on HN

> Sigh, another string library. And this is the reason why I prefer C++ over C. You don't end up writing another string library for the 300th time.

You say that, and yet every C++ project I've ever touched in my life has had its own string class with various levels of horror attached. My favorite was the one that stored everything internally as 32-bit characters to be Unicode safe, and was never used in a codebase that had to deal with Unicode.

Yep, but this one has not the usual layout of struct+pointer (that's the whole point), and I'm not sure it qualifies as "yet another" since it was written in 2006.

You're right, sorry. I only scanned the document, saw the struct, and closed the tab. The way you use a header before the data is similar to how Ruby represents its strings internally.

It appears to me to be "yet another" because the length before the string approach is usually referred to as a b-string.

windows (for example) has had a comprehensive b-string library (type is called BSTR) for about 20 years - due to it's age and provenance it has the downside of thinking a character is a 16-bit value...

Just to add a reference to the parent: http://www.johndcook.com/cplusplus_strings.html

(The article is mildly interesting by itself)

> And this is the reason why I prefer C++ over C. You don't end up writing another string library for the 300th time.

Ironically, in almost all of my uses of C++, I did end up writing or using a custom string library there too. (I was doing mostly console games or language interpreters.)

Yes, many people who have written much C have written some version of this library.

I'm not so happy with C++ string either. Main complaint is that I like the use of C strings as "semi-predicates"- you can test for NULL to indicate a failure. I wrote my own String class at one point to provide this feature:

  // These provide test if assigned/not assigned
  inline operator void*() const
    {
    return (void *)s;
    }

  inline bool operator!() const
    {
    if(s) return 0;
    else return 1;
    }

Better not to pun, and use an actual optional where you want it...

Eh, I like my C idioms. With C++ it's easy to make it safe so that accessing an unassigned string doesn't cause a crash (make NULL equivalent to empty strings except for tests).

I have the same gripe with the way the STL is designed. Too tedious to test for empty first before reading an item.

Not only that.

You also have x vector libraries for helping with bound checking in secure sensitive code and macro based generic data structures.

No thanks.

Reply on news.ycombinator.com