«

Hi all,

I have 2 char* iterators str and end and I'm doing as follows:
string id_str(str, end);
const char * id = id_str.c_str();

but these has 2 problems afaik. One, I'm generating a string as an
intermediate step to get a char*, which seems useless. Two, I don't
know when id gets destroyed or when the chars to where id points to are
cleaned. I could now use strcopy to copy id to a freshly allocated
string but... is there any more direct way? (more efficient perhaps???)

Cheers,

Paulo Matos

"pmatos" <> wrote in message
news: oups.com...

> I have 2 char* iterators str and end and I'm doing as follows:
> string id_str(str, end);
> const char * id = id_str.c_str();
>
> but these has 2 problems afaik. One, I'm generating a string as an
> intermediate step to get a char*, which seems useless. Two, I don't
> know when id gets destroyed or when the chars to where id points to are
> cleaned. I could now use strcopy to copy id to a freshly allocated
> string but... is there any more direct way? (more efficient perhaps???)

Well, you can always do this:

size_t len = end - str;
char* id = new char[len+1];
std::copy(str, end, id);
id[len] = '\0';

However, before following this path, I suggest you think about why you want
a char* in the first place. It just adds to your bookkeeping hassles. Why
not just use the string itself?

Andrew Koenig wrote:
> "pmatos" <> wrote in message
> news: oups.com...
>
> > I have 2 char* iterators str and end and I'm doing as follows:
> > string id_str(str, end);
> > const char * id = id_str.c_str();
> >
> > but these has 2 problems afaik. One, I'm generating a string as an
> > intermediate step to get a char*, which seems useless. Two, I don't
> > know when id gets destroyed or when the chars to where id points to
are
> > cleaned. I could now use strcopy to copy id to a freshly allocated
> > string but... is there any more direct way? (more efficient
perhaps???)
>
> Well, you can always do this:
>
> size_t len = end - str;
> char* id = new char[len+1];
> std::copy(str, end, id);
> id[len] = '\0';
>
> However, before following this path, I suggest you think about why
you want
> a char* in the first place. It just adds to your bookkeeping
hassles. Why
> not just use the string itself?

Oh well, I might well start a flame war... hope not. However, I'm
programming for efficiency and it seems to me handling char * to be
faster than handling strings. Would I be incorrect for any reason?
Anyway, even if using strings, I'd need to use string * a lot since I'd
be passing them around and I don't wish to be passing them by value.

Cheers,

Paulo Matos

"pmatos" <> wrote in message
news: oups.com...

> Oh well, I might well start a flame war... hope not. However, I'm
> programming for efficiency and it seems to me handling char * to be
> faster than handling strings. Would I be incorrect for any reason?

Have you measured it? The differences, if any, would depend on the
particular implementations of strings and memory allocation that you happen
to be using.

> Anyway, even if using strings, I'd need to use string * a lot since I'd
> be passing them around and I don't wish to be passing them by value.

Or perhaps define a class that lets you get the performance characteristics
you want in a more disciplined way.