I recently discovered that
std::strstream has been deprecated in favor of
std::stringstream. It’s been a while since I’ve used it, but it did what I needed to do at the time, so was surprised to hear of its deprecation.
My question is why was this decision made, and what benefits does
std::stringstream provide that are absent from
strstream returned a
char * that was very difficult to manage, as nowhere was it stated how it had been allocated. It was thus impossible to know if you should delete it or call free() on it or do something else entirely. About the only really satisfactory way to deallocate it was to hand it back to the
strstream via the
freeze() function. This was sufficiently non-obvious, that lots of people got it wrong. The
stringstream returns a string object which manages itself, which is far less error prone.
There was also the issue of having to use
ends to terminate the string, but I believe the deallocation problem was the main reason for deprecation.
Easier to understand memory management. (Can someone remember who is responsible for freeing the allocated memory and in which conditions?)
(Note that as strstream still provide something which is not available elsewhere, it will continue to be present in C++0X — at least last time I checked the draft it was).
strstream builds a
char *. A
std::stringstream builds a
std::string. I suppose
strstreams are deprecated becuase of the potential for a buffer overflow, something that
std::string automatically prevents.
From a personal perspective on more than one occasion I’ve seen obscure memory corruptions that took days or weeks to track down and eventually came down to use of
strstream. As soon as it was replaced with
stringstream the corruptions vanished and I didn’t ask any more questions! That was enough for me.