2009/12/28 Boris Dušek <boris.dusek@gmail.com>:
I haven't tried it, but what would prevent one from using std::istreambuf_iterator and std::ostreambuf_iterator on an gzip/bzip2 filtering streambuf created with boost.iostreams?
On Mon, Dec 28, 2009 at 5:05 PM, Jason Felice <jason.m.felice@gmail.com> wrote:
I noticed there's a gzip iostream, but no gzip iterator. This is a kind of precedent for the external dependency aspect.
First, I didn't see that there was a bzip2 filtering streambuf, though that might be because this code is currently stuck on boost 1.34.1 (for at least the next couple months). Second, I've not used the filtering streambufs, but I assumed they must decorate another streambuf. Seems a lot of extra machinery for what I was doing. I was very happy about my solution being very well abstracted such that the unit tests verified this stuff without hitting the disk. I used back_inserter<vector<char> > as the output type. I suppose you could do all that with stringstreams' buffer, but iterator-to-iterator seems like the simplest and most generic place to put it.