Friday 22 April 2005 Talk: Paul and Jez's Stream-a-poloza

A fun filled journey into the Standard Library's dark continent, illuminating the mysteries of IOStreams and revealing the streambufs beneath.

C++ IOStreams - what they do, how they work, why you might want to write your own, and how to do it without pain or fuss.

Slides and speaker notes from a 90 minute session presented with Paul Grenyer at ACCU Conference 2005

---

IOStreams is one of the lesser known areas of the standard library. While most people are now entirely comfortable slinging around std::vectors and std::lists, and know the difference between a std::map and a std::set, IOStreams is still rather opaque.

Sure, writing an operator<< isn't too big a deal, and maybe the bold have even come up with a custom manipulator. But streams? Where do they come from? Where do they go?

In a fun filled journey into the Standard Library's dark continent, we will attempt to illuminate the mysteries of IOStreams and reveal the streambufs beneath.

Starting with a quick tour through IOStreams excitingly diamond shaped class hierarchy, we'll introduce the concepts of sources and sinks (i.e. Streams which read from or write to some external destination) and filters/adaptors (streams that modify data in flight). We'll also talk about some of IOStreams little idiosyncrasies – why some buffers might not actually buffer, and why read-only buffers sometimes need to be writable.

By way of illustration we'll be walking through a logging stream. It writes log messages to an arbitrary number of other streams. The destination streams will be a combination of fstreams, stacks of filters, logging daemons on other machines, and whatever else we need.

---

Tagged code, c++, talk, and accu-conference

---