Re: [Boost-users] [Iostreams] Buffering nonblocking I/O

Subject: Re: [Boost-users] [Iostreams] Buffering nonblocking I/O
From: Nat Linden (nat_at_[hidden])
Date: 2012-01-04 19:37:48

On Wed, Jan 4, 2012 at 5:21 PM, Chris Cleeland <chris.cleeland_at_[hidden]> wrote:

> On Wed, Jan 4, 2012 at 1:44 PM, Nat Linden <nat_at_[hidden]> wrote:

>> Let's postulate a new Boost.Iostreams Concept to describe an object
>> capable of buffering data between producer and consumer.
>> Until we can settle on a better name, as a placeholder, let's just
>> call this a Queue.

> You could call this thing a "conduit".

Interesting! Although the term doesn't quite jump out at me, I can
understand the metaphor, and to my knowledge it has the virtue of not
already being hopelessly overloaded. If other people like it, I could
go with that.

>> ... I actually need an unbounded Queue for this.
>
> AND
>
>> It would also be possible to implement a Queue based on an actual OS pipe.
>
> Beware that OS pipes are bounded.

Yup, I know that. Sorry I wasn't clearer: I intended to suggest three
different plausible implementations of the Queue/Conduit concept, with
variations on the one I'm actually planning to build. For an unbounded
implementation I'm toying with a std::list< std::basic_string<Ch> >.

I consider it interesting that you could fairly simply build an
implementation around an OS pipe, but that doesn't address the use
case that interests me most at the moment. I tossed it out there for
purposes of illustration -- I guess a bit too tersely.