Re: [boost] Re: Sockets - what's the latest?

From: Brian Gray (briangray_at_[hidden])
Date: 2003-02-14 12:36:02

On Friday, February 14, 2003, at 09:12 AM, Peter Dimov wrote:
> Brian Gray wrote:
>>
>> At the very end of it, network programmers should be using a
>> callback-driven interface and not have to worry about multiplexing at
>> all, but I agree that for now a third layer should be deferred until
>> the basic groundwork has been laid out.
>
> Sometimes it pays to design the highest level first, and then ask: what
> lower level components do we need to get here?

Normally I'd agree, but I guess I have this vision that the 2nd tier 
we're talking about would be a fully-usable, platform-independent API 
for those developers who wish to code at a lower level, and then the 
3rd tier would be a simpler -- but with a more advanced design -- 
model.  So they both need designing, with different purposes.  Tier 2 
needs to be powerful enough that anything may be coded using it, 
including new network protocols.  Tier 3 needs to be suitable for all 
but the most advanced purposes, gaining in design what it gives up in 
power.

I lean this way because I spend so much of my time doing very low-level 
things, and C/C++ is the only language that allows me the flexibility 
to do what I want.  The thought of designing a boost library that makes 
the same sacrifices that Java made is a real downer.  I see the need 
for a good OO design, but I'd also like raw access to low-level 
structures in a platform-independent manner.  Which is why I don't 
simply suggest that people like me use Sockets.  They don't exist on 
all platforms.

  -- Brian