Re: [boost] Re: no semaphores in boost::thread

From: Maciej Sobczak (maciej_at_[hidden])
Date: 2003-06-07 02:46:16

Hi,

Alexander Terekhov wrote:

> Show me some code. I mean something that shows why do you need counting
> semas.

This wording is too strong. Going this way, we can *always* say that 
feature X is not deadly needed and can be replaced by two or more 
(probably lower-level) features Y.
We could write assembly as well.

The problem with semas is that many textbooks use them as a mean to 
describe some ideas. A lot of algorithms are explained in terms of semas 
and mutexes, and not in terms of mutexes and condvars.
One could argue that every algorithm can be transformed to use only 
mutexes and condvars, but imposing this additional (and *this* is 
error-prone for sure) redesign work on programmers wanting to just get 
things done is *not* in the mission statement of any high level library 
and Boost in particular. The mission is to provide the programmers with 
the tools they want and IMHO semas fall into this category.
If we rewrite all textbooks (starting with all those where 'Dijkstra' or 
'Agrawal' appear), then maybe in the next century I will answer your 
question: "I don't see any".

If the library has no semas, then programmers will for sure roll their 
own, most likely flawed, implementations. It is much better to do it for 
them, as a joined effort of those who know how to do it well.

(No, "brain-dead pulsing events" are not really needed. ;) )

-- 
Maciej Sobczak
http://www.maciejsobczak.com/
Distributed programming lib for C, C++, Python & Tcl:
http://www.maciejsobczak.com/prog/yami/