Re: [boost] SUBMISSION: Signal/slot library

From: sipan_at_[hidden]
Date: 2000-04-09 20:35:40

--- In boost_at_[hidden], Joseph Davis <joseph.davis_at_i...>  wrote:
>  
> 
> Tomasz Kowalczyk wrote:
> 
> >
> > What needs to be made better ?
> > What part do you think is not portable ?
> > What aspects should be more "open" ?
> >
> 
> I'm sorry, but I don't feel qualified to answer these questions.
> I really don't have any first hand knowledge of libsigc++. The Qt
> implementation seems fine for us.  The advantages touted on the
> libsigc++ web page don't really have any practical impact on
> what we are doing..

libsigc++ is faster, not associated with any GUI toolkit, and
does not require special preprocessor.

> I suppose the better questions is:
> 
> Does signal/slot play a useful role in areas other than GUI widget
> design?

Yes. E.g. when you have multiple modules which have to communicate,
you have to implement your own message passing protocol or use
libsigc++ (or its analogs). The project I am working on is using
its own message scheduling and I can bet that result will be much
slower and more complex application then it could have been if we
were using libsigc++.

The problem is that libsigc++ is complex library. E.g it has its
own complete(almost?) collection of wrapper classes around POSIX
threads.

The question is, does high(or low, if you want) level stuff like
threads, sockets, serialization, signal/slot implementations should
be in "boost", which, as I understand, deals with more generic
things.

  Sergey