RE: [misterhouse-users] Perl question

[Varnavas, F. <fra...@gs...>]

Hi Bill,

I like both of the ideas you mention but my original question:

" ...Is there an architectural reason why mh polls rather then using an
asynchronous event model?"

really refers to your last sentence.  Whenever I've written asynchronous
communication software barring an architectural limitation in the hardware
or underlying OS I've gone with a non-polling model. Polling eats a lot of
cpu and can be a poorer performer then an eventing model. But it may not be
possible to implement a non-polling misterhouse on all platforms where
misterhouse is run. 

It's more of a curiosity about the perl support for serial IO then any
practical concern -- misterhouse seems to perform well enough as currently
implemented.

Thanks
Frank V

> -----Original Message-----
> From: Bill Sobel [mailto:bs...@vi...]
> Sent: Friday, June 02, 2000 4:39 PM
> To: mis...@li...
> Subject: RE: [misterhouse-users] Perl question

 
> what it is today.  I've certainly wanted to somehow add an 
> event mechanism,
> two ideas I've been throwing around are:
> 
> 1) The ability to assign 'code' to an event, something like:
> 
> $x10_item->setevent('on',&::frontlineonsubroutine);
> 
> 2) $voicecommad->addobject($x10_item);
> 
[SNIP]

> Of course, at the core, the system would still poll, but this would be
> hidden from the objects using this system.
> 
> Any thoughts on either of these?
> Bill
> 
> 
> ________________________________________________________
> To unsubscribe from this list, go to: 
> http://sourceforge.net/mail/?group_id=1365
>