Re: exported.writes inside a hook threadlock?

On Mon, 12 Apr 2004 22:19:33 +0400
  Eugene <[email protected]> wrote:
>On Mon, 12 Apr 2004 13:04:35 -0500 (CDT)
>  will guaraldi <[email protected]> wrote:
>
>>OutputEvent.  How much bang for our buck would we get 
>>from that?  Do you
>>think it'd simplify anything?
>>
>>Would it affect startup adversely?
>
>I've made the change in my working copy, and it seems all 
>right so far. I should do mudding a bit more with it, to 
>tell is it really safe. Right now I can think of no 
>pitfalls. About positive effect - the write_ui processing 
>is getting shorter (moving the real output to another 
>thread).

It is not so good really as it appears at first glance. 
The problem is that we don't move the job into another 
thread - it is the same thread puts into the queue as well 
as gets from it. So, in fact, using the queue means just 
scheduling (with several delay).

The negative effect is a big loss in speed when data are 
flooding from the network. Not very common case, but still 
possible.

So, current thread structure does not fit well for this 
solution, and I'd rather let things stay as they are now. 
If only we have one more thread to dispatch this UI queue, 
it would be some better (I tried it, and it works, but I'm 
not proposing it right now). May be in some future 
version.

--
Eugene
---
Professional hosting for everyone - http://www.host.ru


-------------------------------------------------------
This SF.Net email is sponsored by: IBM Linux Tutorials
Free Linux tutorial presented by Daniel Robbins, President and CEO of
GenToo technologies. Learn everything from fundamentals to system
administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click