Re: WSARecv & 0 length buffer
- From: "Arkady Frenkel" <arkadyf@xxxxxxxxxxxxxxxx>
- Date: Fri, 15 Apr 2005 12:20:23 +0200
Why ? Each time the data come IOCP envoke you worker thread so you move
data to your working buffer
from IO buffer , so you don't need a lot of them
Arkady
"qfel" <[q_tmp]@[aster.pl]> wrote in message
news:425e807f$0$1421$f69f905@xxxxxxxxxxxxxxxxxx
> The problem about socket overlapped I/O is that you need to supply
> separate buffer for each request, which is nasty when having many pending
> receives.
> For now, I issue WSARecv with 0 - length buffer (so I just want completion
> routine to be queued when any data is avalible to read). This should be
> better than event-drived model, and seems to work, although I cannot se it
> docummented anywhere.
> Does anybody know if such receives are legal?
>
.
- Follow-Ups:
- Re: WSARecv & 0 length buffer
- From: Arkady Frenkel
- Re: WSARecv & 0 length buffer
- References:
- WSARecv & 0 length buffer
- From: qfel
- WSARecv & 0 length buffer
- Prev by Date: Re: How to stop WZC Service in XP-SP2
- Next by Date: Re: How to stop WZC Service in XP-SP2
- Previous by thread: WSARecv & 0 length buffer
- Next by thread: Re: WSARecv & 0 length buffer
- Index(es):
Relevant Pages
|