Subject RE: [IB-Architect] Event datasets RFD
Author David Schnepper
Jim --
Events now use out-of-band if the stack implementation on both ends supports
it. Otherwise it falls back to the separate socket approach.

Events were (are?) the most trouble prone area of any IB port I can recall.


-----Original Message-----
From: Jim Starkey [mailto:jas@...]
Sent: Tuesday, May 16, 2000 11:07 AM
Subject: Re: [IB-Architect] Event datasets RFD

At 06:58 PM 5/16/00 +0100, Paul Beach wrote:

>3. The initial Microsoft TCPIP stack didn't support out of band messaging -
>something if I remember rightly was pretty much a must for Events, and out
>of band messaging was pretty much a TCPIP standard, this particular lack of
>support from Microsoft was not exactly widely publicised, and caused many
>users to stop using events, mainly because of the problems this introduced.

Does it really try to use out of band messaging? I thought it used
a separate socket with asynchonous notification. When I did the
original implementation, I couldn't find any TCP implementations
that actually support out of band signalling reliably. The symptoms
ranged from error indicating not supported, just plain ignored,
delivered sometimes (phase of moon seem to have something to do
with it), and crash. Not warm and fuzzies.

On threaded systems, a delivery thread listening on a plain ordinary
socket is both the simplest and most reliable asynchronous mechanism.
This is an excellent use for threads.

Jim Starkey

Accurate impartial advice on everything from laptops to table saws.

To unsubscribe from this group, send an email to: