D'OH!!!
Why doesn't this become an apr_sockattr element to the apr_socket_create
or apr_socket_bind call, so that APR determines the before/after thingy?
One side note, you setopt *after* the bind if you want to know you are
the absolute owner of the socket. you setopt *before* the bind on win32
if you know your app owns the socket and now you want to add additional
listeners (processes) to pay attention to it. Dunno how we could map this
all that cleanly.
Bill
William A. Rowe, Jr. wrote:
Jim Jagielski wrote:
On Feb 23, 2006, at 11:53 AM, Remy Maucherat wrote:
Jim Jagielski wrote:
I agree that the change is a big benefit, and for
most OSs we care about, SO_REUSEADDR is available.
The APR call should gracefully fail...
I'll plug this in later on today after some edge-case tests.
In AprEndpoint, there are a lot of Socket.optSet(serverSock,
Socket.APR_SO_REUSEADDR, 1);
Yes, the only rub is that for reasons I'm trying to remind myself
why, we do the setting before the bind under non-Windows, and
afterwards on Windows...
If you SO_REUSEADDR before on win32, you will share any existing port
that's previously opened by any other application. Not cool :)
Bill
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
