gonzalo diethelm wrote:
> Ok, that clears up what error we are talking about. But what would cause
> an inproc socket connection to be refused? If the other end has not come
> up yet, the connection will be retried, right?
The auto-retry for inproc isn't implemented yet :( See issue 6.
> BTW, my copy of WinError.h has
>
> #define WSAECONNREFUSED 10061L
>
> and WinSock2.h has the following (consistent with WinError.h):
>
> #define WSABASEERR 10000
> ...
> #define WSAECONNREFUSED (WSABASEERR+61)
> #define ECONNREFUSED WSAECONNREFUSED
That may differ between the windows versions... In any case zmq.h
belives that ECONREFUSED isn't defined - otherwise it wouldn't define it
itself.
> Regarding zmq_strerror, would you consider a patch that prepends the
> numeric error value to the string that describes it? That way, for
> EADDRINUSE (under Windows) you would get the string "10048 - Address in
> use". Perhaps "EADDRINUSE - Address in use" or "10048 - EADDRINUSE -
> Address in use" would be even more informative. This might help find out
> the cause for errors.
I am trying to be consistent with POSIX strerror function. Thus you have
the numeric value (10048) that can be translated into textual form using
strerror ("Address in use").
Martin
_______________________________________________
zeromq-dev mailing list
[email protected]
http://lists.zeromq.org/mailman/listinfo/zeromq-dev
