Hi again Michal,
I just checked v500b3 code...

Well, I think there is some misunderstanding about winsock1 and winsock2 :

at least the purpose is to have consistant use of winsock1 H files AND lib, OR winsock2 H files AND lib.

Secondly, the purpose is to make use of wsock2 when and WHERE it REALLY exist (ie is available).

To make it more clear : winsock2 and getaddrinfo ARE AVAILABLE and working on WCE : the V434 code was assuming that getaddrinfo was NOT existing in WCE, which is just wrong.
And then, in V434, the code for common.h was containing this :

#ifdef _WIN32_WCE
#include <winsock.h>
#else
#include <winsock2.h>
#include <ws2tcpip.h>
#endif

meaning : use of wsock1 in WCE.
THAT was leading to compilation warnng about various #define that need -strangely- redef in winsock2, but are available in winsock1 (still following me ?...).

Example : this is required in common.h if using wsock2, but NOT if using wsock1. My 2010/11 code was fixing all this stuff.
#define ENOTSOCK WSAENOTSOCK


====

BUT today :

v500b3 common.h is JUST containing this :
#ifdef USE_WIN32
#include <winsock2.h>
#include <ws2tcpip.h>

well ok...nice...even for WCE...BUT inconsistent with EVC.MAK calling winsock1 LIB !!!!!!
and inconsistent with many #defines that are in fact REDEF of existing symbols that lead to compilation warnings.

So, I can help to make it more consistent :
use of winsock.h if using winsock.lib (ie winsock 1 !),
OR use of winsock2.h and ws2_lib (ie winsock2 !),

but not mixing winsock2.h and winsock1 lib !

===
Something else : in log.c, many, I should say "ALL", winsock X error codes are  HARDCODED in s_strerror() : not so good for maintainability...and stability of behavior (if codes are changing in the lib...).

I can help...but would you trust me ? :-)...

I will dive in this, this week-end,

See you soon,
Pierre


Le 06/12/2012 22:08, Michal Trojnara a écrit :