Keith Marshall
keith****@users*****
Sun Jul 29 03:40:45 JST 2018
On 28/07/18 14:12, Eli Zaretskii wrote: > I've upgraded to MinGW runtime 5.0.2, and several programs now fail to > compile because they include winsock2.h and then expect error codes > from winerror.h to be defined. Here's a typical example, from Gawk: > > gcc -c -D__USE_MINGW_ANSI_STDIO -DHAVE_LIBREADLINE -DHAVE_MPFR -O2 -gdwarf-2 -g3 -Wno-deprecated-declarations -DGAWK -I. -I./support -DHAVE_CONFIG_H -DDEFLIBPATH="\"d:/usr/lib/gawk\"" -DSHLIBEXT="\"dll\"" -DLOCALEDIR="\"d:/usr/share/locale\"" gawkmisc.c > In file included from gawkmisc.c:36:0: > pc/gawkmisc.pc: In function 'dlopen': > pc/gawkmisc.pc:756:18: error: 'ERROR_INVALID_PARAMETER' undeclared (first use in this function); did you mean 'WSA_INVALID_PARAMETER'? > last_err = ERROR_INVALID_PARAMETER; > ^~~~~~~~~~~~~~~~~~~~~~~ > WSA_INVALID_PARAMETER > pc/gawkmisc.pc:756:18: note: each undeclared identifier is reported only once for each function it appears in > pc/gawkmisc.pc: In function 'dlsym': > pc/gawkmisc.pc:814:18: error: 'ERROR_INVALID_PARAMETER' undeclared (first use in this function); did you mean 'WSA_INVALID_PARAMETER'? > last_err = ERROR_INVALID_PARAMETER; > ^~~~~~~~~~~~~~~~~~~~~~~ > WSA_INVALID_PARAMETER > pc/gawkmisc.pc: In function 'w32_maybe_set_errno': > pc/gawkmisc.pc:1013:12: error: 'ERROR_BAD_PIPE' undeclared (first use in this function); did you mean 'LZERROR_BADVALUE'? > case ERROR_BAD_PIPE: > ^~~~~~~~~~~~~~ > LZERROR_BADVALUE > pc/gawkmisc.pc:1014:12: error: 'ERROR_PIPE_BUSY' undeclared (first use in this function) > case ERROR_PIPE_BUSY: > ^~~~~~~~~~~~~~~ > pc/gawkmisc.pc:1015:12: error: 'ERROR_NO_DATA' undeclared (first use in this function); did you mean 'IMM_ERROR_NODATA'? > case ERROR_NO_DATA: > ^~~~~~~~~~~~~ > IMM_ERROR_NODATA > pc/gawkmisc.pc:1016:12: error: 'ERROR_PIPE_NOT_CONNECTED' undeclared (first use in this function); did you mean 'WN_NOT_CONNECTED'? > case ERROR_PIPE_NOT_CONNECTED: > ^~~~~~~~~~~~~~~~~~~~~~~~ > WN_NOT_CONNECTED > > This happens because some header included by the source file does > this: > > #if !defined _WIN32_WINNT || _WIN32_WINNT < 0x501 > # undef _WIN32_WINNT > # define _WIN32_WINNT 0x501 > #endif > #include <winsock2.h> > #include <ws2tcpip.h> > > This used to work with MinGW runtime 3.x, but now fails, evidently > because winsock2.h no longer includes windows.h, Actually, it does, but indirectly via <winsock.h>, with which it now shares code which is common to WinSock v1.1 and v2 protocols. > and this doesn't define the ERROR_* macros. Instead, it does this: > > #define WSA_IO_PENDING (ERROR_IO_PENDING) > #define WSA_IO_INCOMPLETE (ERROR_IO_INCOMPLETE) > #define WSA_INVALID_HANDLE (ERROR_INVALID_HANDLE) > > etc, i.e. it redirects winsock errors to "common" errors. It has always done this; I guess it's now more selective about what it actually includes from <winerror.h>. > But it doesn't seem to take care of defining those common errors. That it gratuitously depends on the above three (and three more), which it filters out, is a bug. > So I need to include winerror.h before winsock2.h, in order to have > those error codes defined. It shouldn't matter which order you include them; <winsock2.h> first, followed by <winerror.h> should work just as well. > Do these programs do something that isn't "kosher" nowadays? What > should a program do if it wants to include winsock2.h, but doesn't > want to include windows.h before that? Is that allowed? Yes. In fact, IIRC, it is required for strict MSVC compatibility. What is less clear cut is: should <winsock2.h>, (and indeed <winsock.h>), unconditionally expose all of <winerror.h>, or just those definitions which are relevant for the WinSock protocols? Certainly, the six error codes, on which <winsock2.h> gratuitously depends, must be exposed, but that won't be sufficient to define other error codes, which appear to be undefined in your compilation, unless you also explicitly include <winerror.h>, or its WinSock selectivity filters are removed. On balance, it may be best to just expose all of <winerror.h>, whenever it is included; I think that's what happened, (via <windows.h>), in the earlier version, where the selectivity filters were different: they were in place to filter out blocks of conflicting WinSock specific error codes, which were "duplicated" (inconsistently) in <winsock.h>, <winsock2.h>, and <winerror.h> -- Regards, Keith. Public key available from keys.gnupg.net Key fingerprint: C19E C018 1547 DE50 E1D4 8F53 C0AD 36C6 347E 5A3F -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 836 bytes Desc: OpenPGP digital signature Url : https://lists.osdn.me/mailman/archives/mingw-users/attachments/20180728/46f6a366/attachment-0001.pgp
Fork