Hi, www/sope now uses MSG_NOSIGNAL as flag for send(2), see the Linux manual page: http://www.kernel.org/doc/man-pages/online/pages/man2/send.2.html
After some googling, I found that I can define it to 0, and just go ahead. This way its right now in SOPE. I opened a bug report to get the patch upstream:: http://www.sogo.nu/bugs/view.php?id=1460 The answer I got tells, me, that its needed to fix a crash in openchange/samba4 compatibility. This is for sope/sogo 2.0, we don't have openchange/samba4 for now, so ignoring it, should be no problem for now. Anyone can give me a hint on how to better handle this MSG_NOSIGNAL instead of just defining it to 0? checking the ports tree for examples, I see it differently handled: most of them, just define it to 0, others i.e. dctc/patches/patch-src_dc_xfer_common_c does: +/* compability macros for reading/writing to sockets */ +#ifdef __OpenBSD__ +# define sock_read(sck, buf, size, flags) read(sck, buf, size) +# define sock_write(sck, buf, size, flags) write(sck, buf, size) +#else +# define sock_read(sck, buf, size, flags) recv(sck, buf, size, flags) +# define sock_write(sck, buf, size, flags) send(sck, buf, size, flags) +#endif + and then later: - res=send(sck,buf,sizeof(buf),MSG_NOSIGNAL /* |MSG_WAITALL */ ); + res=sock_write(sck,buf,sizeof(buf),MSG_NOSIGNAL /* |MSG_WAITALL */ ); I'm unsure whether the dctc approach is good? Mac OSX seems to have SO_NOSIGPIPE as setsockopt() option, but we also don't seem to have this. any insights or pointers to further research on my side is apprechiated. cheers, Sebastian
