Re: SIGPIPE (again)

[sjh@zorak.net]

On Sunday August 10, "Kurt D. Zeilenga" wrote:
> I really don't see this approach fixing much...  but it may

How would it not fix the SIGPIPEs going to apps?

> break a few things.  Note that send(2) only works for sockets
> while write(2) works on most any kind of descriptor.  While
> normally the library is used with sockets, it can be used
> with pipes, files, and such.

Hmm, so liblber sometimes talks to files/pipes in sockbuf.c?

This is the patch:

*** sockbuf.c   2003-08-08 10:55:36.000000000 -0700
--- sockbuf.c.orig      2003-08-08 10:56:33.000000000 -0700
***************
*** 487,493 ****
        return nread( sbiod->sbiod_sb->sb_fd, buf, len );
  
  #else
!       return tcpread( sbiod->sbiod_sb->sb_fd, buf, len );
  #endif
  }
  
--- 487,493 ----
        return nread( sbiod->sbiod_sb->sb_fd, buf, len );
  
  #else
!       return read( sbiod->sbiod_sb->sb_fd, buf, len );
  #endif
  }
  
***************
*** 540,546 ****
        return write( sbiod->sbiod_sb->sb_fd, buf,
                (len<MAX_WRITE) ? len : MAX_WRITE);
  #else
!       return tcpwrite( sbiod->sbiod_sb->sb_fd, buf, len );
  #endif   
  }   
     
--- 540,546 ----
        return write( sbiod->sbiod_sb->sb_fd, buf,
                (len<MAX_WRITE) ? len : MAX_WRITE);
  #else
!       return write( sbiod->sbiod_sb->sb_fd, buf, len );
  #endif   
  }   


Plus another patch to change the def of tcpwrite/tcpread on linux.

-Seth