Re: CVS commit: src/sbin/ping

[christos%astron.com@localhost (Christos Zoulas)]

In article <rmik3rv4kwp.fsf%fnord.ir.bbn.com@localhost>,
Greg Troxel  <gdt%ir.bbn.com@localhost> wrote:
>-=-=-=-=-=-
>
>
>I am not understanding why '-s argument is the number of data bytes
>beyond the standard 8-byte icmp header' is complicated.   People who use
>-s are trying to control packet size, and the rule of 8 header + data
>seems to be longstanding.
>
>Looking at -current:
>
>  the size of the timespec should be given, and it's endianness.
>
>  the man page for -C should say that these compat timestamps are
>  instead of the timespec.
>
>so 
>
>ping -s 8: should send 16-byte packets in compat format
>ping -s 16: should send 24-byte packets in new format
>
>It's at best odd to have the packet length change for a given -s
>argument because of timestamp format.  This points out that the original
>decision should have been '-s X means X bytes of IP payload', but that's
>not what was decided, and it's easy enough to stick with 'X bytes beyond
>the 8-byte ICMP header'.
>
>      [10:58am] 2277>ping  quasar
>      PING quasar.astron.com (192.168.2.4): 48 data bytes
>      64 bytes from 192.168.2.4: icmp_seq=0 ttl=255 time=0.027 ms
>
>That's a bug.  All bytes after the 8-byte header are data bytes, even if
>some of them have timestamps.

I agree, are you fixing it?

christos