On Mar 29, 2005, at 7:01 AM, Alan Barrett wrote:

> Three options have been suggested:
>
>   a) drop the packet and return 0
>   b) drop the packet and return ENOBUFS
>   c) delay the packet until it can be sent, then return 0
>
> Jonathan has argued against (c).  But that still leaves a choice
> between (a) and (b), so I don't see why Jonathan says that the
> point is not up for discussion.

(b) (current behavior) is specifically allowed by SUSv3, matches our  
current behavior, and is useful feedback for an applications.   
Therefore, it is my opinion that nothing in our kernel should change.

-- thorpej