[Click] set_ip_header semantics

[Eddie Kohler]

Hi Bart,

Heh, mail delay.

So I looked at this patch and rolled it around.  I then thought maybe the 
better solution would be to rename set_network_header() to 
mark_network_header()... and implemented a patch to that effect.  I showed 
that around and people thought it was even more confusing than the current 
situation.  And it's worth noting that Linux has functions like 
skb_set_mac_header() now, which do exactly analogous things to Click's 
p->set_mac_header().  So I think the status quo stays.  Thanks though!

Eddie


Bart Braem wrote:
> On 04 Nov 2008, at 17:14, Eddie Kohler wrote:
> 
>> I'm interested to look at the patch.  But note that for consistency, 
>> Packet::set_ip_header_anno() would probably imply that that one 
>> accesses the IP header via Packet::ip_header_anno(), and so forth.  We 
>> currently use Packet::ip_header(), which is better....
> 
> 
> Eddie,
> 
> Please find in attachment a patch to do this.
> I do not know whether a ip_header_anno() call is needed, it does not 
> matter whether it is an annotation or not to the user.
> 
> Regards,
> Bart Braem