Re: patch: sockaddr instead of mbuf to carry addresses

[Roy Marples <roy%marples.name@localhost>]

On 25/03/2015 01:06, Christos Zoulas wrote:
> On Mar 24,  8:36pm, rtr%netbsd.org@localhost (Tyler Retzlaff) wrote:
> -- Subject: Re: patch: sockaddr instead of mbuf to carry addresses
> 
> | hey,
> | 
> | On 3/24/2015 7:47 PM, Christos Zoulas wrote:
> | > In article <20150324230053.AB04314A1D3%mail.netbsd.org@localhost>,
> | > Mindaugas Rasiukevicius  <rmind%netbsd.org@localhost> wrote:
> | >> Tyler Retzlaff <rtr%netbsd.org@localhost> wrote:
> | 
> | 
> | > I think so too; the only reason we are not using sockaddr_storage for it
> | > is because we allow sockaddr_un to be up to 253 characters long as an
> | > extension to the spec. I kind of wished that sockaddr_storage was defined
> | > to be 256 bytes instead of 128...
> | 
> | I have no particular objection to doing it this way, I just thought 
> | there would be objection to changing the size of struct sockaddr_storage 
> | which is why I went with the new structure.
> | 
> | If it is better to increase the size of struct sockaddr_storage shoot me 
> | a preferred updated structure definition and I'll go with that.
> 
> I am not sure which is less of a pain to do... (binary compatibility?)

That would only be incompatible if sockaddr_storage was documented as
part of the API of something yes? I would be suprised if we had anything
like that.

Roy