that would not work, because of how dns(1) presents its interface.

if the file server presented a directory of files then you can merge them.

however dns has a single file that you open, write a request to, and then later
read a reply from.

in this later form you merge the directories you just have two request files,
rather than one request file which offers the service of moth files.

I am not sure I have explained this very well, I hope you understand.

-Steve


> On 2 Jan 2016, at 05:17, Marc Boschma <[email protected]> wrote:
> 
> Still getting my head around Plan9 but wouldn’t mounting the unicast and 
> multicast DNS file servers over the top of each other work? (I assume the 
> order of the mount (bind) would lead to resolution order… but maybe no 
> unified responses.
> 
> Marc
> 
>> On 2 Jan 2016, at 2:42 pm, erik quanstrom <[email protected]> wrote:
>> 
>> On Fri Jan  1 19:32:25 PST 2016, [email protected] wrote:
>>> 
>>>> On 2 Jan 2016, at 7:05 am, Steve Simon <[email protected]> wrote:
>>>> anyone done any work to implement mDNS / bonjour on plan9?
>>> 
>>> No, but I have an interest; just starting out with Plan9 :)
>>> 
>>>> my rough plan is to write a file server which generates /lib/ndb/mdns
>>>> which can be included into your /lib/ndb/local.
>>>> 
>>>> I fear the biggest hassle is the clash of UDP port use may mean
>>>> mDNS must become part of dns(1) rather than a separate file server.
>>> 
>>> Shouldn’t dns(1) only bind to unicast UDP port and thus mDNS could bind to 
>>> the multicast UDP port?
>>> 
>>> Are you only considering resolution or also publishing services?
>> 
>> it would make sense to me to make a dnsudp request file server that manages 
>> requests, and
>> fork (ha!) that task off to it.  this file server would not care if it's 
>> querying normal dns,
>> or mdns.
>> 
>> - erik
> 

Reply via email to