Re: [mif] I won't be in Taipei for MIF WG

[Alexandru Petrescu <[e-mail]>]

Le 27/10/2011 17:29, Ted Lemon a écrit :
> On Oct 27, 2011, at 10:33 AM, Alexandru Petrescu wrote:
>>> 1. While having working implementation helps if the design is 
>>> sound, you can't use working implementation to substantiate a 
>>> proposal.
>> 
>> Hm? How does this statement compare to "rough consensus and running
>> code" substantiating a proposal?
> 
> There is no rough consensus for your proposal, Alexandru.

And  is there running code for route-option?

> If your implementation is so complicated that it requires you to 
> contribute code to the ISC, that should already be a clue that it's
> a bad idea: DHCP options are simple, not complicated.

Efficiency sometimes rightfully implies code complexity (check e.g.
PacketBB encoding for wireless settings).

> You aren't supposed to have to change the server source code to add 
> an option. You are supposed to be able to just specify the option 
> format, and the existing server should be able to parse it.

I believe the code format is C code currently (e.g. some quoted text).

> This is a feature of every server I know of, certainly including the
>  ISC server, which has an option format grammar, and the Nominum 
> server, which has something very similar.
> 
> The ISC DHCP client is likewise configurable to allow new options to 
> be described without source code changes.

I am not sure about it.

> So the sum total code required to implement the route option should 
> be zero lines.

I am not sure really.  There should be some code (script? bash? textual
conf lines?)

Please provide some sort of explanation or proof or so, if possible,
thank you.

Alex

>