[dcc2] New draft

[James Ross]

Dan Smith wrote:
> Hi everyone,
> 
> Well I know I said one week, but it has been 4!  I have tried to add all 
> the elements we touched on in our discussions into the draft.  The major 
> changes are:
> 
> 1. Groupings of supported tokens
> 2. Optional Groupings
> 2. Error tokens and error messages
> 3. A section on backwards compatibility with historic dcc
> 4. Application token to specify the application layer protocol
> 5. Network and Transport tokens
> 
> Please take a look at this revision, make sure I didn't miss anything, 
> and if it all looks ok we will send it out to the ietf as an ID for some 
> initial comments.
> 
> http://www.dcc2.org/specifications/draft-smith-irc-dcc2-negotiation-00.htm

Here are my comments on it...


Is there any particular reason why FILE, PORT and SIZE (and MULTI?) are 
specified in all upper-case, different from the other tokens?

I couldn't find any reference to case-sensitivity in that spec, but I'd 
assume the token names weren't case-sensitive. Perhaps it would be an 
idea to explicitly state that?


In the section 4.1.2 Publish Syntax, you have:

   dcc2-publish = `DCC2 Publish` space ConnectionTokens [space AppTokens]

Is there a good reason to force the application-specific tokens to be 
after the connection tokens? I can see they have differently specified 
values, basically just allowing them to be double-quoted, but why not 
simplify this and just allow all tokens to have double-quoted values, or 
the subset allowed without quotes?


   filepunct = %d33 %d35-41 %d43-36 %d61 %d64 %d91 %d93-95 %d123 %d125-126

I can't read that at all, making it really hard to have any idea what's 
valid or not. Also, %d43-36 makes no sense to /me/. Looking at the 
decoded characters, I guess that should be %d43-46.

A comment saying that filepunct is simply the set !#$%&'()+,-.=@[]^_{}~ 
would be very useful. :)


The examples don't include the word "Publish" as required by the ABNF. 
Which is wrong?


In 4.2.2 Accept Syntax, the SID token is being fixed in its position. Why?


In 4.2.3.2 DCC2 SEND, the 3rd example has the line:

   DCC2 CannotAccept SID=1 ErrorToken = NAT

Typo? (spaces around '=')


I found having examples of each part of the conversation useful, but as 
an interested reader a full example conversation wouldn't hurt.


> One thing to note is that we reference the CTCP specification, but it is 
> not really a specification, just a newsgroup post.  To reference it in a 
> draft, it should itself be a standard.  Should we formalize the CTCP 
> protocol documentation?  And if so, would anyone like to help undertake 
> the task of writing it as an Internet Draft?

That would mean we would have to support it properly, wouldn't it? ;)

-- 
James Ross <silver at warwickcompsoc.co.uk>

Sorry for not being null terminatedþ?lÀE¸??F#óóÕ·?{
           -- Bugzilla Quip List