Re: Working group process and the future of the WG.

Ted,

thanks for the summary.

Below a few thoughts and clarifications...:

PROTO

I absolutely agree that the PROTO procedures make a lot of sense, in 
particular the requirements to come up with a detailed write-up when a 
document is to be submitted. If the WG chairs feel that they need help 
in collecting and summarizing this information, they should point that 
out, so that the WG can assist.


Ticketing System

I explicitly disagree on the comments about the ticketing system that 
was installed by Joe H. As far as I can tell, we had a WG consensus on 
how to use it for the WG Last Call for BIND, but for some reasons the WG 
chairs at some point decided not to follow that procedure anymore 
(without actually stating that). Reminder: that procedure was defined in 
<http://lists.w3.org/Archives/Public/w3c-dist-auth/2005JanMar/0001.html>, 
and as far as those rules are concerned, the BIND spec finished last 
call successfully. It would be really good if the WG chairs would either 
acknowledge this, or, alternatively, spell out exactly what they think 
is left to do.


WG future

I think it makes sense to distinguish different types of drafts the WG 
is currently working on.


Group #1: Advanced Collections

BIND and REDIRECT have been through a WG last call back in 2000 (as 
their sibling ORDERED COLLECTIONS, which we successfully finished in 
2003). Both specs are stable in that their authors are not aware of any 
open issues and in that there are implementations out there. As 
mentioned before, BIND went through a second WG last call in January.

At this point, the WG needs to decide whether these specs should be 
published through the WG at all, and if so, at which level. Not 
publishing these specs at all IMHO will negatively affect 
interoperability of existing implementations, though.

BIND is currently implemented in (at least) two server projects (one of 
which open source), and one client. Looking at the RFC Standards Process 
(as pointed out by Jim W on Friday), it really falls into the "Proposed" 
category.

REDIRECT is currently implemented in (at least) one server and two 
clients (one of which only implements discovery, but not creation of 
redirects). Again, this spec seems to fall into the "Proposed" category.

On the other hand, if we can't get consensus for publication as 
"Proposed", I think it would be the right thing to do publish as 
"Experimental". Not publishing at all clearly harms interoperability and 
causes valuable information to disappear from the set of documentation 
implementers usually can find.


Group #2: Quota

This spec has been added (without prior discussion) to the WG's work 
items; and has been extremely slow progressing. It now has reached a 
status without known issues. Unfortunately (but not surprising giving 
the very slow speed it has progressed), both authors do not work anymore 
for their respective companies, so what we should do is to find out 
whether the current document actually meets the requirements of those 
who originally wanted it (Xythos and Apple; as far as I can tell, we 
already got a nod from Apple's Jim Luther in 
<http://lists.w3.org/Archives/Public/w3c-dist-auth/2005AprJun/0069.html>).


Group #3: RFC2518bis

This is certainly the most important piece of work. It's really 
frustrating that all the effort spent on clarifications and 
simplifications (in particular for locking) hasn't led to a publication 
yet. For instance, just a few weeks ago the Subversion developers wasted 
a lot of time trying to implement the deprecated lock null resource 
semantics. I have offered by help in getting things done multiple times; 
and now that BIND and REDIRECT seem to be done I'm really looking for that.


To summarize: I personally think that the problems this WG has are caused by

- Draft editors responding too slow to issues (or not responding at 
all), and

- inability of the WG to manage the document submission once the 
technical and editorial work is done (meaning to either actually submit 
the document, or to precisely point out what's left to do).

If we fix these problems, we can submit BIND/QUOTA/REDIRECT quickly, and 
then focus all of our energy to get RFC2518bis done.


Best regards, Julian

Received on Sunday, 29 May 2005 14:40:23 UTC