W3C home > Mailing lists > Public > spec-prod@w3.org > October to December 2011

Re: References Re: What are the requirements/problems? Re: Working on New Styles for W3C Specifications

From: Marcos Caceres <w3c@marcosc.com>
Date: Wed, 14 Dec 2011 23:18:07 +0000
To: Jim Melton <jim.melton@oracle.com>
Cc: Robin Berjon <robin@berjon.com>, "chairs@w3.org Chairs" <chairs@w3.org>, "spec-prod@w3.org Prod" <spec-prod@w3.org>
Message-ID: <F3CEFC60C8F24020A6F72D35E815971A@marcosc.com>


On Wednesday, 14 December 2011 at 23:03, Jim Melton wrote:

> Marcon,
> 
> At 12/14/2011 03:02 PM, Marcos Caceres wrote:
> 
> > > This brings me to a convention I've been advocating here and 
> > there but that hasn't been formalised. There are use cases both for 
> > point to the latest version, and for pointing to a specific version 
> > that you don't want to see change beneath your feet.
> > Whoa! hang on. No Working Draft should be cited that way (i.e., as 
> > stable!). That's why HTML5 has the *big red warning* and document's 
> > SoTD always says that documents may be obsoleted at any time (unless 
> > they are Recs) and it's inappropriate to cite them as anything but a 
> > work in progress.
> 
> 
> 
> I beg to disagree. My Working Group (in conjunction with a sister 
> WG) publishes up to ten or eleven documents in sync. When we publish 
> a WD for one of these documents, we want their references to other 
> documents in the group to be explicitly to the dated WD, not to the 
> "latest version". Why? Simple -- when we have a need to reference 
> an older version in TR date space, we want its links to be fully 
> consistent with that specific document.

Even if you are explicitly saying "implement this buggy dated unfinished spec"?  
> There is no implication of "stability", but one of consistency.

That does not strike me a as fair to implementers (who are the ultimate recipients of our work). Would you tell implementers that to implement your spec, they should also implement an outdated specification that you reference? And that if they implement a newer version, they need to revert their work to a past version to match your spec's references?  
> 
> You also said:
> 
> > > Perhaps we could also have a [FOO-ED] convention?
> > A lot of editors would prefer the Editor's Draft to actually be the 
> > authoritative draft. Appearing on /TR/ MUST NOT be taken as a sign 
> > of maturity or stability (unless the spec is REC or PR). All other 
> > statuses are as unstable as each other.
> 
> 
> 
> Again, I disagree that this is always the right thing to do. In my 
> WG, we rarely have our many editor's drafts in sync except when we're 
> ready to publish a set of public WDs. I don't object to some WGs 
> making their Editor's Drafts authoritative, but I object strongly to 
> the presumption that all WGs will do so.

Sorry, I did not mean to imply that. My point was simply that a Working Draft on /TR/ has no claim of stability over any other draft (until it becomes rec). Stability is determined by other criteria (e.g., bugs against the spec + complete test coverage of conformance requirements + n number of implementations that demonstratively pass tests in the test suite) . 

 
Received on Wednesday, 14 December 2011 23:18:43 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Saturday, 10 March 2012 06:19:18 GMT