Re: publishing FPWD issue (was Re: Thursday telcon & agenda)

Olivier, Dom --

Thanks for the clarifications of the options for TR.  We need to decide 
this soon.  Should we continue in email? Or do we need to spend time on it 
in telcon?

My overall impression:

** multi-part approach is simpler to implement.
** family approach is conceptually more accurate.

Some in-line comments below...

At 06:24 AM 1/17/02 -0500, Dominique Hazael-Massieux wrote:
>On Thu, Jan 17, 2002, Olivier Thereaux wrote:
> > * if we choose the "family" approach, we'll need authorizations from the
> > director each time and for each part (shortnames and all, see
> > publication rules [2])
>
>each time, that is only for the first publication, not for the following
>ones.

I.e, only need the permissions for FPWD of each part?  Subsequent WDs then 
don't need permission, as I understand.

>
> > * if we choose the "family" approach, we may have trouble linking
> > through dated documents. No problem if we always link to (fragments of)
> > latest versions of the other documents. linking to (fragments of) dated
> > versions of other documents might be a bit painful.
>
>That's certainly the main issue. And even if it is an issue, I strongly
>suggest not to link to the latest version of the documents, since we
>would lose a lot of controls of what the semantic of the given link
>would be over the time. When we cross-reference between a document at a
>date D, it's not very probable that this cross-reference will have the
>same meaning 2 or 3 drafts later.

Yes, that's true.  But on the other hand, don't we need to be sure that all 
of the "latest draft" parts are consistent, when we add a new part or 
change a part?  For example, when we add "Spec Guidelines", then that will 
enable some postponed linking from "Intro".  Or when we change "Proc&Ops 
Example and Techniques", that will certainly required adjustments to 
"Proc&Ops Guidelines".

It seems to me that the all-parts coordination aspect of multi-part is 
applicable, even if we decide that the model is "family".

About Olivier's comment, "painful", for the linking in the family approach 
-- perhaps there are tools to help, and I don't know about them, but it 
seems worse than painful.  As far as I can see, all intra-part links would 
have to be aimed at the TR dated directories, instead of just at the 
filename#anchor, right?

Okay, I could make a 'sed' script for moving from WG space to TR 
space.  E.g., run the script on qaframe-intro and change all 
"qaframe-ops.html#anchor" to 
"http://www.w3.org/TR/2002/WD-qaframe-ops-2002MMYY/qaframe-ops#anchor". 
Does that work?  Is there some other W3C convention to handle these situations?

How does WAI do it?

>
> > "Historically", working groups (CSS, DOM) have chosen the "family"
> > approach. What will we choose?
>
>I think that publishing as separate documents (that is, the "family"
>approach) makes more sense, since they are quite different documents.
>The multipart approach does only work IMHO for parts of the same
>documents.

Conceptually, yes.  Only question is how to manage inter-part links with 
"family" approach.

-Lofton.

Received on Thursday, 17 January 2002 11:58:06 UTC