RE: [dita] possible spec update to include in DITA 1.2 (or fix in 1.1)

["Grosso, Paul" <pgrosso@ptc.com>]

Robert’s description would be OK, but does the DITA 1.1 spec actually say this anywhere?  

Is it really the case that if I have something like the following, that I have to specify collection-type="family" on each level except the last if I want it to apply to all of the descendants?

  topicgroup collection-type="family"  or relcell "collection-type="family"
    topicref href=xxxx
        fopicref href=yyyyy
            topicref href=zzzzz

Here is the text from the DITA 1.1 Language Spec:

Collection types describe how links relate to each other. A family collection represents a tight grouping in which each of the referenced topics not only relates to the current topic but also relate to each other. Sequence indicates that the order of the child topics is significant; output processors will typically link between them in order. Unordered indicates that the order is not significant. Choice indicates that one of the children should be selected. If no value is specified, processors should treat the default as ″unordered″, although no default is specified in the DTD.

If I read this carefully, it seems to be saying something different for collection-type=”family” where it mentions referenced topics, but does not mention children and the other possible values where it does mention children.  

And how does this work within a reltable where you would have something like this:

  reltable
    relheader
    relrow
       relcell
         topicref
            topicref

If I specify collection-type on a relcell, does it apply to just the first level topicref?

If I specific collection-type on a relrow, does it apply to just the relcells (children) and not to the topicrefs within the relcells?  What good is it, if this is true?  Same question for collection-type on reltable?

And this is what it says in the DITA 1.1 Architecture Spec:

The collection-type attribute indicates how a particular set of sibling topicrefs relate to each other. The collection-type attribute is set on the container element for the sibling topicrefs. The collection-type value can indicate whether to generate links among the siblings, and what kind of links to generate (for example, next and previous links for a sequence, or sibling links for a family). The collection-type attribute can also indicate how the parent topic should link to its children (for example, showing the child links as a numbered list when the collection-type is sequence). 

This is a better match with what Robert described. It doesn’t seem to exactly match the Language Spec, or at least the Language Spec isn’t as clear or as explicit.  And I’m still left with the question about collection-type on reltable and relrow.  I guess you could argue that relrow, relcell, and reltable are all containers for the first level contents of relcells, but that starts to deal with containers within containers and once you start down that road where do you stop (topicrefs within other topicrefs, topicrefs within topicgroups or topicheads, ...)?

paul

> -----Original Message-----
> From: Robert D Anderson [mailto:robander@us.ibm.com] 
> Sent: Monday, 2007 March 19 16:03
> To: DITA TC list 
> Subject: RE: [dita] possible spec update to include in DITA 1.2
> 
> The collection-type attribute actually does not inherit anywhere, and
> always applies to the direct children. So when specified on 
> the map, it
> actually specifies a collection type for the top-level 
> topicref elements
> that are immediately within the map element. At least, that 
> has always been
> my understanding.
> 
> Thanks-
> 
> Robert D Anderson
> IBM Authoring Tools Development
> Chief Architect, DITA Open Toolkit
> (507) 253-8787, T/L 553-8787
> 
> "Grosso, Paul" <pgrosso@ptc.com> wrote on 03/19/2007 03:53:40 PM:
> 
> > In addition to navtitle/locktitle are there other attributes
> > that occur on the map element or its specializations that are
> > in the same situation as locktitle (the value isn't inherited
> > by descendants and doesn't have much meaning on its own on the
> > map element itself).
> >
> > There are three attributes in this category: collection-type,
> > locktitle, and chunk.  And we've dealt with chunk already, so
> > that leaves just collection-type and locktitle.
> >
> > paul
> >
> > > -----Original Message-----
> > > From: Robert D Anderson [mailto:robander@us.ibm.com]
> > > Sent: Monday, 2007 March 19 15:48
> > > To: DITA TC list
> > > Subject: RE: [dita] possible spec update to include in DITA 1.2
> > >
> > > Based on the resolution at the January 16 meeting, I was
> > > supposed to follow
> > > up on this on the list, so that we could resolve it as a
> > > public comment in
> > > the 1.1 specification. Based on that, and assuming I manage
> > > to bring the
> > > issue up again, I think it should not be needed on the 1.2 list.
> > >
> > > I will find the notes from January and see if there is any
> > > conclusion on
> > > the topic. I think that one or two things might still be
> > > open. I'll do that
> > > this week (probably not before tomorrow's meeting, but still
> > > well before
> > > the end of the public comment period).
> > >
> > > Robert D Anderson
> > > IBM Authoring Tools Development
> > > Chief Architect, DITA Open Toolkit
> > > (507) 253-8787, T/L 553-8787
> > >
> > >
> > >
> > >
> > >              "Grosso, Paul"
> > >
> > >              <pgrosso@ptc.com>
> > >
> > >
> > >           To
> > >              03/19/2007 03:41          "DITA TC list "
> > >
> > >              PM
> > > <dita@lists.oasis-open.org>
> > >
> > >           cc
> > >
> > >
> > >
> > >      Subject
> > >                                        RE: [dita] possible
> > > spec update to
> > >                                        include in DITA 1.2
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > > I think Navtitle and locktitle discussion to which you refer
> > > is what I meant to reference.
> > >
> > > So, yes, that should be on the 1.2 list too.
> > >
> > > paul
> > >
> > > > -----Original Message-----
> > > > From: Robert D Anderson [mailto:robander@us.ibm.com]
> > > > Sent: Monday, 2007 March 19 14:32
> > > > To: DITA TC list
> > > > Subject: Re: [dita] possible spec update to include in DITA 1.2
> > > >
> > > > Hi Paul -
> > > >
> > > > One item was to clarify the use of locktitle on a map. There
> > > > was a bit of
> > > > remaining discussion, and I did not want to hold up the
> > > > approval of the
> > > > committee draft for such a trivial item, so I agreed to 
> try and get
> > > > consensus on the email list, and then submit the resolution
> > > > as a public
> > > > comment for the spec update. I haven't followed up on that
> > > > yet, and need to
> > > > do so.
> > > > http://lists.oasis-open.org/archives/dita/200701/msg00058.html
> > > >
> > > > I'm not sure if there are others that should be on the 
> list for 1.2.
> > > >
> > > > Robert D Anderson
> > > > IBM Authoring Tools Development
> > > > Chief Architect, DITA Open Toolkit
> > > > (507) 253-8787, T/L 553-8787
> > > >
> > > > "Grosso, Paul" <pgrosso@ptc.com> wrote on 03/19/2007 
> 01:41:25 PM:
> > > >
> > > > >
> > > > > I haven't dug it up, but I believe there was some discussion
> > > > > earlier in the TC of changing or defining the meaning of some
> > > > > attributes when they appear on the root element in a topic
> > > > > (toc was one, I think there are others).
> > > > >
> > > > > Does anyone remember this?  Is this still something we want
> > > > > to do in DITA 1.2?
> > > > >
> > > > > paul
> > > >
> > >
> > >
>