RE: [docbook-tc] Open issues in DocBook transclusion proposal

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

> -----Original Message-----
> From: Jirka Kosek [mailto:jirka@kosek.cz]
> Sent: Thursday, 2010 September 16 8:49
> To: DocBook Technical Committee
> Subject: [docbook-tc] Open issues in DocBook transclusion proposal
> 
> Hi,
> 
> as per my action item from the latest teleconference I'm sending list
> of
> issues which are to my knowledge still open in transclusion proposal
> together with proposed resolutions.
> 
> 1. When the profiling is applied?
> 
> Effectivity attributes can be specified both on def and ref elements.
> Someone pointed out that in DITA something like this was underspecified
> and could lead to different results depending on the mutual order of
> transclusion and profiling processing. For now I can't think of such
> example from the top of my head. So my current thinking is that for def
> and ref elements we should say that effective are only those ones that
> satisfy profiling conditions and that profiling applies after
> transclusion (this is necessary for effectivity attributes used on
> elements inside <def>.).

I guess it depends on what can constitute a profiling
condition, but I'm assuming that something that gets
transcluded could affect profiling.  If so, then which
defs and refs are effective could depend on what gets
transcluded, so you can get into a circular thing here.

If we just say "first transclude, then profile" you may
end up processing defs (and thereby possibly overidding
other defs) only to discover that the def should not
have been effective, but it's too late.

We could say "first profile only defs and refs, then
transclude, then profile everything else".  I think that
may be what Jirka is saying above.  I think that avoids
circularity and is unambiguous, but it does mean two
profiling passes, and it may produce results that are
hard for users to understand.

I don't know what the right answer is here.

paul