Re: [FrameSGML] FM+SGML: everything valid highest level?

[Mark Barratt <markb@xxxxxxxxxxxxxxx>]

> At 08:11 PM 3/23/02 +0000, Mark Barratt wrote:
> 
>>When dealing with arbitrary import-by-refrence file fragments I'm
>>tempted to let everything down to para and list level be valid at the
>>highest level.
>>
>>Is there some reason why this might be a very bad or impractical idea
>>(apart from philosophically, that is)?
>>


Thanks to Dan and Lynne for reasoned and comprehensive responses. Both 
suggest using a fragment wrapper. Lynne suggests SGMLFragment, which is 
a special-case container that unwraps itself on import. Ideal, but she 
points out that the implementation of this container in Frame has a bug 
which causes it to work properly only if the imported fragment is (apart 
from SGMLFragment itself) contained in a single element. No good for me 
as the imported material is often sequences of lower-level stuff 
(typically, three or four paragraphs).

Dan proposes a similar solution, but using an 'insert' element which 
remains on import. This looks good but has the practical disadvantage 
that it adds quite a lot of complication to formatting and export - the 
structure uses relatively few elements with meaning and print formatting 
  implied by relative and absolute positions, and one destination 
requires some already-complex XSL transformations.

In one way the problem is simpler than it may seem: I don't need to hold 
the text insets as SGML fragments, but as structured Frame files, and I 
don't much care about validating the inset files while editing them, 
because they are structurally simple. What happens if I do nothing at 
all to the structure rules is that inserted files wrap themselves in a 
NoName element, and the NoName can be unwrapped in the destination file. 
A change in the inset file causes the NoName element to reappear, but it 
does not reappear if I save and reopen the destination file unless 
insets have changed. For the editor, I could argue that this is a useful 
feature, as it alerts him/her to the need to review changed shared 
material in case it has implications for specific text in the current 
document. I think this is called post-hoc rationalisation.


BTW, a moment's thought or experiment would have showed me that allowing 
  any element to be valid at the highest level doesn't help: a sequence 
of such elements without a parent is invalid in anyone's book.

Thanks for the input. And thanks to the Frame team for allowing files to 
open, display, save and export with invalid content and a warning, 
rather than requiring a valid instance at every stage. An underrated 
feature, I suspect.

best


------------
Mark Barratt
Text Matters
37 Upper Redlands Road,            Information design:
Reading RG1 5JE, UK                We help explain things using
phone +44 (0)118 986 8313          .language
fax   +44 (0)118 931 3743          .design
email markb@textmatters.com        .systems
web   http://www.textmatters.com   .process



** To unsubscribe, send a message to majordomo@omsys.com **
** with "unsubscribe framers" (no quotes) in the body.   **