On Fri, Nov 05, 1999 at 02:48:52PM +0100, Martijn van Beers wrote:
> On Wed, Nov 03, 1999 at 06:13:50PM +0000, Damon Chaplin wrote:
> > Martijn van Beers wrote:
> > > Yes, that's my aim. As Daniel Veillard said it, elements are to define
> > > structure, and the structure should be as close to gtk+'s structure
> > > as possible
> > I'm not sure its a good idea to complete change the XML format now.
> > Maybe if there are big advantages, but I'm not sure using attributes for
> > properties gains us much.
> 
> I'll let Daniel Veillard comment on the exact advantages of using
> attributes, since he's the expert.

  People usually use attribute when:
    - they want the given information to not really appear as such in
      the document tree structure.
    - usually it's for "small" pieces of informations
    - and it's rather fixed, i.e. attributescannot easily evolve in more
      complex structures.
    - and default values for attributes can be inherited if you use a DTD.

and you will not use an attribute but an element instead if:
    - that part of the information may be extended in future versions
      i.e. made more complex than a simple string
    - you want this information to appear in the structure
    - you want to localize the content (it's possbile to provide
      multiple values, one per language, while this cannot be done
      cleanly with just attributes).
    - you want to be able to point to that information with an
      ID/IDREF reference (future pointer for XML will be able to
      address attributes).
    - you're not afraid by a bit more verbosity (compared to an
      attribute based container).

  Then looking at those hints, I was thinking it would make sense
to try to keep the structure close to the actual widget tree. Whether
properties should be encoded in attributes or not is not a fundamental
problem, with the exception that if a property contains a rendered
string, then it should be encoded as an element to be able to have
multiple language values in the xml serialization.

  I don't think I have enough knowledge of the current Glade needs
to really be able to give good advices in one way or the other. 
I would just like to point out that by using an appropriate namespace
and version informations it's perfectly possible to design a file
format which is extensible and if needed such a scheme should be used
as soon as possible since we all expect Glade use to grow exponentially
in the following months, right ;-) ?

  So keep up the good work and if a format redesign is needed,
then the sooner the better !

Daniel

-- 
[EMAIL PROTECTED] | W3C, INRIA Rhone-Alpes  | Today's Bookmarks :
Tel : +33 476 615 257  | 655, avenue de l'Europe | Linux, WWW, rpmfind,
Fax : +33 476 615 207  | 38330 Montbonnot FRANCE | rpm2html, XML,
http://www.w3.org/People/W3Cpeople.html#Veillard | badminton, and Kaffe.

+---------------------------------------------------------------------+
To unsubscribe from this list, send a message to [EMAIL PROTECTED]
with the line "unsubscribe glade-devel" in the body of the message.

Reply via email to