udefine the data model as completely as possible
ucompromises must be fixed
later with much more effort
udefine the semantics as strictly as possible (how
???)
uelements are better to handle than attributes if
authoring is done in SGML
uSGML’s datatyping is not sufficient
uimplement your own based on
architecture attributes
utry to establish linking based on existing data
uit is not always necessary to
follow the link
uuse synergy of natural object
identifiers (e.g. object names)
ukeep flexible in the global data model (locations of
containers)
uthis makes it easier to adapt
on existing tools
ube aware that the files tend to be big (10 - 15
Megabytes):
“terseness is not a design goal for XML (Tim Bray)”