In "Salami Slice" only elements are globally declared, while their types ar=
e still defined anonymously within the element.  In "Venetian Blind" it is =
the reverse.  The complex types are defined globally but the elements are d=
eclared locally within them.  As you are obviously aware, making both eleme=
nts and complex types global maximizes reusability opportunities.  I call t=
his Venetian Salami but it is better known as the Garden of Eden style, coi=
ned by Eve Maler of Sun Microsystems, and I sing it praises in the first ha=
lf of this article.
 
Garden of Eden: The Key to Reuse
http://www.tdan.com/view-articles/7185
 
1.  Some people like to put all their complex types in an enterprise namesp=
ace and reference them from elements declared in small local transactional =
schemas.  This promotes reuse that can be exploited by OO class hierarchies=
 and allows multiple, local synonyms for standardized data structures.  It =
has the advantage of not burdening instance files with multiple namespace p=
refixes since the enterprise namespace is only referenced in the schemas.  =
However, the reuse is not evident from simply inspecting instance files whi=
ch may not be ideal in multi-enterprise data exchange scenarios or with tho=
se who value transparency.
 
2.  Some people like to define everything in one namespace but modularize t=
he schema by partitioning it in a number of schema files, which then refere=
nce each other with xsd:include statements. This can facilitate memory econ=
omies by allowing partitioning by subject matter or commonly needed combina=
tions of elements.  It has the advantage of allowing a change of strategies=
 based on experience without requiring changes of namespace prefixes in ins=
tance files but it entails referencing some metadata to determine in which =
module a given element is declared, unlike option 3.
 
3.  A heavier duty approach is to partition by namespace and reference comp=
onents using xsd:import statements.  This a more explicit approach favoured=
 by international vocabulary standards.  Just try to get it all right the f=
irst time.
 
These are only my initial thoughts.  Perhaps it is time for Mr. Costello to=
 revisit his pioneering best practices?
 
Cheers Jack Lindsey
 
Extensions in another Namespace
http://www.tdan.com/view-articles/7186
 
 
 
 
 
 
 



> Date: Mon, 2 Jun 2008 15:15:55 +0100> From: andrew.j.welch@g...> To:=
 xmlschema-dev@w...> Subject: schema arrangement or architecture> > > Hi,=
> > Given the "salami slice" style of schema (all elements and types are> g=
lobal) are there any established practices for arranging a large> schema? e=
g> > - try to keep to one large file or break it down into several files?> =
- ensure all element definitions are always in a single file, or where> pos=
sible keep element definitions in the same file as their type> definitions?=
> - should all attribute be defined globally?> > etc.> > There seems to be =
quite a lot of information about the different> schema styles out there, bu=
t not much on managing a large schema...> (in this case I'm calling a "larg=
e" schema one with just over 100> different elements each with their own ty=
pe... ~150 attributes)> > Is the secret of maintaining a large schema down =
to good tooling> rather than an intimate knowledge of what's in each file?>=
 > thanks> -- > Andrew Welch> http://andrewjwelch.com> Kernow: http://kerno=
wforsaxon.sf.net/> 
_________________________________________________________________
Try Chicktionary, a game that tests how many words you can form from the le=
tters given. Find this and more puzzles at Live Search Games!
http://g.msn.ca/ca55/207=