Handling large namespace

[WideBoy]

Hi

I am wondering if someone has any ideas on how I might resolve the
following problem.
I have a large namespace file (1.5 MB) generated from a UML logical
data model, which needless to say, is very large. My worry is that
instance documents based on this namespace will take significant
resources in terms of time, memory and CPU to validate against this
namespace schema.

So my questions are as follows:
1. Is there some way of breaking this large namespace file without
creating several individual files?

2. Are there any other alternative ways of addressing this problem?

3, Why do people choose to translate complex data models into even more
complex namespace file(s)?

Any thoughts or ideas on this would be much appreciated.

Many thanks,

Naran

[Joe Kesselman]

WideBoy wrote:

I have a large namespace file (1.5 MB) generated from a UML logical
data model, which needless to say, is very large.
I'm not sure what you mean by a "namespace file". A schema?

If so, breaking it into smaller pieces would not improve validation
speed. It *is* possible to divide a schema into smaller units, but they
all have to get reassembled into a single model of the grammar before
validation takes place, so it wouldn't save you anything.
3, Why do people choose to translate complex data models into even more
complex namespace file(s)?

Again: Since I don't know what you mean by a "namespace file", I can't
answer the question.
--
() ASCII Ribbon Campaign | Joe Kesselman
/\ Stamp out HTML e-mail! | System architexture and kinetic poetry

[Joe Kesselman]

> I have a large namespace file (1.5 MB) generated from a UML logical

data model, which needless to say, is very large. My worry is that
instance documents based on this namespace will take significant
resources in terms of time, memory and CPU to validate against this
namespace schema.

I forgot to add: An IBM team recently published a paper demonstrating
that schema-aware parsing can actually run *FASTER* than non-validated
parsing, by taking advantage of the schema information to produce a
parser specifically optimized for that kind of document.
--
() ASCII Ribbon Campaign | Joe Kesselman
/\ Stamp out HTML e-mail! | System architexture and kinetic poetry

[Soren Kuula]

WideBoy wrote:

Hi

I am wondering if someone has any ideas on how I might resolve the
following problem.
I have a large namespace file (1.5 MB)

Hi, what is a namespace file?

Soren

[Andy Dingley]

Joe Kesselman wrote:

I forgot to add: An IBM team recently published a paper demonstrating
that schema-aware parsing can actually run *FASTER* than non-validated
parsing, by taking advantage of the schema information to produce a
parser specifically optimized for that kind of document.

Interesting, but that's a "can" not "will". I doubt this is a common
technology any time soon. I also imagine that it would rely on
regularly seeing documents to the same schema - the overhead of
optimizing the parser is surely heavy for the first document.

Nice idea though - I expect we'll see more of it in the future.

[WideBoy]

Joe,

Thanks for taking the time to reply to my questions.
sorry for my tardy response, have been on hol. for a few days.

Basically, what I have is a schema file with a specific namespace which
we hope that other interested parties working in our domain will use.
Pretty much like the w3c schema file with it's namespace. Only in my
case the schema/namespace file is stupidly large with a crazy,
umimaginable set of elements and structures resulting from an OO model
with tonnes of inheritance being employed for no obvious reason. The
OO model is basically a corporate logical data model. That is not
intended to ever translate into any physical database. However, the
intention is that with the creation of this namespace it will be
possible to exchange data via XML messages between compliant
applications.

I gather from respondents that breaking up this file into several
smaller files is not going to improve validation times of XML
documents. I can indeed concur with this from bitter experience. Are
there any other ways of achieving this goal?
I'm not entirely sure why people elect to go down the namespace route
to solve this data-interchange problem, anyway?

Regards,

Naran

[steve_marjoribanks]

>I'm not entirely sure why people elect to go down the namespace route

to solve this data-interchange problem, anyway?

Because it guarantees absolute conformity to the schema. If it doesn't
conform it won't validate. There is no better way for ensuring
standardisation of data than making all the xml instance documents
refer to a single schema.