On Sun, 10 Jun 2007 23:29:35 -0700, obiwanjacobi
<obiwanjacobi@[EMAIL PROTECTED]
> wrote:
>Right. So the only difference between a schema that will split the
>records and a schema that will read in the whole message is that the
>'record' node has a maxOccurs of 1 instead of unbounded. Is that
>correct?
Yes. In your schema, you need to describe ONE instance - not something
that will parse the entire file.
>> Well, this one is (potentially) more difficult than the previous one.
>> Are there ALWAYS exactly two lines? Or can there be some groups with
>> one record, some groups with ten records, etc.?
>There can be one or more records per ID. I'm currently working on a
>solution where I split all the records of the incoming message and
>then aggregate them back using a serial convoy orchestration and some
>custom code.
That is one solution, yes.
>Is there another solution using just the disassembler?
You'd have to build your own custom disassembler component :-/
Another solution is to have a map that uses muenchian grouping
technique to group the incoming elements into records with unique IDs
- and in your orchestration, you can then split up the message like I
did here:
http://blog.eliasen.dk/PermaLink,guid,6c7ac8ec-3f3e-49e4-a15a-76c736d30654.aspx
--
eliasen, representing himself and not the company he works for.
Private blog: http://blog.eliasen.dk
Private email: jan@[EMAIL PROTECTED]
|
