[Exist-open] Quick questions on Node passing, Xquery Storing and Tunning

[<wma...@AV...>]

Hello.
I'm actually trying to test eXist update/query capabilities and wanted =
to make things right. Here are some newbie questions that may help me:

1. In one scenario, I have a document stored in the database, and a =
string in Java containing a document with some <a> element as a =
children. I need to insert all <a> in the document into the stored =
document. I'm actually storing the document in the string into a =
temporal document in the database, and passing to the query its =
name($source_doc), so the query would look like:
for $n in fn:doc($source_doc)//a
return
update insert $n into fn:doc($dest_doc)//father_of_a=20

I've read messages from 2004 saying there is problem passing nodes from =
Java to be queried. Is that solved? Does it affect the above scenario? =
Since I have the string to start with, what would the be best approach =
for performance: parsing String to DOM and passing the dom, or doing it =
like I'm doing it (storing the string as a document and then query =
between persisted documents)?
Also, I should stored the document in the string at each run, but I'm =
simply storing using the same name of the temporal document, thus =
replacing it. Would it be better to delete the document at the end of =
the call so the next call will put a new document in? I ask this to know =
if a replace is less efficient than a delete/insert procedure.

2. I'm actually running the queries (as small and simple as the one in =
point 1) using a compiled query (exQryService.compile(sQuery)). There is =
the option to store a module (*.xq file) and perform imports. Since that =
would need in java also a query to be created (doing the import and =
performing the call to the function), I wonder which of the two =
approaches would give me the best performance (having the code in =
compiled queries or having it in modules). Note that the scenario will =
perform that query several times (call the query from Java).

3.  Are there any tips/tricks to improve performance in the following =
scenarios?
a. Inserting big XML documents.
b. inserting/deleting/replacing elements that are 6-10 levels deep in =
the hierarchy.
c. Query elements that are 6-10 levels deep in the hierarchy based on an =
element's content value.
d. Returning big XML documents.

Thanks for your help!.

_______________________________
William Mart=EDnez Pomares
Architect
Avantica Technologies