[xsd-users] Some performance problems with XSD.

[Dmitry Trifilov]

Hello,
I have some performance problems with code generated by XSD.

Code generated by XSD parses my XML file during 35 - 40 seconds, this 
time is too long for my task.
My XML file have one feature: It contains element "<words_sequence>" 
which contains about 6000 subelements "<Word>".

I use flag "xml_schema::flags::dont_validate" during parsing.
I have foound out that control flow spends a lot of time within function
"void WordsSequence::parse (::xsd::cxx::xml::dom::element< char > const& 
e,  ::xsd::cxx::tree::flags f)" from
KMParsingSettingsSchema.cpp generated by XSD.

I have created simple test which parses my XML file using xerces and 
puts content of  "<Word>" elements to "std::vector<std::string>"
object and I have found out that test works approximately for 2 seconds.

(XSD schema and XML file are gzipped and attached).

My environment is:
Xerces version is "2.6.0".
XSD version is "2.1.0"
gcc compiler version is "Red Hat Linux 3.2.3-54"

I am not familiar with XML schemas, may be it is my bag in XML schema 
definition?
How can I improve performance and where is the source of my performance 
problems?

Thanks in advance
-Dmitry

-------------- next part --------------
A non-text attachment was scrubbed...
Name: KMParsingSettingsSchema_xsd.tgz
Type: application/octet-stream
Size: 1043 bytes
Desc: not available
Url : http://codesynthesis.com/pipermail/xsd-users/attachments/20060809/012a25ed/KMParsingSettingsSchema_xsd.obj
-------------- next part --------------
A non-text attachment was scrubbed...
Name: KMParsingSettingsSchema_xml.tgz
Type: application/octet-stream
Size: 47960 bytes
Desc: not available
Url : http://codesynthesis.com/pipermail/xsd-users/attachments/20060809/012a25ed/KMParsingSettingsSchema_xml.obj