XML parsing using DOM


this should be changed. Currently DDOCUtils loads the whole DDOC into memory. all XML handling should be changed to Pull type readers ( XMLReader and friends ) to better handle large documents.
Also doc hashing calcs could use streamed reading.
IFilter uses SAX Push model to read the data, again, as filter itself operates under Pull model pull mode reading should be a better match and conserve memory.


Melborp wrote Apr 17, 2009 at 11:32 AM

From big file signing performance point of view this has high impact. As soon as we get time, we will try rewriting the whole thing (signing and verifying) on streams. Patches are of course also welcome :)

kaidokert wrote Apr 18, 2009 at 5:35 AM

XMLLite pull parser: ( alternative is open-source LLamaXML )
IFilterImpl class:

wrote Apr 19, 2009 at 5:13 PM

wrote Apr 19, 2009 at 5:14 PM

wrote Feb 13, 2013 at 8:00 PM