Hi Jim,

I cannot say I'm a local folk, I am just keeping one eye to see how life returns here... :)

Pulling the whole thing into the memory may happen, and it depends on how the XSLT code is designed: whether it is based more on "push" or "pull" approach, see e.g. http://www.ibm.com/developerworks/library/x-xdpshpul.html . If your XSLT code is more of "pull", then you are most likely to get the whole thing into memory first...

Sometimes it is possible to convert "pull" to "push", but not always. And generally "push" code is better suited for big files. If you could post your XSLT code here, and if it is not too complicated, I might have a look. But we better move to the adjacent "j-users@xalan.apache.org" list first, as the "dev" list is intended for other sort of topics, I'm afraid...


On 03.07.13 22:01, Jackson, Jim wrote:


Good Afternoon Folks


I had a question if you are open for one.  Found your contact information on the web.


We process on the i-series (as400) and use the command to convert XML to CSV


Do you know of any way to speed up this process (even with small (30mb files)) it takes upwards of ten plus minutes to convert the files.  It also seems like it loads the entire thing to memory first and then exports it (perhaps this is what is causing the problem) ?


I ask this because I have been investigating other PC type XML to CSV converters and they fly through similar size files.


Any thoughts or input ?




Jim Jackson


QSH        CMD('java org.apache.xalan.xslt.Process -in +

              /usr/fil3a/CDUSAGCUSACS23101.xml + 

              -xsl /usr/fil3a/tapdtl.xml -out +