We’re evaluating the apose Cells java interface for some dynamic report generation with smartmarkers - we’re invoking the interface from coldfusion, using a small .xls or .xslx file (with 8 nearly identical sheets), and using process() to fill in some very simple sheets with some resultsets.
With aspose 2.2.1, as we double the number of rows to to the query, the processing time is nearly 4 times as slow, whether we use .xls or .xlsx files. This problem seems to persists in 2.1.2, but not in 2.1.0 - at this point, the processing speed on the .xls files is very fast. (We didn’t evalulate 2.1.1 to see if that contains the regression.)
The cold fusion instance is running with a gig of memory, and this is the only page that is currently running on that dev server.
As an example, using a resultset with 6400 rows (per each of the 8 sheets), we’re seeing the following times just to invoke process() -
aspose 2.2.1 .xls - 380s #similar results with 2.1.2
aspose 2.2.1 .xlsx - 383s
aspose 2.1.0 .xls - 0.7s
We were unable to use our sample on .xlsx on 2.1.0 - it didn’t complain, but generated corrupt .xlsx output. (I assume it might be trying to process the file as an .xls)
That’s nearly a 540 times slowdown on our sample code, just looking at the times for process().
I can provide the ColdFusion code if you like, but the basic steps are:
- instantiate a workbookdesigner
- .open() the file
- .setDataSource() for the various sheets (each with its own query)
Any help you can provide to track down this performance issue would be greatly appreciated - at the moment, this slowdown makes it impossible to use.