DO NOT REPLY TO THIS EMAIL, BUT PLEASE POST YOUR BUG RELATED COMMENTS THROUGH THE WEB INTERFACE AVAILABLE AT <http://nagoya.apache.org/bugzilla/show_bug.cgi?id=796>. ANY REPLY MADE TO THIS MESSAGE WILL NOT BE COLLECTED AND INSERTED IN THE BUG DATABASE.
http://nagoya.apache.org/bugzilla/show_bug.cgi?id=796 In multi-level sort, must raise warning if any level is bad [EMAIL PROTECTED] changed: What |Removed |Added ---------------------------------------------------------------------------- AssignedTo|[EMAIL PROTECTED] |[EMAIL PROTECTED] ------- Additional Comments From [EMAIL PROTECTED] 2002-01-22 13:49 ------- The example given in testcase sorterr10.xsl should not be considered as a negative test. sorterr10.xsl provides a reference to a non-existent xml-element mo: key('MonthNum',mo) instead of key('MonthNum',month) This should be considered as an error of the stylesheet writer, not matching the structure of the xml data to be processed. What is returned by the select-expression is an emtpy node-set, not establishing an order between the nodes. In this case processing should proceed on the next deeper level. Often you have uncomplete or erroneous data. If Xalan is forced to stop sorting, you won't get what you expect: At least the elements with valid data entries should be sorted in the appropriate order. (It would be desirable to control more explicit the handling of empty node-sets in <xsl:sort> using additional arguments on xsl:sort, maybe XSLT 2.0 spec. ? ) As in the context of XSLT 1.0/1.1 I would consider the bug described as invalid. The testcase is leading to a wrong behaviour and should be replaced. ------- Additional Comments From [EMAIL PROTECTED] 2002-01-22 14:10 ------- Xalan shows an inconsistent behaviour in multi-level sorting. If in the first <xsl:sort> an empty node-set is selected, the first and second levels are skipped, but the entires are sorted against any deeper levels if present. This is due to the implementation specific caching of the values returned by evaluating the first two selection criteria. It would be reasonable to provide a further testcase with three levels with the first one beeing unusuable.
