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=12073>. 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=12073 Regression: Extreme performance issue introduced with 2.3.0 ------- Additional Comments From [EMAIL PROTECTED] 2003-03-07 18:32 ------- Sorry that no one looked into this problem more promptly; of course performance problems are of great interest. The problem appears to be caused by the parameters to the FastStringBuffer object that is used to store the character data of the input document. Currently, it is allocating space in increments of 32 Java chars, resulting in very slow processing of the input document, as you noted. In the xslt20 branch, storage is allocated in increments of 1024 Java chars; on the XSLTC_DTM branch it is allocated in increments of 1024 or 128 Java chars, depending on whether the document is actually a result-tree fragment. I will attach a patch presently that improves the situation, but I'd prefer not to apply it to CVS, because it could degrade the performance for result-tree fragments. I believe the preferred solution will be for the fix from the XSLTC_DTM branch to be applied to the MAIN branch.
