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=12182>. 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=12182 long xpath expression crashes with ArrayIndexOutOfBoundsException in org.apache.xpath.compiler.Lexer.addToTokenQueue() [EMAIL PROTECTED] changed: What |Removed |Added ---------------------------------------------------------------------------- Status|UNCONFIRMED |NEW everconfirmed|0 |1 Keywords| |APIBug, XSLTBug Summary|calling xalan:evaluate on |long xpath expression |long xpath expression |crashes with |crashes with |ArrayIndexOutOfBoundsExcepti |ArrayIndexOutOfBoundsExcepti|on in |on |org.apache.xpath.compiler.Le | |xer.addToTokenQueue() ------- Additional Comments From [EMAIL PROTECTED] 2002-09-05 06:07 ------- [Changed summary from "calling xalan:evaluate on long xpath expression crashes with ArrayIndexOutOfBoundsException" to "long xpath expression crashes with ArrayIndexOutOfBoundsException in org.apache.xpath.compiler.Lexer.addToTokenQueue()"] Ok, I've taken a look at the code and found the problem, but am hesitating as to the best way to fix it. (By the way, it has nothing to do with xalan:evaluate, it's just a problem with long xpaths in general). In org/apache/xpath/compiler/OpMap.java there's a variable MAXTOKENQUEUESIZE that's set to 500. An array m_tokenQueue in org.apache.xpath.compiler.OpMap is allocated with this size. org.apache.xpath.compiler.Lexer.addToTokenQueue() adds to this array, overrunning it (throwing the exception) when it hits the end, without checking it first. My question is now this: how public is org.apache.xpath.compiler.OpMap? Would changing its public m_tokenQueue attribute from an Object[] to a Vector (and all places that use it in the xalan package) be acceptable, or is that public enough to be used by someone who'd have code break on them as a result? If not, are people ok with allocating a new larger m_tokenQueue array and copying over the old when we reach the max size? I don't believe having a set limit is acceptable, so just bumping the limit from 500 up doesn't cut it, nor does throwing a different exception (though addToTokenQueue should obviously never throw an AIOOBE). Comments?
