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=26119>. 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=26119 Crash in FormatterToXML on Solaris, having compiled Xalan using the CC compiler (Forte 6u2), when Hoard is loaded. Summary: Crash in FormatterToXML on Solaris, having compiled Xalan using the CC compiler (Forte 6u2), when Hoard is loaded. Product: XalanC Version: 1.6 Platform: Sun OS/Version: Solaris Status: NEW Severity: Critical Priority: Other Component: XalanC AssignedTo: [EMAIL PROTECTED] ReportedBy: [EMAIL PROTECTED] System setup: Using XalanTransform sample application, with simple XML & XSL, with LD_PRELOAD set to load the hoard memory manager (I'm using libhoard.so.1.0.7). I've used both self-built binaries, and prebuilt binaries from the Xalan site. A SIGSEGV results, and running the XalanTransform with watchmalloc enabled (LD_PRELOAD set to watchmalloc.so.1, MALLOC_DEBUG set to WATCH) results in a TRAP on the way out of the FormatterToXML constructor. Hoard makes this problem more visible - without Hoard loaded no crash occurs, although judging from the results of running watchmalloc, a problem still exists. Further investigation revealed that the initialization within that constructor of m_encoding using the ?: 'conditional' operator is responsible - I am of the belief, having discussed this issue with a colleague, that the Forte compiler has problems with scoping wrt. the ?: operator, and that this is almost certainly the cause of this problem. I have personally fixed this issue by moving the m_encoding initialization into the constructor, and replacing the ?: with the equivalent if{} else {}. A similar fix, assuming it meets with your approval, should probably find it's way into the Xalan source. Thanks.
