It's worse than I thought. I confirmed that the code runs correctly under 1.5.0_05, and it does not run correctly under 1.4.2_06 or 1.4.2_10 unless run in interpreted mode. It appears there's a problem with HotSpot under 1.4.2 that was never fixed and which is causing this issue.
I'm a bit burned out right now to start preparing a bug report for Sun, but there you have it. It does not appear to be an issue with the compiler, since I did not rebuild for 1.5.0_05. How irritating. - Charlie On 1/14/06, Charles O Nutter <[EMAIL PROTECTED]> wrote: > Ugh. I've just spent the past two days hunting down impossible causes > for a null pointer, and it now appears likely that it's an Eclipse > compiler bug. When built using ant under Eclipse, which uses Eclipse's > Java compiler, I'm getting a null pointer exception on > state.getThreadContext().getRubyClass().whatever within the > ConstNodeVisitor. If I run that same built code with -Xint, > interpreted mode, it runs fine. It also runs fine in all modes when > built using Sun's compiler. > > I found one bug that might be related that's fixed in >3.1.1. > > https://bugs.eclipse.org/bugs/show_bug.cgi?id=111898 > > I'm not sure if this is the cause or if it's something else, but if > you intend to play with my exception handling patch, you'll want to > use the Sun compiler instead of Eclipse's. > > I hate wasting time chasing other peoples' bugs. Remind me to try > building things both in and out of Eclipse when I find seemingly > impossible bugs. > > - Charlie > ------------------------------------------------------- This SF.net email is sponsored by: Splunk Inc. Do you grep through log files for problems? Stop! Download the new AJAX search engine that makes searching your log files as easy as surfing the web. DOWNLOAD SPLUNK! http://ads.osdn.com/?ad_idv37&alloc_id865&op=click _______________________________________________ Jruby-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/jruby-devel
