On Wed Nov 05 19:31:36 2008, chrisdolan wrote: > It took a while to understand the code, but the solution was a > one-liner. Patch attached. Unfortunately, the patch produced a bunch of spectest failures. However, it wasn't because the patch was wrong, but rather a bug in some code generation in PCT. I fixed that, and then the patch worked and did the Right Thing, so it's applied in r32546. Plus got us spectesting this.
Thanks! Jonathan