I believe someone recently checked in a fix that suppresses the trailing zeros, since their presence was reported as an error. It sounds like the gold file should have been updated at that time but wasn't. (If the zeros *should* be present, we need to undo that fix... but as far as I can tell the bug report is correct and the patch was a desirable one.)
General reminder to committers: Always run the regression tests before you check in changes -- and if you've updated the tests, check those in too. ______________________________________ Joe Kesselman / IBM Research
