: looking at the current code for DateField.toObject(Fieldable) it seems : inheriently broken, attempting to parse a string right after concating 'Z' : on the end even though the parser expects the Z to already be gone -- i'm : not sure how this could path could *ever* work, regardless of the input.
Duh, i forgot a key aspect of DateFormat.parse(String): "The method may not use the entire text of the given string." ... which is why the code can work sometimes, the parser will happily ignore the 'Z' ... the exception is about the fact that there are no milliseconds (which are suppose to be optional) so this really is the exact same bug as SOLR-470 ... the problem is just more significant then i thought -- I assumed the only way to trigger this was a DateMath expression that didn't have milliseconds, but DateField.toObject(Fieldable) has the same bug so direct usage of it (by code like DocumentBuilder.loadStoredFields) will also cause this problem if the orriginal date didn't have any milliseconds. -Hoss