Eike Rathke:
> we currently define that MIN/MAX return 0 if there is no value at all,
> either no parameter or no value in a referenced range. I know that all
> apps except OOo see it different, but IMHO the minimum/maximum of no
> values should not be 0 but an error instead.
Since most implementation return 0, I don't think they should be REQUIRED to produce an error. I don't see the value to implementors in modifying the apps to detect this odd case, especially since 0 is such a widely-held view.
I think we should either (a) define as 0, as now, or (b) make this implementation-defined, stating that apps MUST produce EITHER 0 or Error in this case. I'd be fine with (b).
--- David A. Wheeler
