Jeroen,
>I guess someone in the designtime of VFP thought it was a great idea to have millisecond resolution in the datetime-vars. However they did not bother to adapt the conversion routines to work with the milliseconds.
This datetime standard was in place well before VFP implemented datetimes. I'd have to say the thought was to "shield" us from the milliseconds portion so that us database developers could just think in terms of integral seconds between two datetimes. We already had seconds() to give us sub-second time spans.
The problem is there are a couple of holes in the shield. Once you know they are there it's not difficult to adapt code accordingly.