>Very good!
>
>Now the question is, is it my problem or Microsoft's? Ultimately it is mine, but is it a bug?

Replying would be poking my hand into bee nest :) Anyways I'll try to explain why as softly as I can :)

Datetime is held as 2 integers, one for date and other is for time of day (fractional). IOW time portion also keeps milliseconds but you didn't say what the milliseconds was !!! It might be any one from 0 to 999 from the seemingly same datetime values (you don't see the milliseconds part). However on comparison milliseconds part comes into play with a 999/1000 miss chance (actual resolution like 1/18 secs narrows the gap but in theory it's 1000 possible different values for same YYYYMMDDhhmmss).

If it's a bug it's not VFP's bug IMHO. It wasn't VFP's decision to keep datetime as 8 byte value (of course I'm not sure about this but have a great sense it was result of SQL datetime and C datetime structs - I don't know exactly those structs and might be wrong). Basing my idea that Foxpro was holding date as plain YYYYMMDD and didn't choose to do same for datetime. Currently I don't care the roots and carefull to write workaround code ( because it hit me before as did to you now :)
Cetin