timestamps (was: TIP, how to fight with corrupted indexing)

[Robert Brenstein]

>  > There
>  > is clearly a need for a mechanism that ensures that indexes are in
>>  sync with data.
>
>Yes, such mechanism will be so called TIMESTAMP -- internal time of
>database. Each time is changed at least one field, value of tine is
>incremented.
>
>In 2.0 TIMESTAMP will present in each table in any case, for each record.
>Because this is needed for our transaction model.
>

You mean that v-kernel will keep checking the timestamps and 
automatically rebuild index when it is out of sync (besides using the 
timestamps for transactions)? So each record will have a timestamp 
and then the table will have its own, right? And table's timestamp 
will reflect newest timestamp of its records and any 
structural/property changes to table, right?

Robert