RE: TimeFilter edits

At 02:21 PM 1/21/2005, David B Harrington wrote:
> 
>
>> If a column is deleted but not the whole row, then that will
>> cause the lastChangeTime for the row to be updated.  There is
>> nothing special about TimeFilter and deleted rows.  Agents 
>> never download deleted rows, so I don't see why the text
>> is confusing.  
>
>Well the current text apparently **is** confusing. 
>The issue has been raised in the IEEE, and as I try to interpret the
>text for them, I understand why they find it confusing. 
>Your spec says that a deleted row is considered a changed row, and
>changed rows can be downloaded.

I disagree with this interpretation of the text.
This is a theoretical issue.  You cannot download
rows that don't exist -- TimeFilter or not.  Users
should not think TimeFilter causes rows to be preserved
after they are deleted.  

Andy


>> 
>> NEW (v2):
>> 
>>         "To be used for the index to a table.  Allows an application
>>         to download only those rows changed since a particular time.
>>         A row is considered changed if the value of any object in
>the
>>         row changes or if the row (or any object in the row) 
>> is created
>>         or deleted."
>> 
>
>Let's walk through your proposed text backwards to see the syllogism.
>Premise 1:
>        "if the row ... is created or deleted"
>        then the "row is considered changed"
>
>Premise 2:
>        if "rows changed since a particular time"
>        then application allowed "to download ... those rows"
>
>Therefore 
>        if "row is deleted"
>        application is allowed to download the row
>
>This is incorrect; as you say above "Agents never download deleted
>rows"
>But your text says the opposite! 
>
>dbh