Re: a methodical approach to defining what date elements we need

[Graham <dtcd@xxxxxxx>]

On 12 Jul 2004, at 8:44 pm, Arve Bersvendsen wrote:

> Also: Keep in mind that any date can be wrong or falsified, 
> machine-generated or not.

Yes.

> I firmly believe that "issued" should be set in stone. To me, it 
> simply doesn't make sense that an entry issued three weeks ago 
> suddenly gets this date changed. The entry is three weeks old. Period.

This is a single entry that has been modified over time:

 http://www.tbray.org/ongoing/When/200x/2004/02/20/GenxStatus

What is its issued date? (NB Suggesting it be broken into multiple 
entries is not an option)

> 1. This is equivalent to republishing in a different feed than the 
> original. Nothing changes, including dates
> 2. This is a re-issue, with a new issued date, a new modification 
> date, a new ID. In effect: a "new entry" that supersedes the former 
> revision.

The ID should never change, otherwise the element is useless (we might 
as well just generate a hash of the content) and we should forget it 
entirely, and Atom for that matter.

Graham

Attachment: smime.p7s
Description: S/MIME cryptographic signature