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

[Asbjørn Ulsberg <asbjorn@xxxxxxxxxxxxxx>]

On Tue, 13 Jul 2004 16:22:21 -0600, Antone Roundy <antone@xxxxxxxxxxxxxx>  
wrote:

> This, and the fact that people ARE going to republish heavily modified  
> entries without creating a new one are my big reasons for not favoring  
> any kind of "replaces" system.

Sam's problematic use case is just as problematic now as with any  
«supersede» system. An entry might be lost in void no matter what we do.  
The only solution to this is to be able to fetch the entry on a public  
repository, like Google perhaps. USENET's MessageID's already work like  
this. Perhaps atom:id could too.

> I'd prefer a modified (big change)/updated (small change) combo

What about keeping the semantics of 'modified' as is, and add a simple  
solution to the 're-issued' cases, if people are having difficulties with  
accepting <relation>; Allow for several <issued> elements. I would  
definately prefer a <relation>[1]-mechanism over mulitple <issued>  
elements, but it's a good compromise, and <relation> might still be  
supported on top of that <issued> semantic.

What the different <issued> elements would mean, is however not very  
clear. I'm not sure we can do or say anything but «If you re-issue an  
entry, you SHOULD (or MUST) append a new atom:issued element to atom:entry  
that reflects the new issue date. The original atom:issied should be left  
intact». The order of the <issued> element is of no significance; they  
could easilly be ordered chronologically with a string sort (if they are  
in the same timezone).

So, simple solution: multiple <issued> elements. A bit more complicated  
solution: <relation>. The bonus with adding <relation> is, though, that we  
get «for free» all the other attributes of the element, which obsoletes  
current @rel="alternate" with much more explicit attributes; 'IsFormatOf'  
(when an entry is an alternative representation for an HTML article),  
'References' (threading) and 'IsBasedOn' (translations of an entry).

____
[1] <url: http://dublincore.org/documents/relation-element/>

--
Asbjørn Ulsberg         -=|=-        asbjornu@xxxxxxxxxxx
«He's a loathsome offensive brute, yet I can't look away»