On Wednesday, November 3, 2004, at 03:45 PM, Bob Wyman wrote:
However, I would much prefer if we used distinct mechanisms to indicate theIf there is a viable way to separate the two, I don't see a problem with it, but the method proposed is problematic:
time of the change and the significance of the change. Let's stop
overloading this date value as I proposed in:
http://www.imc.org/atom-syntax/mail-archive/msg11110.html
<modified>...</modified><importance>high</importance>
Or
<modified importance="high">...</modified>
The problem with both stems from the probability of a polling aggregator missing an important modification if an entry is first modified "importantly" and then trivially between polls. A small change would fix that problem:
<modified>...</modified><important-revision-id>2</important-revision> Or <modified important-revision-id="2">...</modified>
Any value that hasn't been used yet for that entry could appear in the place of "2". Serial numbers would be the most obvious method, but as long as the value is unique for that entry, it could be compared to a cached value to determine whether the publisher had made a change they wanted to flag since the last time the entry was seen.
