Just to place my statements in the recent synchronization thread in
the right place.
And a restatement of the fact that I have changed my mind.
On 5 Nov 2005, at 01:01, Thomas Broyer wrote:
<http://www.intertwingly.net/wiki/pie/
PaceOrderCollectionsByAppModified>
== Abstract ==
+1
To summarize my argument.
If we don't add app:modified timestamps, the cheer force of the atom
protocol will make atom:updated mean atom:modified. For a long time I
thought this was not a bad thing, and so I argued that there was no
need for app:modified, since atom:updated really is that thing.
But clearly a lot of people want to be able to have atom:modified
change only when a significant change occurs. Ie. They want to be
able to be able to signal the difference between a major change ("re-
read my post") and a minor editorial change "I just fixed a typo,
there is nothing else to see here".
So if we do not add app:modified there will be no way to distinguish
these two use cases. Publishing servers will be stuck with the
following dilemma: if they allow the user to determine :updated, it
will be very difficult for a blog editor (such as BlogEd) to do
correct synchronization, as demonstrated so well by Thomas Broyer in
the "synchronization" thread. On the other hand if therefore the
publishing system takes atom:updated in hand and sets it
automatically for any change, readers of the feeds will feel they are
being spammed.
app:modified solves the problem. It is low cost, and very much in
demand.
Henry
