Sunday, May 22, 2005, 2:08:57 AM, Tim Bray wrote:
>>>> change from a unicode combined char to single + combining >>>> diacritic, >>> >>> No. >>> >>>> change in paragraph 27 of an article that doesn't show up in a >>>> summaries-only feed, >>> >>> No. >> >> Dave: In my case, and seemingly in the case of most of the tools >> surveyed, both of those would get a "yes". > Existence proof, if any were needed, that the notion that there is > such a thing as an "objective change" is just silly. Anybody who has > worked with the deployment of more than two professional publishing > systems knows full well that there is very little agreement on what > constitutes a change. Attempting to specify a protocol element to > capture a mythological quantity strikes me as bad design. -Tim Did you get my response to Rogers' mail? I said that atom:modified "does not need" to be updated in these cases. However, if a publishing client has made the effort to post the changes to a server, and a server has made an effort to commit these changes to the database, then as with the case of a publisher committing a null change, it is not expected that the implementation makes a judgement call over whether there was really an update. atom:modified is much simpler than that. PaceDateModified2 deliberately doesn't prohibit this, nor does this prevent the proposal from fulfilling its goal to provide a temporal ordering for entry instances. -- Dave
