RE: Reader 'updated' semantics
Thanks for your thoughtful analysis on the update issue. I just want to follow up with reader implementers the importance of supporting the Atom update mechanism. News is one large application of a feed, but there are potentially many others in which the update mechanism is a requirement. For feeds, with respect to email, this is clearly one differentiating factor in which delivery is possible without creating a new message/entry. At this point, most Atom readers do not support updated, so the feed generator is stuck with new = new id. -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of James Holderness Sent: Wednesday, January 11, 2006 11:02 AM To: atom-syntax@imc.org Subject: Re: Reader 'updated' semantics Stephane Bortzmeyer: OP. In Atom, it seems to me that 2) is the only reasonable choice (1 or 3 would require to store the content - or at least a hash - and, if applied blindly, would create many false positives since a simple reformatting of the XML would trigger a change). Read my response again. Option 3 can't possibly trigger any more false positives than 2 since it requires option 2's conditions to be satisfied (i.e. the updated element changed) in addition to the content having been changed. Regards James
Re: Reader 'updated' semantics
On Tue, Jan 10, 2006 at 04:58:12PM -, James Holderness [EMAIL PROTECTED] wrote a message of 31 lines which said: There are a couple of options for an aggregator author. They can mark an entry as having changed when 1) the content of the entry has changed; 2) the updated element has changed; 3) the updated element has changed as well as the content having been changed; or 4) do nothing at all. I'm sure there are other possibilities, but I think those are the main issues. This is, IMHO, a completely different issue from the one asked by the OP. In Atom, it seems to me that 2) is the only reasonable choice (1 or 3 would require to store the content - or at least a hash - and, if applied blindly, would create many false positives since a simple reformatting of the XML would trigger a change). The question asked by the OP was: when updated was changed, what the reader should do? (And my reply was Interesting question but a bit out of scope for the atom-syntax group.)
Re: Reader 'updated' semantics
On Tue, Jan 10, 2006 at 10:35:27AM -0800, Tim Bray [EMAIL PROTECTED] wrote a message of 14 lines which said: There's a word for behavior of RssBandit and Sage: WRONG. Read again my message. Sage does *not* ignore changes of updated. Its behaviour is perfectly right. It just displays updated entries in a more discreet way than new entries. Software that chooses to hide this fact from users is broken Sage does not hide it. Test before criticizing.
Re: Reader 'updated' semantics
On Tue, Jan 10, 2006 at 05:32:08PM +0100, A. Pagaltzis [EMAIL PROTECTED] wrote a message of 37 lines which said: I don???t think it???s controversy, so much as that most people apparently simply don???t care whether an entry they???ve already seen has changed. It is also may be because feed writers are too eager to set updated when the change is insignificant (for instance, when there is simply a reformatting). In Atom, the specification is, IMHO, very clear: RFC 4287, 4.2.15. The atom:updated Element The atom:updated element is a Date construct indicating the most recent instant in time when an entry or feed was modified in a way the publisher considers significant. Therefore, not all ^ modifications necessarily result in a changed atom:updated value. But it is much more muddy in the RSS world which may explain why feed reader programmers tend to ignore updated. Developing best practices in this area is complicated because there is a strong interaction between the feed writers and the reader software. PS: Wikipedia allow authors to check a box Minor update when they modify a page and they don't want it to be regarded as a significant change. I wonder how many authors set it. It would be a nice Usability study to examine that.
Re: Reader 'updated' semantics
Tim Bray wrote: On Jan 10, 2006, at 9:07 AM, James M Snell wrote: In RSS there is definite confusion on what constitutes an update. In Atom it is very clear. If updated changes, the item has been updated. No controversy at all. Indeed. There's a word for behavior of RssBandit and Sage: WRONG. Atom provides a 100% unambiguous way to say this is the same entry, but it's been changed and the publisher thinks the change is significant. Software that chooses to hide this fact from users is broken - arguably dangerous to those who depend on receiving timely and accurate information - and should not be used. -Tim I agree with this, up to the point that flipping the updated bit becomes a spam vector. cheers Bill
Re: Reader 'updated' semantics
Stephane Bortzmeyer: OP. In Atom, it seems to me that 2) is the only reasonable choice (1 or 3 would require to store the content - or at least a hash - and, if applied blindly, would create many false positives since a simple reformatting of the XML would trigger a change). Read my response again. Option 3 can't possibly trigger any more false positives than 2 since it requires option 2's conditions to be satisfied (i.e. the updated element changed) in addition to the content having been changed. Regards James
RE: Reader 'updated' semantics
According to an exchange on the topic at RSSBandit's support site between Dare and myself, he writes: RSS Bandit does not provide an indication of an updated entry since many differ on what constitutes an update and what level of feedback should be provided to the user. I'm led to believe there is some controversy here. James' description is what I would expect, and seems straight-forward enough. Is there any thing more to this? Thanks -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Tim Bray Sent: Monday, January 09, 2006 9:31 PM To: James M Snell Cc: James Yenne; atom-syntax@imc.org Subject: Re: Reader 'updated' semantics On Jan 9, 2006, at 5:08 PM, James M Snell wrote: The updated element is used to indicate when a significant update has occurred to the entry. If you are updating the updated element when you update your entry, you are doing the right thing. If RSSBandit and FeedDemon are not picking up the fact that the entry has been modified based on a changed updated value (even if the updated element is the only thing that is different), then I would say those readers have a bug. You should never be changing your id element value. James is exactly right. The whole reason 'updated' exists is *exactly* so you can tell the downstream software unambiguously when some entry has changed and this ought to be brought to the consumers' attention. -Tim
Re: Reader 'updated' semantics
On Tue, Jan 10, 2006 at 07:06:59AM -0800, James Yenne [EMAIL PROTECTED] wrote a message of 39 lines which said: RSS Bandit does not provide an indication of an updated entry since many differ on what constitutes an update and what level of feedback should be provided to the user. Sage displays entries sorted according to updated but does not highlight updated entries, only new ones (new == new id). I believe it is partly an User Interface and Usability issue and therefore out of scope for the Atom-syntax group.
RE: Reader 'updated' semantics
I haven't used Sage, but I would anticipate that entries which have been retrieved initially by a reader are continually checked for changes to the updated element until the entries age to some drop off point, regardless of the UI display conventions of them. The problem is that there is data loss in the reader that doesn't update entries that have actually changed, again, regardless of how you display these changes. Perhaps semantics is indeed outside of the scope of a syntax discussion list, however, the updated element syntax implies some semantic understanding. I am not a reader implementor, so I don't have any insight on why readers do not pick up these changed updated elements as a cue for entry refresh. Otherwise, please direct me to a group that discusses these issues. Thanks -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Stephane Bortzmeyer Sent: Tuesday, January 10, 2006 7:35 AM To: James Yenne Cc: atom-syntax@imc.org Subject: Re: Reader 'updated' semantics On Tue, Jan 10, 2006 at 07:06:59AM -0800, James Yenne [EMAIL PROTECTED] wrote a message of 39 lines which said: RSS Bandit does not provide an indication of an updated entry since many differ on what constitutes an update and what level of feedback should be provided to the user. Sage displays entries sorted according to updated but does not highlight updated entries, only new ones (new == new id). I believe it is partly an User Interface and Usability issue and therefore out of scope for the Atom-syntax group.
Re: Reader 'updated' semantics
James Yenne wrote: I'm led to believe there is some controversy here. James' description is what I would expect, and seems straight-forward enough. Is there any thing more to this? There are a couple of options for an aggregator author. They can mark an entry as having changed when 1) the content of the entry has changed; 2) the updated element has changed; 3) the updated element has changed as well as the content having been changed; or 4) do nothing at all. I'm sure there are other possibilities, but I think those are the main issues. With option 1 the poor user is flooding with notifications for all sorts of insignificant updates like single character spelling corrections. With option 2 the user is even worse off because they get notified of changes to an entry when in fact nothing at all in the content has changed (this is not uncommon). Option 3 is probably the least annoying, but can still end up flagging insignificant updates and can still miss significant updates if the publisher forgets to set the updated element. Option 4, well, it's easy if nothing else. FWIW Snarfer does something based on option 3. Only changes to the actual message content are considered important enough to notify the user of an update. The title, author and other metadata aren't considered significant. Changes to the HTML markup in the content aren't considered significant either. Not perfect perhaps, but I think it strikes a fair balance between being useful and being annoying. Regards James
RE: Reader 'updated' semantics
As an example of the problem, if I'm using a reader as the sole means to get my news, I would have missed the update to the coal miners story, from miracle to tragedy, which was a very significant update. -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of James Yenne Sent: Tuesday, January 10, 2006 8:13 AM To: 'Stephane Bortzmeyer' Cc: atom-syntax@imc.org Subject: RE: Reader 'updated' semantics I haven't used Sage, but I would anticipate that entries which have been retrieved initially by a reader are continually checked for changes to the updated element until the entries age to some drop off point, regardless of the UI display conventions of them. The problem is that there is data loss in the reader that doesn't update entries that have actually changed, again, regardless of how you display these changes. Perhaps semantics is indeed outside of the scope of a syntax discussion list, however, the updated element syntax implies some semantic understanding. I am not a reader implementor, so I don't have any insight on why readers do not pick up these changed updated elements as a cue for entry refresh. Otherwise, please direct me to a group that discusses these issues. Thanks -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Stephane Bortzmeyer Sent: Tuesday, January 10, 2006 7:35 AM To: James Yenne Cc: atom-syntax@imc.org Subject: Re: Reader 'updated' semantics On Tue, Jan 10, 2006 at 07:06:59AM -0800, James Yenne [EMAIL PROTECTED] wrote a message of 39 lines which said: RSS Bandit does not provide an indication of an updated entry since many differ on what constitutes an update and what level of feedback should be provided to the user. Sage displays entries sorted according to updated but does not highlight updated entries, only new ones (new == new id). I believe it is partly an User Interface and Usability issue and therefore out of scope for the Atom-syntax group.
Re: Reader 'updated' semantics
In RSS there is definite confusion on what constitutes an update. In Atom it is very clear. If updated changes, the item has been updated. No controversy at all. James Yenne wrote: According to an exchange on the topic at RSSBandit's support site between Dare and myself, he writes: RSS Bandit does not provide an indication of an updated entry since many differ on what constitutes an update and what level of feedback should be provided to the user. I'm led to believe there is some controversy here. James' description is what I would expect, and seems straight-forward enough. Is there any thing more to this? Thanks -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Tim Bray Sent: Monday, January 09, 2006 9:31 PM To: James M Snell Cc: James Yenne; atom-syntax@imc.org Subject: Re: Reader 'updated' semantics On Jan 9, 2006, at 5:08 PM, James M Snell wrote: The updated element is used to indicate when a significant update has occurred to the entry. If you are updating the updated element when you update your entry, you are doing the right thing. If RSSBandit and FeedDemon are not picking up the fact that the entry has been modified based on a changed updated value (even if the updated element is the only thing that is different), then I would say those readers have a bug. You should never be changing your id element value. James is exactly right. The whole reason 'updated' exists is *exactly* so you can tell the downstream software unambiguously when some entry has changed and this ought to be brought to the consumers' attention. -Tim
Re: Reader 'updated' semantics
On Jan 10, 2006, at 9:07 AM, James M Snell wrote: In RSS there is definite confusion on what constitutes an update. In Atom it is very clear. If updated changes, the item has been updated. No controversy at all. Indeed. There's a word for behavior of RssBandit and Sage: WRONG. Atom provides a 100% unambiguous way to say this is the same entry, but it's been changed and the publisher thinks the change is significant. Software that chooses to hide this fact from users is broken - arguably dangerous to those who depend on receiving timely and accurate information - and should not be used. -Tim
Reader 'updated' semantics
I'm looking for info about what the expected 'updated' datetime element semantics look like, andhaven't found anything in old email and the wiki on this topic. Is there a description somewhere? RSSBandit and FeedDemon don't pick up entry changes based on my changing the feed and entry 'updated' element to a later datetime. Instead I change the id element and get a 'new' entry, which is not desirable, since id for an entry should be a permanent uuid. I don't know how updates are supposed to occur. Thanks for any help,James
Re: Reader 'updated' semantics
On Jan 9, 2006, at 5:08 PM, James M Snell wrote: The updated element is used to indicate when a significant update has occurred to the entry. If you are updating the updated element when you update your entry, you are doing the right thing. If RSSBandit and FeedDemon are not picking up the fact that the entry has been modified based on a changed updated value (even if the updated element is the only thing that is different), then I would say those readers have a bug. You should never be changing your id element value. James is exactly right. The whole reason 'updated' exists is *exactly* so you can tell the downstream software unambiguously when some entry has changed and this ought to be brought to the consumers' attention. -Tim