Thomas Broyer wrote:
2006/6/15, James M Snell <[EMAIL PROTECTED]>:
Absolutely +1 on not allowing the client to change the atom:id.
For the record, absolutely +1 also.
But the question was also: how about other fields?
I think there could be a real problem wrt authors and contributors: in
some constrained environments, server implementations might want to
disallow changing the author and/or contributor and managing these
informations itself (i.e. the author is the user who submitted the
doc, he can add other authors and turn contributors into authors and
authors into contributors, but cannot remote himself from the list nor
turn himself from an author to a contributor; other users might be
able to update the content, they're not allowed to change authors or
contributors, the system automatically adds them as contributors when
they PUT an updated version if they're not already listed; the list of
possible authors and contributors is constrained by the system based
on a configuration table).
On the other hand, on other environments, server implementations might
let anyone change authors and contributors without any constraint.
Similar problems could also happen wrt categories, particularly if
they're not used to "tag" the entry but rather to "specialize" its
type (à la GData).
I think for us, these issues are the moral equivalent of the ones the
syntax wg went through with dates and publish states. Assuming we could
ever get agreement, they won't be sanely standardized by us in any
reasonable amount of time because they increase the scope of the work.
I think we should allow people to innovate in these areas. I also think
that we should not have situations where people cannot migrate their
data across services, which is why I would to see WG buyin, now, that
servers should take on markup they can't necessarily process, but could
reasonably store. That's because while the new server may have no use
for the data, the client might well need it and have built assumptions
around it, and with any other set of constraints they won't be able to
migrate without data loss. IMO it's quite serious we do our level best
to not have a situation where people get their data locked into services
for this reason. Think of extensions as junk DNA, a feature of being
able to evolve.
cheers
Bill
