Alex Karasulu a écrit :
On Sat, Dec 12, 2009 at 12:03 PM, Emmanuel Lecharny <[email protected]>wrote:
Another Q :
what should we do if we create an OC containing an AT (in MAY or MUST,
that's irrelevant) which is already present in one of it's parents ? Should
it be considered as an error ?
Yes I think we should flag an error just to make sure the user understands
this AT is present and available thanks to an ancestor OC. If anything it
shows the user to be more aware while dealing with schema design.
Ok.
In general if there is any anomaly I would take the policy of rejection even
if it is recoverable or not an issue. If there is even the slightest smell
of something not being right with the schema or the awareness of the schema
designer we should take the stance of caution.
BTW it would be nice to be able to tell the user, 'hey guy this AT is
already in the MAY/MUST list thanks to ancestor foo'.
It's difficult to have this level of precision : right now, the message
tells the user that the AT is already declared somewhere else (MUST or
MAY, or one of the parent's MUST/MAY).
