[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: RE: [wsrp-interfaces] copyPortlet - additional use case
|
It was my understanding that when a clone
is performed in the 1.0 spec, there is no relationship between the original and
the clone. Though, reading the spec, it seems that this is left up to the
implementation: “No relationship between the
supplied Portlet and the new Portlet is defined by this specification.” Though,
any relationship seems a little awkward, since it leads to behavior which the
consumer would not expect. I
definitely agree that there’s a use case for breaking relationships
between clones and copies. In fact, I would argue that clone should
explicitly not lead to any relationship. Is making this requirement more
stringent in the 2.0 spec an option? Unfortunately, it’s likely to
impact current implementations. If we
can’t change this requirement, than I would agree with Andre that this is
a valid argument for having a separate method. However, I’m not
sure that it’s strong an argument, since I would suspect most producers
do not create this relationship and, therefore, don’t have this problem. Scott
From: Andre
Kramer [mailto:andre.kramer@eu.citrix.com] Say the portlet (A) in
the page designer use case to be cloned/copied has properties and values: x=1
and y=2. The new copy/clone (B) would initially have the same property values
for simple non-computed properties, but what about after: A.x = 3; B.y = 4; What is the expected
resultant value of A.y? So far, we have made no assumptions here (the write to
A.x could also modify A.y if y was computed from x), but I would expect a
copy/clone across registrations to help preserve A.y == 2. If we named the
operation "copy" then I would be further encouraged to expect no post
copy relationship between A and B, even if the two registration handles happen
to reference the same registration, and this would, I think, be expected by
Rich's use case, but not our necessarily our 1.0 clonePorlet. Do we agree that there
are such use cases where breaking any relationships between a portlet and its
copies/clones is a requirement? If so, then a new copyPortlet operation makes
sense (rather than overloading clonePortlet). Computed properties can still
take on any value they like, but we would be encouraging producer/portlet
property relationships to respect registrations. Regards, Andre From:
Goldstein, Scott [mailto:Scott.Goldstein@vignette.com] If I understand
correctly, the only difference between clone and copy is the relevance of the
registration. Therefore, I would say that this use case is applicable for
both clone and copy. The clone() method, I believe, currently supports
this use case due to the requirement of copying the current state from the
original portlet to the clone being created. Is this not true for copy()
as well? Scott From: Andre Kramer
[mailto:andre.kramer@eu.citrix.com] Could be
applicable, I agree. In selecting the use case, focused on requiring two
registrations and involving an end user. Regards, Andre From: Rich
Thompson [mailto:richt2@us.ibm.com]
|
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]