What about something like an internalRemove() that does everything needed when the component is removed, but NOT versioning stuff. And a remove() method, that does call internalRemove() and do the versioning stuff.
Then the internalRemove() of MarkupContainer could call each internalRemove() of each child component ... something like that ... (don't know if it is a good point or not, just an idea out of my brain). OR, in the remove method of the MarkupContainer, take care of browsing all the children and calling setParent(null) on them : a compromise between calling remove() on all children, and just getting rid of the children reference as it is done currently (I know that wicket is "unmanaged", but I think that a container component should be responsible of the lifecycle/management of its children components. On 12/28/05, Igor Vaynberg <[EMAIL PROTECTED]> wrote: > so what you are saying is that we should not optimize removeall() and just > make it loop over components and call container.remove() on them? > > this would be much simpler then the current implementation that performs an > optimization that saves a lot of what would be wasted session space. calling > removeall() on a container with 20 children using the simple implementation > would cause 20 change objects created and added to the version history, > while using the current implementation only a single change object is > created. the net effect is the same. > > so this is a tradeoff between user's assumption of how the code works and > efficiency ( how it actually works ). > > if you feel strongly about this then start a vote to change the removeall() > impl to cascade remove() on each component. > > -Igor > > > > On 12/28/05, Gili <[EMAIL PROTECTED]> wrote: > > > > I (partially) disagree. Invoking removeAll() should be equivilent > to > > invoking remove() on all children individually -- assuming no > > subclassing has occured. removeAll() might be more efficient, but if it > > doesn't get the job done what's the point? If subclassing has entered > > into the discussion then both remove() and removeAll() will have to be > > subclassed as a team (similar to hashCode() and equals()). > > > > Gili > > > > Igor Vaynberg wrote: > > > i do not think removeAll() should call the regular remove(). removeAll() > > > generates a single change object and clears out the children, i think > > > that is enough. it does what it says in a simple and efficient way. > > > > > > -Igor > > > > ------------------------------------------------------- This SF.net email is sponsored by: Splunk Inc. Do you grep through log files for problems? Stop! Download the new AJAX search engine that makes searching your log files as easy as surfing the web. DOWNLOAD SPLUNK! http://ads.osdn.com/?ad_idv37&alloc_id865&op=click _______________________________________________ Wicket-user mailing list Wicket-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wicket-user
