[Laszlo-dev] design question, what to do if someone wants to instantiate a new dataset with same name as the old one?

P T Withington ptw at pobox.com
Tue Nov 20 09:32:52 PST 2007


I would vote for not supporting that, and simply saying that it is  
'undefined' what happens.

On 2007-11-20, at 11:14 EST, Henry Minsky wrote:

> This is a question regarding an issue raised by this bug which Andre
> reported
>
> http://jira.openlaszlo.org/jira/browse/LPP-4688
>
> In his test case, he has code which creates a (globally)  named  
> dataset ,
> creates a view which uses it in a datapath,
> then destroys the datapath and creates a new one with the same name
>
>        <method name="createDataset" args="name, local, text" ><! 
> [CDATA[
>            var ds = new LzDataset(local ? this : canvas,  
> {"name":name});
>            ds.appendChild(new LzDataElement("test", {}, [new
> LzDataText(text)]));
>            return ds;
>        ]]></method>
>
>        <method name="test_1" ><![CDATA[
>            var ds = this.createDataset("dsdyn1", false, "foo");
>            var text = new lz.text(this, {id:"leak1",
> datapath:"dsdyn1:/test/text()", visible:false});
>            ds.destroy();
>            ds = this.createDataset("dsdyn1", false, "bar");
>
> There's two issues, one is how to make sure that when you destroy a  
> dataset,
> all the views which are bound to it
> release their pointers.
>
> The other issue is should there be some mechanism so that if a new  
> dataset
> is created with the same name, that
> views which reference it in their datapath should rebind to the new  
> dataset
> automatically? That seems like a hard
> problem to me, and aside from the cost and complexity to maintain  
> it. I
> don't know how often people
> destroy and create datasets, but I always considered them as  
> permanent, in
> that you can
> reload them or clear out their contents, but destroying and  
> recreating one
> with the same
> name is something I had not considered.
>
>
>
>
> -- 
> Henry Minsky
> Software Architect
> hminsky at laszlosystems.com



More information about the Laszlo-dev mailing list