GWT CellTable ListProvider Patterns with Lightstreamer

[flybyGWT]

Hi flybyGWT,

I confirm you that the Lightstreamer JavaScript Client library already handles the process of update at Item level.
Please refer to this piece of code of our GWT demo:

mySub.addListener({<br /> onItemUpdate: function(itemUpdate) {<br /> for (i = 0; i < fieldList.length; i++) {<br /> <br /> // check whether the schema value is changed. If it's not, skip it<br /> // unless a full data snapshot is being sent.<br /> if (!itemUpdate.isValueChanged(fieldList[i]) && !itemUpdate.isSnapshot()) {<br /> continue;<br /> }<br /> <br /> var itemName = null;<br /> itemName = itemUpdate.getItemName();<br /> <br /> var value = null;<br /> value = itemUpdate.getValue(fieldList[i]);<br /> <br /> if (value != null) {<br /> that.@gwtdemo.client.LSGWTSubscription::onItemUpda te(Ljava/lang/String;Ljava/lang/String;Ljava/lang/String;)(itemName, fieldList[i], value);<br /> }<br /> }<br /> }});<br />


The event with which you are receiving an update is unique regardless of how many fields have changed for that Item. The object ItemUpdate allows you to navigate between fields that have actually undergone a change of value. At that point, as you have already noticed, our demo calls for each of these fields a single event to update the GWT FlexTable.

To achieve your goal, you have to change this behavior, by updating your internal cache of the Item for every updated field and then passing to the CellTable the complete row with updated values.
Something like this:

mySub.addListener({<br /> onItemUpdate: function(itemUpdate) {<br /> for (i = 0; i < fieldList.length; i++) {<br /> <br /> // check whether the schema value is changed. If it's not, skip it<br /> // unless a full data snapshot is being sent.<br /> if (!itemUpdate.isValueChanged(fieldList[i]) && !itemUpdate.isSnapshot()) {<br /> continue;<br /> }<br /> <br /> var itemName = null;<br /> itemName = itemUpdate.getItemName();<br /> <br /> var value = null;<br /> value = itemUpdate.getValue(fieldList[i]);<br /> <br /> if (value != null) {<br /> that.@gwtdemo.client.LSGWTSubscription::updateMyIn ternalCahce(Ljava/lang/String;Ljava/lang/String;Ljava/lang/String;)(itemName, fieldList[i], value);<br /> }<br /> }<br /> <br /> that.@gwtdemo.client.LSGWTSubscription::onItemUpda te(Ljava/lang/String;)(itemName);<br /> }});

By the way, I confirm you that the Lightstreamer server does not use JSON to send updates to the client, but a specific and optimized protocol.

Thank you for the reply. In the example you provided you would then on the GWT implementation side define the onItemUpdate(Ljava/lang/String make a call to the LightStreamer client cache with mySub.getItems(itemName) and then pass off the data to the various DataProviders?


My thought was to use

mySub.addListener({ onItemUpdate : function(itemUpdate) {<br /> var itemName = null;<br /> itemName = itemUpdate.getItemName();<br /> <br /> var changeFields = [];<br /> var changeValues = [];<br /> var chngIndex = 0;<br /> <br /> for (i = 0; i < fieldList.length; i++) {<br /> // check whether the schema value is changed. If it's not, skip it<br /> // unless a full data snapshot is being sent.<br /> if (itemUpdate.isValueChanged(fieldList[i]) || itemUpdate.isSnapshot()) {<br /> var value = null;<br /> value = itemUpdate.getValue(fieldList[i]);<br /> <br /> if(null != value){ <br /> changedValues[chngIndex] = value;<br /> changedFields[chngIndex] = fieldList[i];<br /> chngIndex++;<br /> }<br /> <br /> }<br /> }<br /> <br /> if (changedValues.length > 0) {<br /> <br /> <br /> that.@gwtdemo.client.LSGWTSubscription::onItemUpda te(Ljava/lang/String;Lcom/google/gwt/core/client/JsArrayString;Lcom/google/gwt/core/client/JsArrayString;)(itemName, changeFields, changeValues);<br /> }<br /> <br /> }<br /> });

Then do field/value pair update -> GWT java bean updates though deferred binding field/value mapping package.


In the case of having a super data set and multiple subData sets (views of the super set). Would you suggest having a separate subscription for each data set including the sub and suppers, or have the super set be the only subscriber which in client pub/sub down to the subDatasets? Or is there a even more efficient way build into LS already?