Yesterday, we pushed an update of the service. We are still improving the pivottable and we focused this time our attention on sorting.
This one was tricky because :
* we had to deal with special cases with the time dimension. Indeed, the OLAP result we get contains only cells like: January / product A / Sales. Consequently, we don't have anymore the date information that would have allowed easy sorting. So we had to find a way to be sure that 'wednesday' comes before 'sunday' (or january before april...) in all results.
* We don't know in advance the depth of the sort. As you can cross join several attributes like product / sub category / category etc... you have to sort all the members and cells value accordingly on each of this attributes. In our example, first sort products then sub category and finaly category.
We are pretty happy with the current result. It enables users to reallly navigate in data intuitively and coherently. We can do much more regarding value sorting (for now, it is a bit messy when you have several cross joins) but, from a technical point of view, we built some utilities that enable us to really manipulate the OLAPResult in a very flexible and easy way.
This opens a lot of possibilities and we will take advandage of that in this week work:
* chart overlay
=> first level is display directly charts on top of the pivot table to analyse visually the result. Then add a feature to send the result directly to an existing dashboard tab.
=>second level is doing real pivot chart that is replace text cells rendering by charts one. We strongly believe, it is a really powerfull way of displaying data as it make comparison and trends much more obvious than to have only one chart with a lot series / points in it.
* value formating => for text rendering. Offer some formating like $ or 0, 1, 2 or 3 decimal points. We will try to make sensible defaults depending of the data displayed. => We will try to implement auto formating with colors level in the background as you can find in Excel 2007.