Feature request - 'batch' axis label formatting


I’m using a rather small data set (less than 50 points) and wherever possible, I want users to experience 60 fps horizontal panning in the graphs.  Unfortunately, even on an iPhone 5S, i’m unable to do so.  Full disclosure though, i am panning multiple charts at the same time. 

The first simple switch I made was changing how I provided data to the charts SDK. That is instead of using this on-demand provisioning of data to the graph:

- (id<SChartData>)sChart:(ShinobiChart *)theChart dataPointAtIndex:(NSInteger)dataIndex forSeriesAtIndex:(NSInteger)seriesIndex

give all the data up front by using this:

- (NSArray *)sChart:(ShinobiChart*)theChart dataPointsForSeriesAtIndex:(NSInteger)seriesIndex

but I saw basically no change in the frame rate.    To keep a long story short, the way I was able to get it pegged at 60 on an iPhone 5S was to comment out all label formatters for X, Y axis.  I noticed that if even if I hard coded the results of my label formatters,

-(NSString *)stringForObjectValue:(id)obj onAxis:(SChartAxis *)axis

returning @“h” for all objects, the performance really no different than when my proper formatting code was in play. That is, it still dropped 5 fps. That surprised me - does that seem right to you guys?

I’m guessing you provided the ability to provide all data up front for those of us that are sensitive to pan/zoom performance (and have relatively small data sets).  If that’s true, could i also request that something similar be done for label formatting?  

Any other suggestions/tips on how to squeeze the most FPS out of the graphs? 



We appreciate the suggestions and it’s defintely something we can take a look at. The datasource changes you made will only really affect render time - as once the data is collected we don’t ask for it again (but the array method is much more efficient).

The less work the chart has to do to render (or manipulate) the drawn plot area the faster it will be. Therefore, keep styling to a minimum by not having area fills etc. 

There is always going to be an overhead when using deleagte methods that are called very often such as isPanning and therefore the efficiency of these calls is paramount.  

I’ll make sure your comments on the tick labels are noted - we are aware that the calcualtions of tickmarks is going to be an overhead. To this end you could try setting a fixed major tick mark frequency too.  We are aware that we can offer more options to those who would like a bare bones approach. 

Thanks for your post.