I'm trying to update downstream libraries for v1.1.0rc1. There was a breaking change in the c_api.h to have an additional training bool parameter in this code change:

hcho3@f100b8d

* Add `training` parameter for prediction method.
* [Breaking]: Add `training` to C API.

I'm looking at the patch but cannot tell when training should be true or false in the XGBoosterPredict() call. Could somebody offer some guidance?

I read that

* \param training Whether the prediction value is used for training.

but that doesn't really tell me enough to understand what the consequences of training true (1) will be.

Thanks for you help.