Add parallel processing support about heemod HOT 6 CLOSED

So far this part relies on dplyr::do(). There is some hope to see parallel processing supported by this function in the future, but maybe not soon.

from heemod.

Some experimental work on this by Romain François : https://github.com/romainfrancois/parallelGroupBy.

Not production-ready, and not worked on for one year...

from heemod.

A parallel back-end to dplyr https://github.com/hadley/multidplyr. Needs to be installed from Github though, so not usable in production so far. Hopefully soon on CRAN?

from heemod.

Other solution: use a different approach with something like mclapply. While it is probably possible to refactor the code to play nice l with this, I'd rather stay with the clean&simple dplyr interface.

from heemod.

To run multidplyr, eval_newdata() needs to be loaded on every node. To do that it should be defined as a standalone function (with a more explicit name such as eval_newdataline()) and not defined at runtime in eval_model_newdata().

from heemod.

Closed by 854dcda.

from heemod.