When n.cores is NULL, gbmCluster() detects multiple core automatically and
creates a cluster when appropriate. The code introduced in commit
9519046fe3bb49ad44dc to fix an issue with cluster detection checks n.cores
directly instead of checking the output of gbmCluster(). So what was
automatically parallelised before isn't anymore.
The current code checks the output of gbmCluster and still uses the
non-parallel lapply() on machines with only one core.
When n.cores is NULL, gbmCluster() detects multiple core automatically and creates a cluster when appropriate. The code introduced in commit 9519046fe3bb49ad44dc to fix an issue with cluster detection checks n.cores directly instead of checking the output of gbmCluster(). So what was automatically parallelised before isn't anymore.
The current code checks the output of gbmCluster and still uses the
non-parallel lapply() on machines with only one core.