You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The idea is to have the future record the full recover stack in the FutureResult object. Something like the info in:
tb<- .traceback(NULL)
if (is.null(tb)) tb<- .traceback(2L)
utils::dump.frames(dumpto="frames", to.file=FALSE)
and then regenerate this call stack when relaying the error on the parent R process. I'm not sure if/not the latter can be done, but if we can figure it out, it'll be super cool.
It might be expensive to transfer all call stacks back from the worker to the main R session when running in parallel, so maybe this needs to be enabled via an option, e.g. options(future.recover.enable = TRUE). It might be that we can enable it by default for sequential futures.
Idea
The idea is to have the future record the full recover stack in the FutureResult object. Something like the info in:
and then regenerate this call stack when relaying the error on the parent R process. I'm not sure if/not the latter can be done, but if we can figure it out, it'll be super cool.
It might be expensive to transfer all call stacks back from the worker to the main R session when running in parallel, so maybe this needs to be enabled via an option, e.g.
options(future.recover.enable = TRUE)
. It might be that we can enable it by default forsequential
futures.See also
#253
The text was updated successfully, but these errors were encountered: