For example, I define a function f.

f[x_] := x^2;

And when I execute

ParallelTable[f[i], {i, 1, 10}]

I get

{1, 4, 9, 16, 25, 36, 49, 64, 81, 100}

as expected. (my function definition has propagated to the parallel kernels)

Now, if I do

Print = Null;


ParallelDo[Print[i], {i, 1, 10}]

I expect nothing to be printed out because I have changed the Print function. But instead, I get the printout.


Why did my function definition propagate to the parallel kernels in the first example, but not in the second example?

  • $\begingroup$ Seemingly minor but important point: Definitions are shared among subkernels. They are "distritbuted" (copied to) the subkernels directly from the main kernel. I changed the title to avoid confusion with SetSharedVariable/SetSharedFunction. $\endgroup$
    – Szabolcs
    Oct 12, 2016 at 7:13

1 Answer 1


Print didn't get distributed because it was in the System` context. By default only Global` (or rather $Context) symbols are distributed.

Definitions are sent to subkernels using DistributeDefinitions. Both this function and those that call is have the DistributedContexts options which usually defaults to $DistributedContexts, which in turn is set to

$DistributedContexts := $Context

You can set these to choose which contexts will get distributed.

Auto-distributing everything by default would lead to breakage soon due to many different reasons (loading a package often does more than issue definitions, some contexts hold mutable state such as caches, etc.)


That said, attempting to distribute (automatically or manually) any System` symbols is a patently bad idea that could very easily lead to breakage. While I haven't tried to find out if it is so, I wouldn't be surprised if the system protected itself against this even beyond the fact that these symbols are Protected on the subkernels as well, which alone would break distribution anyway.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.