I'm developing a package that contains a function that requires evaluation of an expensive Table, so I use ParallelTable instead. It works correctly, but always gives an error message, repeatedly, for each kernel, like this:
Table::iterb: Iterator {mypackage`Private`m,1,mypackage`Private`nsrc$663} does not have appropriate bounds
How can I avoid this error? Do I need to distribute variables explicitly?
The iterator variable in the package lives in the context that your package sets (in this case, mypackage`Private`), but the call to ParallelTable from inside your package does not distribute the definitions in your package, because the parallel functions only distribute contexts that are listed in $DistributedContexts. This is by default set to $Context, which will most probably be Global` if you're running the package in a notebook. So the definitions in your package are unknown to the worker kernels, which is the cause of the error.
Instead, change the function call to
ParallelTable[..., {m, nsrc}, DistributedContexts -> {"mypackage`Private`"}] Distribute contexts that other symbols might depend on (you don't need to distribute System` and the current context).
