Module vs. Namespace

[L33T]

Currently, the Python project defines modules based on the core sub-projects (arch, common, compiler, symex) which requires specific module imports (or wildcard) to obtain classes. (i.e. from vtil.arch import basic_block)

This does not follow the core project where "imports" are usually inside the main namespace (VTIL). For example, basic_block is accessible from the main namespace (vtil::basic_block) unlike the python project which puts the class inside vtil::arch::basic_block virtually.

This is arguably good for organizing the massive projects into submodules, but can result in a confusing import perspective. Take the following for example:

from vtil.arch import routine, basic_block
from vtil.common import tagged_order

def on_block(bbl: basic_block) -> tagged_order:
  print(bbl)
  return tagged_order.obreak

rtn = routine.load('./hello_world.vtil')
rtn.for_each(on_block)

tagged_order is actually vtil::enumerator::tagged_order which is contained inside the common project and requires to be explicitly imported, instead of being inside the global vtil namespace.

The following is considered:

from vtil import routine, basic_block, tagged_order

def on_block(bbl: basic_block) -> tagged_order:
  print(bbl)
  return tagged_order.obreak

rtn = routine.load('./hello_world.vtil')
rtn.for_each(on_block)

And the submodules would be replaced with explicitly nested namespaces inside the vtil namespace. (i.e. debugger, symbolic)