ipdb: IPython-enabled pdb

Warnings

• breaking Critical IPython dependency versioning: `ipdb` relies heavily on `IPython`. Ensure correct `IPython` versions are installed for your Python version to avoid conflicts or missing features. For Python 3.6, install `7.16.3 <= IPython < 7.17.0`. For Python > 3.6, install `IPython >= 7.31.1`. Consult `ipdb`'s `HISTORY.txt` for detailed historical requirements if using older Python versions.
  Fix: Check `ipdb`'s `HISTORY.txt` or PyPI `Requires-Python` and `Requires-Dist` for exact `IPython` version constraints for your Python environment and `pip install 'ipython<X.Y'` or `'ipython>=X.Y'` accordingly.
• gotcha Do not leave `ipdb.set_trace()` calls in production code. These breakpoints will halt program execution and expose the interactive debugger, which is a security risk and an operational blocker.
  Fix: Remove `ipdb.set_trace()` before deploying code to production. Consider using environment variables or conditional logic to enable debuggers only in development environments.
• gotcha In Jupyter Notebooks on Windows, `ipdb` may occasionally fail to display the current line and context (e.g., with the 'l' command) due to inconsistent directory separator handling (mixed slashes).
  Fix: No official fix yet; a reported workaround involves a minor change to `debugger.py` to canonicalize filenames. Consider using `pdb` or external IDE debuggers for complex Windows/Jupyter setups if this issue persists.
• deprecated Older `ipdb` versions dropped support for various Python versions. `ipdb` version 0.6 dropped Python 2.4 support, version 0.8 dropped Python 2.5 support, and version 0.10.3 dropped Python 3.2 support. Current versions require Python >= 2.7, excluding 3.0-3.3.
  Fix: Upgrade to the latest `ipdb` version compatible with your Python interpreter, or ensure your Python version meets `ipdb`'s `requires_python` specification.
• gotcha When using `ipdb` in environments that manipulate standard output (e.g., certain testing frameworks like older versions of `nose`), `ipdb`'s display might be affected. `ipdb` versions <= 0.9.4 had brittle strategies for handling such scenarios.
  Fix: Update `ipdb` to the latest version. If issues persist, consult the documentation of the conflicting tool or consider alternative debugging approaches for that specific environment.
• gotcha Changes made to variables or objects within an IPython interactive shell launched from `ipdb` (e.g., using magic functions like `%ipdb`) might not always propagate back or affect the `ipdb` shell's context, leading to unexpected behavior if you expect state changes to persist across shells.
  Fix: Be aware that the IPython sub-shell might have a separate scope for certain operations. For direct manipulation of variables in the `ipdb` context, use `ipdb`'s native commands (e.g., `!variable = new_value`) or `p`rint to confirm state.

Install

• pip install ipdb Install ipdb

Imports

• set_trace

  import ipdb; ipdb.set_trace()

  The most common way to set a breakpoint in your code.
• pm

  import ipdb; ipdb.pm()

  Initiates post-mortem debugging after an uncaught exception, similar to IPython's %debug magic.
• launch_ipdb_on_exception

  from ipdb import launch_ipdb_on_exception

  Context manager to automatically launch ipdb if an exception is raised within the 'with' block.
• iex

  from ipdb import iex

  Decorator to launch ipdb if an exception is raised in the decorated function.

Quickstart

To use `ipdb.set_trace()`, simply place it where you want to pause execution. Run your script, and it will drop you into an interactive debugging session at that point. You can then inspect variables, step through code, and execute commands. Alternatively, run your script directly with `python -m ipdb your_script.py` to start debugging from the beginning or upon an exception.

import ipdb

def divide(a, b):
    try:
        # Set a breakpoint to inspect variables before the division
        ipdb.set_trace()
        result = a / b
        return result
    except ZeroDivisionError:
        print("Cannot divide by zero!")
        return None

print(divide(10, 2))
# To trigger the debugger on error, uncomment the line below and remove the try-except:
# divide(10, 0)