Skip to main content

Restart debugging your program in the function that failed.

Project description

hot_restart

Restart debugging your program in the function that failed.

Installation:

pip install hot_restart

hot_restart currently has no required dependencies outside the standard library. If pydevd is already imported, hot_restart will use it to allow debugging in VS Code (and other IDEs using pydevd / debugpy).

Usage:

Wrap any function you expect to crash in @hot_restart.wrap:

import hot_restart

@hot_restart.wrap
def my_func():
    ...

@hot_restart.wrap_class
class MyClass:

    def forward(self, x):
        super().forward(x * 2)

Alternatively, call hot_restart.wrap_module() after your functions and classes are defined.

When a wrapped function crashes, hot_restart will open a post mortem debugger in the wrapped function, with the rest of the stack still live above that function call.

Then you can modify the source code of your crashing function, and (c)ontinue from the debugger, and hot_restart will reload the source of that function (and only that function).

You can use @hot_restart.no_wrap to indicate that a function or class should not be wrapped.

hot_restart uses AST transformations to find new definitions, to avoid causing "import-time" side effects, and to match line numbers with the real source code while keeping each version of a function (except the initial load) in its own source file.

Edge Cases

You can also trigger a full module reload with hot_restart.reload_module(). This may trigger "import-time" side effects and duplicate (i.e. conflicting) class definitions. To avoid restarting your main function when reloading your main module, check hot_restart.is_restarting_module().

hot_restart uses source re-writing to handle the common cases, and does not patch the byte-code of functions. This limits hot_restart from handling adding new variables to a closure or adding methods to existing class instances, but simplifies the implementation.

Line numbers and Surrogate Sources

hot_restart generates a surrogate source file to compile. This avoids import-time side effects on reload, but means that line numbers may become slightly different than the source on disk. When using pdb, by default the source of the reloaded file will be set to the surrogate source file, ensuring that line numbers in pdb match the executed code. In other debuggers (or when hot_restart.DEBUG_ORIGINAL_PATH_RELOADED_CODE is True), the original source will be used. This may result in line numbers in the executed code not matching line numbers in the debugger.

super() Calls

hot_restart rewrites super() into super(<classname>, <first argument>) in reloaded source (original source is loaded intact). This prevents methods from acquiring a closure variable __class__, which would break in many cases, but slightly changes lookup behavior if the class is redefined.

Closures and Nested Functions

hot_restart will patch in old closure variables into newly defined functions, as long as hot_restart.wrap is the innermost decorator. However, functions cannot add new closure variables without a full module reload. Functions can still gain new arguments.

For nested functions, hot_restart.wrap_module() cannot find the inner function, so hot_restart.wrap() must be used manually.

If hot_restart.wrap is not the inner-most decorator, then closure variables will be lost.

Alternative / Complementary Tools:

If you just want more complete hot reloading (for example, the ability to add new methods to existing instances of classes), you should consider jurigged. You should also be able to use jurigged with hot_restart. If you want to use jurigged's code reloading with hot_restart's function restarting, you can disable hot_restart's automatic code reloading by setting:

hot_restart.RELOAD_ON_CONTINUE = False

This module implements a workflow similar to "edit-and-continue". If you're looking for a more maximalist implementation of this workflow with IDE integration, you may be interested in Reloadium.

Although these useful alternatives exist, I created hot_restart because I found it difficult to debug failures in those systems. hot_restart intentionally implements a more minimal set of reloading routines that I can easily wrap my head around.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

hot_restart-0.2.2.tar.gz (14.5 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

hot_restart-0.2.2-py2.py3-none-any.whl (11.5 kB view details)

Uploaded Python 2Python 3

File details

Details for the file hot_restart-0.2.2.tar.gz.

File metadata

  • Download URL: hot_restart-0.2.2.tar.gz
  • Upload date:
  • Size: 14.5 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: python-requests/2.32.3

File hashes

Hashes for hot_restart-0.2.2.tar.gz
Algorithm Hash digest
SHA256 7f376aa0c6929c2b322e20c30b0352fc8cb3f4edd393cf8608848ae78a75462b
MD5 aa43ada992d73a0d7e8a5e78046a2d43
BLAKE2b-256 d552e58d9f29082f211a1f681ae484a904f5e4c038faa003010ec6b7721f0937

See more details on using hashes here.

File details

Details for the file hot_restart-0.2.2-py2.py3-none-any.whl.

File metadata

  • Download URL: hot_restart-0.2.2-py2.py3-none-any.whl
  • Upload date:
  • Size: 11.5 kB
  • Tags: Python 2, Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: python-requests/2.32.3

File hashes

Hashes for hot_restart-0.2.2-py2.py3-none-any.whl
Algorithm Hash digest
SHA256 a5f9c25c8cb78e67fa11358738dededc924835579348822f9d6b6d90a0d4ee7e
MD5 64fb7e970d152af1e349e277582d84f7
BLAKE2b-256 21f56a0d96b97a769ff320ee4fd3d7b36b64271442805cc262c7eb5a0478e149

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page