Issue 35900: Add pickler hook for the user to customize the serialization of user defined functions and types. (original) (raw)
Created on 2019-02-05 14:40 by pierreglaser, last changed 2022-04-11 14:59 by admin. This issue is now closed.
Messages (7)
Author: Pierre Glaser (pierreglaser) *
Date: 2019-02-05 14:40
Pickler objects provide a dispatch_table attribute, where the user can specify custom saving functions depending on the object-to-be-saved type. However, for performance purposes, this table is predated (in the C implementation only) by a hardcoded switch that will take care of the saving for many built-in types, without a lookup in the dispatch_table.
Especially, it is not possible to define custom saving methods for functions and classes, although the current default (save_global, that saves an object using its module attribute path) is likely to fail at pickling or unpickling time in many cases.
The aforementioned failures exist on purpose in the standard library (as a way to allow for the serialization of functions accessible from non-dynamic (*) modules only). However, there exist cases where serializing functions from dynamic modules matter. These cases are currently handled thanks the cloudpickle module (https://github.com/cloudpipe/cloudpickle), that is used by many distributed data-science frameworks such as pyspark, ray and dask. For the reasons explained above, cloudpickle's Pickler subclass derives from the python Pickler class instead of its C class, which severely harms its performance.
While prototyping with Antoine Pitrou, we came to the conclusion that a hook could be added to the C Pickler class, in which an optional user-defined callback would be invoked (if defined) when saving functions and classes instead of the traditional save_global. Here is a patch so that we can have something concrete of which to discuss.
(*) dynamic module are modules that cannot be imported by name as traditional python file backed module. Examples include the main module that can be populated dynamically by running a script or by a, user writing code in a python shell / jupyter notebook.
Author: Antoine Pitrou (pitrou) * 
Date: 2019-02-05 15:40
FYI, I've removed the duplicate message :-) Also adding Serhiy as cc.
Author: Olivier Grisel (Olivier.Grisel) *
Date: 2019-02-13 10:54
Adding such a hook would make it possible to reimplement cloudpickle.CloudPickler by deriving from the fast _pickle.Pickler class (instead of the slow pickle._Pickler as done currently). This would mean rewriting most of the CloudPickler method to only rely on a save_reduce-style design instead of directly calling pickle._Pickler.write and pickle._Pickler.save. This is tedious but doable.
There is however a blocker with the current way closures are set: when we pickle a dynamically defined function (e.g. lambda, nested function or function in main), we currently use a direct call to memoize (https://github.com/cloudpipe/cloudpickle/blob/v0.7.0/cloudpickle/cloudpickle.py#L594) so as to be able to refer to the function itself in its own closure without causing an infinite loop in CloudPickler.dump. This also makes possible to pickle mutually recursive functions.
The easiest way to avoid having to call memoize explicitly would be to be able to pass the full closure attribute in the state dict of the reduce call. Indeed the save_reduce function calls memoize automatically after saving the reconstructor and its args but prior to saving the state:
https://github.com/python/cpython/blob/v3.7.2/Modules/_pickle.c#L3903-L3931
It would therefore be possible to pass a (state, slotstate) tuple with the closure in slotstate that so it could be reconstructed at unpickling time with a setattr:
https://github.com/python/cpython/blob/v3.7.2/Modules/_pickle.c#L6258-L6272
However, it is currently not possible to setattr closure at the moment. We can only set individual closure cell contents (which is not compatible with the setattr state trick described above).
To summarize, we need to implement the setter function for the closure attribute of functions and methods to make it natural to reimplement the CloudPickler by inheriting from _pickle.Pickler using the hook described in this issue.
Author: Pierre Glaser (pierreglaser) *
Date: 2019-03-11 14:51
Update:
Instead of changing permission on some attributes of function objects (globals and closure), we added an optional argument called state_setter to save_reduce. This expects a callable that will be saved inside the object's pickle string, and called when setting the state of the object instead of using the default way in load_build. This allows for external flexibility when setting custom pickling behavior of built-in types (in our use-cases: function and classes). I updated the patches so that anyone interested can take a look.
Also, we tested the cloudpickle package against these patches (see https://github.com/cloudpipe/cloudpickle/pull/253). The tests run fine, and we observe a 10-30x speedup for real-life use-cases. We are starting to hit convergence on the implementation :)
Author: Antoine Pitrou (pitrou) *
Date: 2019-05-08 19:40
New changeset 65d98d0f53f558d7c799098da0abf376068c15fd by Antoine Pitrou (Pierre Glaser) in branch 'master': bpo-35900: Add a state_setter arg to save_reduce (GH-12588) https://github.com/python/cpython/commit/65d98d0f53f558d7c799098da0abf376068c15fd
Author: Antoine Pitrou (pitrou) *
Date: 2019-05-08 21:08
New changeset 289f1f80ee87a4baf4567a86b3425fb3bf73291d by Antoine Pitrou (Pierre Glaser) in branch 'master': bpo-35900: Enable custom reduction callback registration in _pickle (GH-12499) https://github.com/python/cpython/commit/289f1f80ee87a4baf4567a86b3425fb3bf73291d
Author: Antoine Pitrou (pitrou) *
Date: 2019-05-08 21:08
Both PRs are now merged. Thank you Pierre!
History
Date
User
Action
Args
2022-04-11 14:59:10
admin
set
github: 80081
2019-05-08 21:08:54
pitrou
set
status: open -> closed
resolution: fixed
messages: +
stage: patch review -> resolved
2019-05-08 21:08:30
pitrou
set
messages: +
2019-05-08 19:40:28
pitrou
set
messages: +
2019-03-27 14:39:40
pierreglaser
set
pull_requests: + <pull%5Frequest12530>
2019-03-22 21:51:41
pierreglaser
set
stage: patch review
pull_requests: + <pull%5Frequest12449>
2019-03-11 14:51:39
pierreglaser
set
files: + pickler_hook.patch
messages: +
2019-02-13 10:54:45
Olivier.Grisel
set
nosy: + Olivier.Grisel
messages: +
2019-02-05 15:40:42
pitrou
set
nosy: + serhiy.storchaka
messages: +
2019-02-05 15:40:07
pitrou
set
messages: -
2019-02-05 15:02:16
SilentGhost
set
title: Add pickler hoor for the user to customize the serialization of user defined functions and types. -> Add pickler hook for the user to customize the serialization of user defined functions and types.
2019-02-05 14:42:21
pierreglaser
set
files: + pickler_hook.patch
keywords: + patch
messages: +
2019-02-05 14:40:22
pierreglaser
create