Add a method on importlib.SourceLoader for creating bytecode file format/container

[brettcannon]

BPO	16494
Nosy	@brettcannon, @jcea, @asvetlov

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = 'https://github.com/brettcannon'
closed_at = <Date 2014-07-18.18:50:42.350>
created_at = <Date 2012-11-17.16:00:17.112>
labels = ['type-feature', 'library']
title = 'Add a method on importlib.SourceLoader for creating bytecode file format/container'
updated_at = <Date 2014-07-18.18:50:42.349>
user = 'https://github.com/brettcannon'

bugs.python.org fields:

activity = <Date 2014-07-18.18:50:42.349>
actor = 'brett.cannon'
assignee = 'brett.cannon'
closed = True
closed_date = <Date 2014-07-18.18:50:42.350>
closer = 'brett.cannon'
components = ['Library (Lib)']
creation = <Date 2012-11-17.16:00:17.112>
creator = 'brett.cannon'
dependencies = []
files = []
hgrepos = []
issue_num = 16494
keywords = []
message_count = 3.0
messages = ['175762', '222782', '223425']
nosy_count = 4.0
nosy_names = ['brett.cannon', 'jcea', 'asvetlov', 'BreamoreBoy']
pr_nums = []
priority = 'normal'
resolution = 'rejected'
stage = 'test needed'
status = 'closed'
superseder = None
type = 'enhancement'
url = 'https://bugs.python.org/issue16494'
versions = ['Python 3.5']

[brettcannon]

With issue bpo-15031 providing a way to take a bytecode file and get a code object out of it, there should probably be a comparable method that takes a code object and a path stats dict and returns the bytes to write out to a file.

This could also be a slight performance win for subclasses of SourceLoader which can read bytecode files but not write them (e.g. zip files where the cost of modifying a zip file is too high). That way the method could raise NotImplementedError (or return None) to signify it doesn't want anything written out by SourceLoader.get_code() and thus skip the entire step of preparing the bytecode file bytes instead of what happens now where set_data() raises NotImplementedError after the whole set of bytes is constructed.

More importantly, it would clean up py_compile once issue bpo-15627 is in with an 'optimize' flag as py_compile.compile() would essentially become source = loader.get_source(...); code = loader.compile_source(source, optimize=...); pyc_bytes = loader.create_bytecode_file(code, loader.path_stats(...)). Otherwise get_code() will need to grow an 'optimize' flag and that doesn't seem necessary.

[BreamoreBoy]

@brett as bpo-15627 is in would you like to follow up on this.

[brettcannon]

I think I want to keep bytecode files as optimizations as much as possible, so exposing anything that explicitly writes them is just going to lead to headaches.