hgdemandimport: apply lazy module loading to sys.meta_path finders
Python's `sys.meta_path` finders are the primary objects whose job it
is to find a module at import time. When `import` is called, Python
iterates objects in this list and calls `o.find_spec(...)` to find
a `ModuleSpec` (or None if the module couldn't be found by that
finder). If no meta path finder can find a module, import fails.
One of the default meta path finders is `PathFinder`. Its job is to
import modules from the filesystem and is probably the most important
importer. This finder looks at `sys.path` and `sys.path_hooks` to do
its job.
The `ModuleSpec` returned by `MetaPathImporter.find_spec()` has a
`loader` attribute, which defines the concrete module loader to use.
`sys.path_hooks` is a hook point for teaching `PathFinder` to
instantiate custom loader types.
Previously, we injected a custom `sys.path_hook` that told `PathFinder`
to wrap the default loaders with a loader that creates a module object
that is lazy.
This approach worked. But its main limitation was that it only applied
to the `PathFinder` meta path importer. There are other meta path
importers that are registered. And in the case of PyOxidizer loading
modules from memory, `PathFinder` doesn't come into play since
PyOxidizer's own meta path importer was handling all imports.
This commit changes our approach to lazy module loading by proxying
all meta path importers. Specifically, we overload the `find_spec()`
method to swap in a wrapped loader on the `ModuleSpec` before it
is returned. The end result of this is all meta path importers should
be lazy.
As much as I would have loved to utilize .__class__ manipulation to
achieve this, some meta path importers are implemented in C/Rust
in such a way that they cannot be monkeypatched. This is why we
use __getattribute__ to define a proxy.
Also, this change could theoretically open us up to regressions in
meta path importers whose loader is creating module objects which
can't be monkeypatched. But I'm not aware of any of these in the
wild. So I think we'll be safe.
According to hyperfine, this change yields a decent startup time win of
5-6ms:
```
Benchmark #1: ~/.pyenv/versions/3.6.10/bin/python ./hg version
Time (mean ? ?): 86.8 ms ? 0.5 ms [User: 78.0 ms, System: 8.7 ms]
Range (min ? max): 86.0 ms ? 89.1 ms 50 runs
Time (mean ? ?): 81.1 ms ? 2.7 ms [User: 74.5 ms, System: 6.5 ms]
Range (min ? max): 77.8 ms ? 90.5 ms 50 runs
Benchmark #2: ~/.pyenv/versions/3.7.6/bin/python ./hg version
Time (mean ? ?): 78.9 ms ? 0.6 ms [User: 70.2 ms, System: 8.7 ms]
Range (min ? max): 78.1 ms ? 81.2 ms 50 runs
Time (mean ? ?): 73.4 ms ? 0.6 ms [User: 65.3 ms, System: 8.0 ms]
Range (min ? max): 72.4 ms ? 75.7 ms 50 runs
Benchmark #3: ~/.pyenv/versions/3.8.1/bin/python ./hg version
Time (mean ? ?): 78.1 ms ? 0.6 ms [User: 70.2 ms, System: 7.9 ms]
Range (min ? max): 77.4 ms ? 80.9 ms 50 runs
Time (mean ? ?): 72.1 ms ? 0.4 ms [User: 64.4 ms, System: 7.6 ms]
Range (min ? max): 71.4 ms ? 74.1 ms 50 runs
```
Differential Revision: https://phab.mercurial-scm.org/D7954
# discovery.py - protocol changeset discovery functions
#
# Copyright 2010 Matt Mackall <mpm@selenic.com>
#
# This software may be used and distributed according to the terms of the
# GNU General Public License version 2 or any later version.
from __future__ import absolute_import
import collections
from .i18n import _
from .node import (
nullid,
short,
)
from . import (
error,
pycompat,
)
def findcommonincoming(repo, remote, heads=None, force=False):
"""Return a tuple (common, fetch, heads) used to identify the common
subset of nodes between repo and remote.
"common" is a list of (at least) the heads of the common subset.
"fetch" is a list of roots of the nodes that would be incoming, to be
supplied to changegroupsubset.
"heads" is either the supplied heads, or else the remote's heads.
"""
knownnode = repo.changelog.hasnode
search = []
fetch = set()
seen = set()
seenbranch = set()
base = set()
if not heads:
with remote.commandexecutor() as e:
heads = e.callcommand(b'heads', {}).result()
if repo.changelog.tip() == nullid:
base.add(nullid)
if heads != [nullid]:
return [nullid], [nullid], list(heads)
return [nullid], [], heads
# assume we're closer to the tip than the root
# and start by examining the heads
repo.ui.status(_(b"searching for changes\n"))
unknown = []
for h in heads:
if not knownnode(h):
unknown.append(h)
else:
base.add(h)
if not unknown:
return list(base), [], list(heads)
req = set(unknown)
reqcnt = 0
progress = repo.ui.makeprogress(_(b'searching'), unit=_(b'queries'))
# search through remote branches
# a 'branch' here is a linear segment of history, with four parts:
# head, root, first parent, second parent
# (a branch always has two parents (or none) by definition)
with remote.commandexecutor() as e:
branches = e.callcommand(b'branches', {b'nodes': unknown}).result()
unknown = collections.deque(branches)
while unknown:
r = []
while unknown:
n = unknown.popleft()
if n[0] in seen:
continue
repo.ui.debug(b"examining %s:%s\n" % (short(n[0]), short(n[1])))
if n[0] == nullid: # found the end of the branch
pass
elif n in seenbranch:
repo.ui.debug(b"branch already found\n")
continue
elif n[1] and knownnode(n[1]): # do we know the base?
repo.ui.debug(
b"found incomplete branch %s:%s\n"
% (short(n[0]), short(n[1]))
)
search.append(n[0:2]) # schedule branch range for scanning
seenbranch.add(n)
else:
if n[1] not in seen and n[1] not in fetch:
if knownnode(n[2]) and knownnode(n[3]):
repo.ui.debug(b"found new changeset %s\n" % short(n[1]))
fetch.add(n[1]) # earliest unknown
for p in n[2:4]:
if knownnode(p):
base.add(p) # latest known
for p in n[2:4]:
if p not in req and not knownnode(p):
r.append(p)
req.add(p)
seen.add(n[0])
if r:
reqcnt += 1
progress.increment()
repo.ui.debug(
b"request %d: %s\n" % (reqcnt, b" ".join(map(short, r)))
)
for p in pycompat.xrange(0, len(r), 10):
with remote.commandexecutor() as e:
branches = e.callcommand(
b'branches', {b'nodes': r[p : p + 10],}
).result()
for b in branches:
repo.ui.debug(
b"received %s:%s\n" % (short(b[0]), short(b[1]))
)
unknown.append(b)
# do binary search on the branches we found
while search:
newsearch = []
reqcnt += 1
progress.increment()
with remote.commandexecutor() as e:
between = e.callcommand(b'between', {b'pairs': search}).result()
for n, l in zip(search, between):
l.append(n[1])
p = n[0]
f = 1
for i in l:
repo.ui.debug(b"narrowing %d:%d %s\n" % (f, len(l), short(i)))
if knownnode(i):
if f <= 2:
repo.ui.debug(
b"found new branch changeset %s\n" % short(p)
)
fetch.add(p)
base.add(i)
else:
repo.ui.debug(
b"narrowed branch search to %s:%s\n"
% (short(p), short(i))
)
newsearch.append((p, i))
break
p, f = i, f * 2
search = newsearch
# sanity check our fetch list
for f in fetch:
if knownnode(f):
raise error.RepoError(_(b"already have changeset ") + short(f[:4]))
base = list(base)
if base == [nullid]:
if force:
repo.ui.warn(_(b"warning: repository is unrelated\n"))
else:
raise error.Abort(_(b"repository is unrelated"))
repo.ui.debug(
b"found new changesets starting at "
+ b" ".join([short(f) for f in fetch])
+ b"\n"
)
progress.complete()
repo.ui.debug(b"%d total queries\n" % reqcnt)
return base, list(fetch), heads