keepalive: ensure `close_all()` actually closes all cached connections
While debugging why LFS blob downloads are getting corrupted with workers, I
noticed that prior to spinning up the workers, the ConnectionManager has 2
connections to the server and calling `KeepAliveHandler.close_all()` left one
behind. The reason is the value component of `self._cm.get_all().items()` is a
list, and `self._cm.remove()` modifies said list while the caller is iterating
over it. Now `get_all()` is a deep copy of both the dict and lists in all
cases.
import os
from mercurial import (
dispatch,
ui as uimod,
)
from mercurial.utils import stringutil
# ensure errors aren't buffered
testui = uimod.ui()
testui.pushbuffer()
testui.writenoi18n(b'buffered\n')
testui.warnnoi18n(b'warning\n')
testui.write_err(b'error\n')
print(stringutil.pprint(testui.popbuffer(), bprefix=True).decode('ascii'))
# test dispatch.dispatch with the same ui object
hgrc = open(os.environ["HGRCPATH"], 'wb')
hgrc.write(b'[extensions]\n')
hgrc.write(b'color=\n')
hgrc.close()
ui_ = uimod.ui.load()
ui_.setconfig(b'ui', b'formatted', b'True')
# we're not interested in the output, so write that to devnull
ui_.fout = open(os.devnull, 'wb')
# call some arbitrary command just so we go through
# color's wrapped _runcommand twice.
def runcmd():
dispatch.dispatch(dispatch.request([b'version', b'-q'], ui_))
runcmd()
print("colored? %s" % (ui_._colormode is not None))
runcmd()
print("colored? %s" % (ui_._colormode is not None))