bundle2: don't try to recover from a GeneratorExit (issue4785)
GeneratorExit means the other end of the conversation has already
stopped listening, so don't try and yield out error
information. Instead, just let the GeneratorExit propagate normally.
This should resolve esoteric issues observed with servers that have
aggressive timeouts waiting for data to send to clients logging
internal Python errors[0]. This has been observed with both gunicorn's
gevent worker model and with scm-manager's built-in webserver (which
itself is something sitting inside jetty.)
0: Exception RuntimeError: 'generator ignored GeneratorExit' in <generator object getchunks at 0x7fd2f6c586e0> ignored
http://mercurial.selenic.com/bts/issue2137
Setup:
create a little extension that has 3 side-effects:
1) ensure changelog data is not inlined
2) make revlog to use lazyparser
3) test that repo.lookup() works
1 and 2 are preconditions for the bug; 3 is the bug.
$ cat > commitwrapper.py <<EOF
> from mercurial import extensions, node, revlog
>
> def reposetup(ui, repo):
> class wraprepo(repo.__class__):
> def commit(self, *args, **kwargs):
> result = super(wraprepo, self).commit(*args, **kwargs)
> tip1 = node.short(repo.changelog.tip())
> tip2 = node.short(repo.lookup(tip1))
> assert tip1 == tip2
> ui.write('new tip: %s\n' % tip1)
> return result
> repo.__class__ = wraprepo
>
> def extsetup(ui):
> revlog._maxinline = 8 # split out 00changelog.d early
> revlog._prereadsize = 8 # use revlog.lazyparser
> EOF
$ cat >> $HGRCPATH <<EOF
> [extensions]
> commitwrapper = `pwd`/commitwrapper.py
> EOF
$ hg init repo1
$ cd repo1
$ echo a > a
$ hg commit -A -m'add a with a long commit message to make the changelog a bit bigger'
adding a
new tip: 553596fad57b
Test that new changesets are visible to repo.lookup():
$ echo a >> a
$ hg commit -m'one more commit to demonstrate the bug'
new tip: 799ae3599e0e
$ hg tip
changeset: 1:799ae3599e0e
tag: tip
user: test
date: Thu Jan 01 00:00:00 1970 +0000
summary: one more commit to demonstrate the bug
$ cd ..