wireprotoserver: ensure that output stream gets flushed on exception
Previously flush was happening due to Python finalizer being run on
`BufferedWriter`. With upgrade to Python 3.11 this started randomly
failing.
My guess is that the finalizer on the raw `FileIO` object may
be running before the finalizer of `BufferedWriter` has a chance to run.
At any rate, since we're not relying on finalizers in the happy case
we should also not rely on them in case of exception.
Create a repo, set the username to something more than 255 bytes, then run hg amend on it.
$ unset HGUSER
$ cat >> $HGRCPATH << EOF
> [ui]
> username = aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa <very.long.name@example.com>
> [extensions]
> amend =
> [experimental]
> evolution.createmarkers=True
> evolution.exchange=True
> EOF
$ hg init tmpa
$ cd tmpa
$ echo a > a
$ hg add
adding a
$ hg commit -m "Initial commit"
$ echo a >> a
$ hg amend 2>&1 | grep -E -v '^(\*\*| )'
transaction abort!
rollback completed
Traceback (most recent call last):
*ProgrammingError: obsstore metadata value cannot be longer than 255 bytes (value "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa <very.long.name@example.com>" for key "user" is 285 bytes) (glob)