John Snow posted [PATCH 09/12] qapi/gen: move write method to QAPIGenC, make fname a str
QAPIGenC and QAPIGenH in particular depend on fname being defined, but we have a usage of QAPIGenCCode that isn't intended to be associated with a particular file. No problem, move the write method down to the class that actually needs it, and keep QAPIGenCCode more abstract. Signed-off-by: John Snow <js...@redhat.com> --- scripts/qapi/commands.py | 2 +- scripts/qapi/gen.py | 54 ++++++++++++++++++++-------------------- 2 files changed, 28 insertions(+), 28 deletions(-) There is just one user of QAPIGen without a file name, and it's awkward. Let's get rid of it. Since my work to get rid of it depends on parts of John's series, and I'm pressed for time, I include the parts I need in my series. John, feel free to pick this into your complete series. John Snow (6): qapi/commands: assert arg_type is not None qapi/events: fix visit_event typing qapi/main: handle theoretical None-return from re.match() qapi/gen: assert that _start_if is not None in _wrap_ifcond qapi/gen: use './builtin' for the built-in module name qapi/gen: write _genc/_genh access shims Markus Armbruster (5): qapi/gen: Replace ._begin_system_module() qapi/gen: Expose a single module name space qapi/gen: Support for switching to another module temporarily qapi/commands: Simplify command registry generation qapi/gen: Drop support for QAPIGen without a file name scripts/qapi/commands.py | 60 ++++++++++++++++----------------- scripts/qapi/events.py | 14 ++++---- scripts/qapi/gen.py | 72 +++++++++++++++++++++++----------------- scripts/qapi/main.py | 2 ++ scripts/qapi/types.py | 2 +- scripts/qapi/visit.py | 2 +- 6 files changed, 82 insertions(+), 70 deletions(-) -- 2.26.2