[PATCH v2 00/12] qapi: static typing conversion, pt1.5

John Snow posted 12 patches 3 years, 4 months ago
Test checkpatch passed
Patches applied successfully (tree, apply log)
git fetch https://github.com/patchew-project/qemu tags/patchew/20201217015927.197287-1-jsnow@redhat.com
Maintainers: Michael Roth <michael.roth@amd.com>, Markus Armbruster <armbru@redhat.com>
There is a newer version of this series
scripts/qapi/commands.py                 |  11 ++-
scripts/qapi/events.py                   |  14 +--
scripts/qapi/gen.py                      | 108 ++++++++++++-----------
scripts/qapi/main.py                     |   2 +
scripts/qapi/mypy.ini                    |   1 -
scripts/qapi/schema.py                   |  35 ++++----
scripts/qapi/source.py                   |  20 ++++-
scripts/qapi/types.py                    |  11 +--
scripts/qapi/visit.py                    |   8 +-
tests/qapi-schema/comments.out           |   4 +-
tests/qapi-schema/doc-good.out           |   4 +-
tests/qapi-schema/empty.out              |   4 +-
tests/qapi-schema/event-case.out         |   4 +-
tests/qapi-schema/include-repetition.out |   8 +-
tests/qapi-schema/include-simple.out     |   6 +-
tests/qapi-schema/indented-expr.out      |   4 +-
tests/qapi-schema/qapi-schema-test.out   |   8 +-
tests/qapi-schema/test-qapi.py           |   2 +-
18 files changed, 145 insertions(+), 109 deletions(-)
[PATCH v2 00/12] qapi: static typing conversion, pt1.5
Posted by John Snow 3 years, 4 months ago
Hi, this patchset enables strict optional checking in mypy for
everything we have typed so far.

In general, this patchset seeks to eliminate Optional[T] in favor of T
wherever possible. Optional types used for object properties,
function/method parameters and return values where we expect, in most
cases, to be non-None is troublesome as it requires peppering the code
with assertions about state. If and whenever possible, prefer using
non-Optional types.

Ironing out these issues allows us to be even stricter with our type
checking, which improves consistency in subclass interface types and
should make review a little nicer.

This series is based on (but does not require) the 'qapi: sphinx-autodoc
for qapi module' series.

V2:

001/12:[----] [--] 'qapi/commands: assert arg_type is not None'
002/12:[----] [--] 'qapi/events: fix visit_event typing'
003/12:[0003] [FC] 'qapi/main: handle theoretical None-return from re.match()'
004/12:[----] [--] 'qapi/gen: assert that _start_if is not None in _wrap_ifcond'
005/12:[0003] [FC] 'qapi/gen: use './builtin' for the built-in module name'
006/12:[0004] [FC] 'qapi/source: Add builtin null-object sentinel'
007/12:[0024] [FC] 'qapi/schema: make QAPISourceInfo mandatory'
008/12:[----] [--] 'qapi/gen: write _genc/_genh access shims'
009/12:[----] [--] 'qapi/gen: move write method to QAPIGenC, make fname a str'
010/12:[----] [--] 'tests/qapi-schema: Add quotes to module name in test output'
011/12:[0004] [FC] 'qapi/schema: Name the builtin module "" instead of None'
012/12:[----] [--] 'qapi: enable strict-optional checks'

- This revision isn't quite complete, but due to deadlines and
  timezones, opted to send the "revision so far". It keeps some
  imperfect fixes that Markus is devising better alternatives for.

John Snow (12):
  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/source: Add builtin null-object sentinel
  qapi/schema: make QAPISourceInfo mandatory
  qapi/gen: write _genc/_genh access shims
  qapi/gen: move write method to QAPIGenC, make fname a str
  tests/qapi-schema: Add quotes to module name in test output
  qapi/schema: Name the builtin module "" instead of None
  qapi: enable strict-optional checks

 scripts/qapi/commands.py                 |  11 ++-
 scripts/qapi/events.py                   |  14 +--
 scripts/qapi/gen.py                      | 108 ++++++++++++-----------
 scripts/qapi/main.py                     |   2 +
 scripts/qapi/mypy.ini                    |   1 -
 scripts/qapi/schema.py                   |  35 ++++----
 scripts/qapi/source.py                   |  20 ++++-
 scripts/qapi/types.py                    |  11 +--
 scripts/qapi/visit.py                    |   8 +-
 tests/qapi-schema/comments.out           |   4 +-
 tests/qapi-schema/doc-good.out           |   4 +-
 tests/qapi-schema/empty.out              |   4 +-
 tests/qapi-schema/event-case.out         |   4 +-
 tests/qapi-schema/include-repetition.out |   8 +-
 tests/qapi-schema/include-simple.out     |   6 +-
 tests/qapi-schema/indented-expr.out      |   4 +-
 tests/qapi-schema/qapi-schema-test.out   |   8 +-
 tests/qapi-schema/test-qapi.py           |   2 +-
 18 files changed, 145 insertions(+), 109 deletions(-)

-- 
2.26.2



Re: [PATCH v2 00/12] qapi: static typing conversion, pt1.5
Posted by Markus Armbruster 3 years, 3 months ago
John Snow <jsnow@redhat.com> writes:

> Hi, this patchset enables strict optional checking in mypy for
> everything we have typed so far.
>
> In general, this patchset seeks to eliminate Optional[T] in favor of T
> wherever possible. Optional types used for object properties,
> function/method parameters and return values where we expect, in most
> cases, to be non-None is troublesome as it requires peppering the code
> with assertions about state. If and whenever possible, prefer using
> non-Optional types.
>
> Ironing out these issues allows us to be even stricter with our type
> checking, which improves consistency in subclass interface types and
> should make review a little nicer.
>
> This series is based on (but does not require) the 'qapi: sphinx-autodoc
> for qapi module' series.

Just two issues left that aren't entirely trivial: 1. mission creep
[PATCH 6+7], 2. integrating my "[PATCH 00/11] Drop support for QAPIGen
without a file name".