From: John Snow <jsnow@redhat.com>
To: Markus Armbruster <armbru@redhat.com>
Cc: Michael Roth <michael.roth@amd.com>,
Cleber Rosa <crosa@redhat.com>,
qemu-devel@nongnu.org, Eduardo Habkost <ehabkost@redhat.com>
Subject: Re: [PATCH v6 19/19] qapi/introspect.py: add SchemaMetaType enum
Date: Tue, 16 Feb 2021 10:08:38 -0500 [thread overview]
Message-ID: <ff534bd4-ad04-1afe-b820-cc1482c8566b@redhat.com> (raw)
In-Reply-To: <87pn10nz57.fsf@dusky.pond.sub.org>
On 2/16/21 4:24 AM, Markus Armbruster wrote:
> John Snow <jsnow@redhat.com> writes:
>
>> Follows the qapi/introspect.py definition of the same; this adds a more
>> precise typing to _gen_tree's mtype parameter.
>>
>> NB: print(SchemaMetaType.BUILTIN) would produce the string
>> "SchemaMetaType.BUILTIN", but when using format strings (.format or f-strings),
>> it relies on the __format__ method defined in the Enum class, which uses the
>> "value" of the enum instead, producing the string "builtin".
>>
>> For consistency with old-style format strings (which simply call the
>> __str__ method of an object), a __str__ dunder is added, though it is
>> not actually used here in this code.
>>
>> Signed-off-by: John Snow <jsnow@redhat.com>
>> ---
>> scripts/qapi/introspect.py | 38 +++++++++++++++++++++++++++++---------
>> 1 file changed, 29 insertions(+), 9 deletions(-)
>>
>> diff --git a/scripts/qapi/introspect.py b/scripts/qapi/introspect.py
>> index c6f5cf8d874..008a21f5c4c 100644
>> --- a/scripts/qapi/introspect.py
>> +++ b/scripts/qapi/introspect.py
>> @@ -11,6 +11,7 @@
>> See the COPYING file in the top-level directory.
>> """
>>
>> +from enum import Enum
>> from typing import (
>> Any,
>> Dict,
>> @@ -79,6 +80,23 @@
>> SchemaInfoCommand = Dict[str, object]
>>
>>
>> +class SchemaMetaType(str, Enum):
>> + """
>> + Mimics the SchemaMetaType enum from qapi/introspect.json.
>> + """
>> + BUILTIN = 'builtin'
>> + ENUM = 'enum'
>> + ARRAY = 'array'
>> + OBJECT = 'object'
>> + ALTERNATE = 'alternate'
>> + COMMAND = 'command'
>> + EVENT = 'event'
>> +
>> + def __str__(self) -> str:
>> + # Needed for intuitive behavior with old-style format strings.
>> + return str(self.value)
>> +
>> +
>
> The fanciness compared to plain Enum('SchemaMetaType', 'BUILTIN ...')
> avoids extra code to map the enum values to the strings with need.
>
I wasn't even aware there was a short form. (TIL!)
This form allows me to inherit from str and pass the value around
anywhere strings are used. Due to the generalized nature of
tree_to_qlit, using the short constructor form (which creates ints)
would need additional magic to be useful.
You can almost replicate it:
_values = ('builtin', 'enum', 'array')
_nv_pairs = [(value.upper(), value) for value in _values]
SchemaMetaType = enum.Enum('SchemaMetaType', _nv_pairs, type=str)
though this loses out on the __str__ method hack, and I don't think mypy
will be able to introspect into this functional constructor.
>> _ValueT = TypeVar('_ValueT', bound=_Value)
>>
>>
>> @@ -251,7 +269,8 @@ def _gen_features(features: Sequence[QAPISchemaFeature]
>> ) -> List[Annotated[str]]:
>> return [Annotated(f.name, f.ifcond) for f in features]
>>
>> - def _gen_tree(self, name: str, mtype: str, obj: Dict[str, object],
>> + def _gen_tree(self, name: str, mtype: SchemaMetaType,
>> + obj: Dict[str, object],
>> ifcond: Sequence[str] = (),
>> features: Sequence[QAPISchemaFeature] = ()) -> None:
>> """
>> @@ -299,7 +318,7 @@ def _gen_variant(self, variant: QAPISchemaVariant
>>
>> def visit_builtin_type(self, name: str, info: Optional[QAPISourceInfo],
>> json_type: str) -> None:
>> - self._gen_tree(name, 'builtin', {'json-type': json_type})
>> + self._gen_tree(name, SchemaMetaType.BUILTIN, {'json-type': json_type})
>>
>> def visit_enum_type(self, name: str, info: Optional[QAPISourceInfo],
>> ifcond: Sequence[str],
>> @@ -307,7 +326,7 @@ def visit_enum_type(self, name: str, info: Optional[QAPISourceInfo],
>> members: List[QAPISchemaEnumMember],
>> prefix: Optional[str]) -> None:
>> self._gen_tree(
>> - name, 'enum',
>> + name, SchemaMetaType.ENUM,
>> {'values': [Annotated(m.name, m.ifcond) for m in members]},
>> ifcond, features
>> )
>> @@ -316,8 +335,8 @@ def visit_array_type(self, name: str, info: Optional[QAPISourceInfo],
>> ifcond: Sequence[str],
>> element_type: QAPISchemaType) -> None:
>> element = self._use_type(element_type)
>> - self._gen_tree('[' + element + ']', 'array', {'element-type': element},
>> - ifcond)
>> + self._gen_tree('[' + element + ']', SchemaMetaType.ARRAY,
>> + {'element-type': element}, ifcond)
>>
>> def visit_object_type_flat(self, name: str, info: Optional[QAPISourceInfo],
>> ifcond: Sequence[str],
>> @@ -330,14 +349,14 @@ def visit_object_type_flat(self, name: str, info: Optional[QAPISourceInfo],
>> if variants:
>> obj['tag'] = variants.tag_member.name
>> obj['variants'] = [self._gen_variant(v) for v in variants.variants]
>> - self._gen_tree(name, 'object', obj, ifcond, features)
>> + self._gen_tree(name, SchemaMetaType.OBJECT, obj, ifcond, features)
>>
>> def visit_alternate_type(self, name: str, info: Optional[QAPISourceInfo],
>> ifcond: Sequence[str],
>> features: List[QAPISchemaFeature],
>> variants: QAPISchemaVariants) -> None:
>> self._gen_tree(
>> - name, 'alternate',
>> + name, SchemaMetaType.ALTERNATE,
>> {'members': [Annotated({'type': self._use_type(m.type)},
>> m.ifcond)
>> for m in variants.variants]},
>> @@ -361,7 +380,7 @@ def visit_command(self, name: str, info: Optional[QAPISourceInfo],
>> }
>> if allow_oob:
>> obj['allow-oob'] = allow_oob
>> - self._gen_tree(name, 'command', obj, ifcond, features)
>> + self._gen_tree(name, SchemaMetaType.COMMAND, obj, ifcond, features)
>>
>> def visit_event(self, name: str, info: Optional[QAPISourceInfo],
>> ifcond: Sequence[str], features: List[QAPISchemaFeature],
>> @@ -370,7 +389,8 @@ def visit_event(self, name: str, info: Optional[QAPISourceInfo],
>> assert self._schema is not None
>>
>> arg_type = arg_type or self._schema.the_empty_object_type
>> - self._gen_tree(name, 'event', {'arg-type': self._use_type(arg_type)},
>> + self._gen_tree(name, SchemaMetaType.EVENT,
>> + {'arg-type': self._use_type(arg_type)},
>> ifcond, features)
>
> Gain: _gen_tree()'s second argument's type now serves as documentation,
> and passing crap to it becomes harder.
>
> Gut feeling: too much notational overhead for too little gain.
>
> Opinions?
>
No strong feelings one way or the other, honestly. I wrote this mostly
to see how much the overhead would be based on your comment about the
loose typing of meta-type as str.
If it's too much for too little for you, I'm okay without it too.
--js
next prev parent reply other threads:[~2021-02-16 15:10 UTC|newest]
Thread overview: 42+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-02-16 2:17 [PATCH v6 00/19] qapi: static typing conversion, pt2 John Snow
2021-02-16 2:17 ` [PATCH v6 01/19] qapi: Replace List[str] with Sequence[str] for ifcond John Snow
2021-02-16 8:43 ` Markus Armbruster
2021-02-16 15:19 ` John Snow
2021-02-16 15:58 ` Markus Armbruster
2021-02-16 2:17 ` [PATCH v6 02/19] qapi/introspect.py: assert schema is not None John Snow
2021-02-16 2:17 ` [PATCH v6 03/19] qapi/introspect.py: use _make_tree for features nodes John Snow
2021-02-16 2:17 ` [PATCH v6 04/19] qapi/introspect.py: add _gen_features helper John Snow
2021-02-16 2:17 ` [PATCH v6 05/19] qapi/introspect.py: guard against ifcond/comment misuse John Snow
2021-02-17 12:01 ` Markus Armbruster
2021-02-16 2:17 ` [PATCH v6 06/19] qapi/introspect.py: Unify return type of _make_tree() John Snow
2021-02-16 2:17 ` [PATCH v6 07/19] qapi/introspect.py: replace 'extra' dict with 'comment' argument John Snow
2021-02-16 2:17 ` [PATCH v6 08/19] qapi/introspect.py: Always define all 'extra' dict keys John Snow
2021-02-16 2:17 ` [PATCH v6 09/19] qapi/introspect.py: Introduce preliminary tree typing John Snow
2021-02-16 2:18 ` [PATCH v6 10/19] qapi/introspect.py: create a typed 'Annotated' data strutcure John Snow
2021-02-16 2:18 ` [PATCH v6 11/19] qapi/introspect.py: improve _tree_to_qlit error message John Snow
2021-02-16 2:18 ` [PATCH v6 12/19] qapi/introspect.py: improve readability of _tree_to_qlit John Snow
2021-02-16 2:18 ` [PATCH v6 13/19] qapi/introspect.py: remove _gen_variants helper John Snow
2021-02-16 2:18 ` [PATCH v6 14/19] qapi/introspect.py: add type hint annotations John Snow
2021-02-16 8:55 ` Markus Armbruster
2021-02-16 8:58 ` Markus Armbruster
2021-02-16 15:33 ` John Snow
2021-02-16 16:08 ` Markus Armbruster
2021-02-16 16:19 ` John Snow
2021-02-17 9:21 ` Markus Armbruster
2021-02-18 18:56 ` Markus Armbruster
2021-02-18 19:01 ` Markus Armbruster
2021-02-16 2:18 ` [PATCH v6 15/19] qapi/introspect.py: Add docstrings to _gen_tree and _tree_to_qlit John Snow
2021-02-17 9:39 ` Markus Armbruster
2021-02-17 16:07 ` John Snow
2021-02-17 16:35 ` Markus Armbruster
2021-02-17 16:55 ` John Snow
2021-02-18 7:53 ` Markus Armbruster
2021-02-16 2:18 ` [PATCH v6 16/19] qapi/introspect.py: Update copyright and authors list John Snow
2021-02-16 2:18 ` [PATCH v6 17/19] qapi/introspect.py: Type _gen_tree variants as Sequence[str] John Snow
2021-02-16 9:10 ` Markus Armbruster
2021-02-16 15:13 ` John Snow
2021-02-16 2:18 ` [PATCH v6 18/19] qapi/introspect.py: set _gen_tree's default ifcond argument to () John Snow
2021-02-16 2:18 ` [PATCH v6 19/19] qapi/introspect.py: add SchemaMetaType enum John Snow
2021-02-16 9:24 ` Markus Armbruster
2021-02-16 15:08 ` John Snow [this message]
2021-02-16 16:09 ` Markus Armbruster
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=ff534bd4-ad04-1afe-b820-cc1482c8566b@redhat.com \
--to=jsnow@redhat.com \
--cc=armbru@redhat.com \
--cc=crosa@redhat.com \
--cc=ehabkost@redhat.com \
--cc=michael.roth@amd.com \
--cc=qemu-devel@nongnu.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).