On 10/7/20 8:43 AM, Markus Armbruster wrote:
John Snow <js...@redhat.com> writes:
This is true by design, but not presently able to be expressed in the
type system. An assertion helps mypy understand our constraints.
Signed-off-by: John Snow <js...@redhat.com>
Reviewed-by: Eduardo Habkost <ehabk...@redhat.com>
Reviewed-by: Cleber Rosa <cr...@redhat.com>
---
scripts/qapi/visit.py | 12 +++++++-----
1 file changed, 7 insertions(+), 5 deletions(-)
diff --git a/scripts/qapi/visit.py b/scripts/qapi/visit.py
index 14f30c228b7..4f11fd325b8 100644
--- a/scripts/qapi/visit.py
+++ b/scripts/qapi/visit.py
@@ -22,7 +22,7 @@
mcgen,
)
from .gen import QAPISchemaModularCVisitor, ifcontext
-from .schema import QAPISchemaObjectType
+from .schema import QAPISchemaEnumType, QAPISchemaObjectType
def gen_visit_decl(name, scalar=False):
@@ -84,15 +84,17 @@ def gen_visit_object_members(name, base, members, variants):
ret += gen_endif(memb.ifcond)
if variants:
+ tag_member = variants.tag_member
+ assert isinstance(tag_member.type, QAPISchemaEnumType)
I'm curious: do you need the local variable to make the assertion stick?
No, but it only sticks to the binding and not the data. i.e. assertions
to downcast work on the *name*.
(This comes up somewhere in the schema.py patches where I make a change
that looks completely pointless, but it makes mypy happy.)
I could have left it alone. I just saw a lot of repeated multi-dots and
habitually created a temporary local for the purpose.
+
ret += mcgen('''
switch (obj->%(c_name)s) {
''',
- c_name=c_name(variants.tag_member.name))
+ c_name=c_name(tag_member.name))
for var in variants.variants:
- case_str = c_enum_const(variants.tag_member.type.name,
- var.name,
- variants.tag_member.type.prefix)
+ case_str = c_enum_const(tag_member.type.name, var.name,
+ tag_member.type.prefix)
ret += gen_if(var.ifcond)
if var.type.name == 'q_empty':
# valid variant and nothing to do
