Fill PyTypeInfo::TYPE_HINT for built-in types #5619
Conversation
|
Not sure if it's fine to change the values of |
Tpt
force-pushed
the
tpt/builtin-type-hint
branch
from
85eedf0 to
9fbb660
Compare
Tpt
force-pushed
the
tpt/builtin-type-hint
branch
from
9fbb660 to
a91b0df
Compare
davidhewitt
left a comment
davidhewitt
left a comment
There was a problem hiding this comment.
Thanks, I am happy with the changes this adds, just one tidy up I think we can do with the macro implementation before we merge.
src/types/mod.rs
Outdated
| unsafe impl<$($generics,)*> $crate::type_object::PyTypeInfo for $name { | ||
| const NAME: &'static str = stringify!($name); | ||
| const MODULE: ::std::option::Option<&'static str> = $module; | ||
|
|
There was a problem hiding this comment.
Idea: rather than duplicating the whole macro, what we might be able to do is have $crate::pyobject_type_info_type_hint!($type_hint_module, $type_hint_expr) macro which we call here, and that macro can be the one with two defintitions. A noop one without the feature enabled, and one which expands to const TYPE_HINT = ... with the feature enabled.
Makes internal macros always require a module and a name These values are then used to fill PyTypeInfo::MODULE and PyTypeInfo::TYPE_HINT (PyTypeInfo::NAME stays unchanged) This changes some values of PyTypeInfo::MODULE
Head branch was pushed to by a user without write access
|
Rebase done |
Tpt
force-pushed
the
tpt/builtin-type-hint
branch
from
d9a3e18 to
39c7c60
Compare
github-merge-queue
bot
removed this pull request from the merge queue due to failed status checks
|
CI seems to have failed because of an internal error in the runner system |
3 participants
Makes internal macros always require a module and a name (using
builtinsfor Python builtins)These values are then used to fill
PyTypeInfo::MODULEandPyTypeInfo::TYPE_HINT(PyTypeInfo::NAMEstays unchanged)This changes some values of
PyTypeInfo::MODULE