You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This allows users to use CPython specific flags via HPy and reduce portability -- I suggest to test the flags against a mask that contains only the flags supported by HPy (currently: HPy_TPFLAGS_BASETYPE and HPy_TPFLAGS_HAVE_GC).
Would it be safer to treat 0 as the default value? We can still also treat what was HPy_TPFLAGS_DEFAULT as another way of specifying it, but without exposing a constant for it (and maybe with a warning?).
alternatively we can fail for 0 and suggest to use HPy_TPFLAGS_DEFAULT
The text was updated successfully, but these errors were encountered:
In general, it is a question if we should use flags in HPy. Maybe HPy should have bool (or int?) fields in HPyType_Spec for individual flags instead: that is safer, easier to use and migration from CPython API should be simple enough.
Currently HPy just copies
HPyType_Spec
flags toPyType_Spec
:and the user is expected to use at least
HPy_TPFLAGS_DEFAULT
in theirHPyType_Spec
, which exposes CPython specific flags in HPy headers:HPy_TPFLAGS_BASETYPE
andHPy_TPFLAGS_HAVE_GC
).0
as the default value? We can still also treat what wasHPy_TPFLAGS_DEFAULT
as another way of specifying it, but without exposing a constant for it (and maybe with a warning?).0
and suggest to useHPy_TPFLAGS_DEFAULT
The text was updated successfully, but these errors were encountered: