Track for new and join on both str and unicode #40
Conversation
| @@ -105,6 +105,7 @@ whose size is determined when the object is allocated. | |||
| */ | |||
| typedef struct _object { | |||
| PyObject_HEAD | |||
| Py_ssize_t ob_bstate; | |||
There was a problem hiding this comment.
Don't we want to put this only in strings/bytes, rather than in every object? Or is that not possible?
There was a problem hiding this comment.
Some methods use the abstract object type see: #39 , so it is necessary.
There was a problem hiding this comment.
I'm unsure what it means to ask for the bstate of something that isn't a string. Maybe I'm missing something!
| @@ -1646,6 +1646,15 @@ string_join(PyStringObject *self, PyObject *orig) | |||
| * original sequence can be iterated over | |||
| * again, so we must pass seq here. | |||
| */ | |||
| if (!PyString_Check(item) && !item->ob_bstate == NULL) { | |||
There was a problem hiding this comment.
What I now realise is we probably need a Py_GetBState(...) function that returns (say) -1 for "not string/unicode" and then 0/1/2/3/whatever for the actual bstate. That way we can avoid putting bstate on every object.
There was a problem hiding this comment.
Today I started to add Py_GetBState, I broke other tests that I am still fixing.
|
Lets figure out tracking in before we get back here |
Tracking for new and join operations, applied to both unicode and bytes.
This tracking breaks ordinal configurations for unicode, so before I reconfigure, let us confirm with this PR if this is the intended design from our discussion of the minimal example in the google doc on the string assumptions.