On 08/02/2020 11:49 am, Nick Coghlan wrote:
Unfortunately, the simplifications you propose would be backwards incompatible - it's existing behaviour that there's a real shared dict (even on optimised frames) where arbitrary extra attributes can be stored (even though they don't become accessible as Python variables). I don't want to make frame objects any bigger than they already are, so the natural solution is to store the mapping proxy as `f_locals`, and then bypass the proxy in order to make `PyEval_GetLocals` still "work" (at least as well as it ever did).
The proposed changes in PEP 558 are also backwards incompatible. I thought that was the point. The current implementation is broken in weird ways and we want to fix that. Since we need to break backward compatibility anyway, why not do it in a way the makes the behaviour as well defined and maintainable as possible.
I think that PEP 558, as it stands, is still a bit fragile because of the handling of cycles between the locals proxy and the frame.
PyObject_GetAttr(string) also doesn't do that same thing as the proposed C functions, since it invokes the Python descriptor machinery. (Note that the discussion at https://discuss.python.org/t/pep-558-defined-semantics-for-locals/2936/ is more up to date than the PEP text where the C API is concerned)
`PyObject_GetAttr("attr")` has the same semantics as the Python operator `x.attr` which is under the control of `type(x)`, in this case the frame object class. The descriptor protocol is irrelevant.
The reference to tracing mode dependent semantics puzzles me, as that was removed in December: https://github.com/python/peps/commit/54888058ce8ad5257114652d9b41e8d1237b8e...
That was my misreading. The behaviour of `f_locals` in the PEP is not very clear, as it is buried in the discussion of CPython changes. Could you add it to the proposal section?
On Wed, 22 Jan 2020 at 21:59, Mark Shannon firstname.lastname@example.org wrote:
First of all I want to say that I'm very much in favour of the general idea behind PEP 558. Defined semantics is always better than undefined semantics :)
However, I think there are a few changes needed:
Don't add anything to the C API, please. Frame attributes can be accessed via `PyObject_GetAttr[String]`.
Don't make the behaviour dependent on whether "tracing" is turned on. Doing so forces debuggers to use sys.settrace which is horribly
slow. It also makes the implementation more complex, and has no benefit AFAICT.
- Don't store write-through proxies in the frame, but make proxies
retain a reference to the frame. This would reduce the size and complexity of code for handling frames. Clean up of the frame would occur naturally via reference count when all proxies have been reclaimed.
The proposed implementation is hard to reason about and I am not confident that it will not introduce some new subtle bugs to replace the ones it seeks to remove. Any implementation that has functions with "Borrow" and "BreakCycles" in their names makes me nervous.
A simpler conceptual model, which I believe could be made reliable, would be:
No change for non-function frames (as PEP 558 currently proposes).
Each access to `frame.f_locals` (for function frames) should return a new proxy object.
This proxy object would have dict-like, write-through semantics for variables in the frame. For keys that do not match variable names, an exception would be raised. This means that all proxies for a single frame will have value equivalence; object equivalence is not needed. I.e. for a frame `f`, `f.f_locals == f.f_locals` would be True, even though `f.f_locals is f.f_locals` would be False.
Cheers, Mark. _______________________________________________ Python-Dev mailing list -- email@example.com To unsubscribe send an email to firstname.lastname@example.org https://mail.python.org/mailman3/lists/python-dev.python.org/ Message archived at https://email@example.com/message/TSHCB4ZH... Code of Conduct: http://python.org/psf/codeofconduct/