On 26 April 2017 at 02:56, Guido van Rossum <gvanrossum@gmail.com> wrote:
In the end I agree with the OP that we should fix this. I don't see a reason to require a PEP or require updating whatever PEP described this behavior originally -- PEPs generally describe what should be done to a specific version of Python, they don't prevent future alterations, and they essentially represent the historical record, not current documentation.
I'm a little worried about breaking existing code, but only a little bit, and this is clearly a gray area, so I think it's okay to change in 3.7 without deprecations. (But I've been overruled on such matters before, so if you disagree, speak up now and show us your code!)
This is really obscure behaviour to be relying on, so a porting note + the 3.7 pre-release testing cycles seems like sufficient notice to me. It's potentially also worth checking how PyPy handles these cases - for the only other similar case I'm aware of (the quirks with the relative priority of the nb_* and sq_* slots at the C layer), enough projects relied on the CPython behaviour for them to decide to replicate it, so if they *haven't* replicated the quirk described in the OP, it's a solid data point suggesting there aren't a lot of major projects relying on it. Cheers, Nick. -- Nick Coghlan | ncoghlan@gmail.com | Brisbane, Australia