When debug specifier _and_ format_spec are present, use !s as the conversion default

[davepeck]

Currently, t"{1=:foo}".interpolations[0] is Interpolation(1, '1', None, 'foo').

The section of PEP750 on the debug specifier says it should be Interpolation(1, '1', 's', 'foo').

Assuming we think the spec is right, here's a fix. (Otherwise, LMK and I'll fix the PEP.)

[lysnikolaou]

I think that's a spec issue. Having !s as the default would be a departure from what f-strings do, which is that they provide no default when there's a format_spec so that __format__ is called.

I'm wondering now how we can change the spec with the PEP having been accepted.

[davepeck]

I'm wondering now how we can change the spec with the PEP having been accepted.

I’ll open a PR and find out. 😅

Annoyed at myself for not catching this earlier — I've been trying to go over spec vs implementation in detail and this is the first thing I’ve run across with the spec. Luckily it’s pretty minor.

[AA-Turner]

I'm wondering now how we can change the spec with the PEP having been accepted.

The intent is that "Accepted" is different from "Final" to allow for changes made during the implementation of the feature

[lysnikolaou]

The intent is that "Accepted" is different from "Final" to allow for changes made during the implementation of the feature

Oh got it. Thanks for the clarification!

[davepeck]

Closing without merge in favor of the PR on PEP750