Did you consider using a PyStructSequence instead? It might be a better fit for C code.

I'm not sure about issue itself, see #142595 (comment)

But as an excuse for refactoring this sounds much better. I'll rework the pr.

Hmm, no. I don't think using PyStructSequence is a good idea.

We can't create this type just with PyStructSequence_NewType. The namedtype DecimalType supports three arguments, while new type - only one (a sequence). Probably, I can override tp_new, but at the end - things looks more complex with all this, not simpler.

Thanks @skirpichev for the PR, and @encukou for merging it 🌮🎉.. I'm working now to backport this PR to: 3.13.
🐍🍒⛏🤖

Thanks @skirpichev for the PR, and @encukou for merging it 🌮🎉.. I'm working now to backport this PR to: 3.14.
🐍🍒⛏🤖

GH-142622 is a backport of this pull request to the 3.13 branch.

GH-142623 is a backport of this pull request to the 3.14 branch.

In general, it might be good to add a way to create PyStructSequences that are more namedtuple-like. But, yeah, I agree it's too much work for this issue.

I don't think that the amount of work is the problem: it seems, that interfaces are different. The namedtype has additional class method _make() with signature of PyStructSequences constructor.

Perhaps, some PyStructSequence_NewType2(), that creates a type with the right constructor? If not, I think we should just document the difference. I was certainly misguided by sentence:

Struct sequence objects are the C equivalent of namedtuple() objects

Yeah, that should reference the glossary term, not collections.namedtuple specifically.
Doc fix here: #142626

Perhaps, some PyStructSequence_NewType2(), that creates a type with the right constructor?

Yeah, this, with a new versioned variant of PyStructSequence_Desc rather than hardcoding the different behaviour.