[DRAFT] Tweak annotation population to better maintain field definition ordering

[AlecRosenbaum]

Description

This change explores tweaking field annotation population to better maintain field ordering.

As-is (without this change), this is the behavior:

import strawberry

@strawberry.type
class A:
    type_1a: str = "type_1a"
    type_1b: str = strawberry.field()
    type_1c: str = strawberry.field(resolver=lambda: "type_1c")
    
    @strawberry.field
    def type_2(self) -> str:
        return "type_2"

    another_type_1: str
    another_type_1b: str = strawberry.field()
    another_type_1c: str = strawberry.field(resolver=lambda: "type_1c")

print(
    [f.name for f in A.__strawberry_definition__.fields]
)
# ['type_1a', 'type_1b', 'type_1c', 'another_type_1', 'another_type_1b', 'another_type_1c', 'type_2']

Ideally type_2 would be defined in the middle, matching the order of the field on the model.

Exploring potential solutions, it seems like without adding a metaclass or doing AST traversal it's pretty hard to get the actual definition order. In this PR I've tried to approximate it a little better without a major change.

The proposed change here does not work for maintaining ordering of bare annotations defined in the middle of other fields (e.g. another_type_1: str above). However, for other field definition styles it appears to work correctly. So if ordering is important in the schema there's then a simple solution to add = strawberry.field() to otherwise bare annotations.

Types of Changes

• Core
• Bugfix
• New feature
• Enhancement/optimization
• Documentation

Issues Fixed or Closed by This PR

None, question was asked on discord.

Checklist

• My code follows the code style of this project.
• My change requires a change to the documentation.
• I have updated the documentation accordingly.
• I have read the CONTRIBUTING document.
• I have added tests to cover my changes.
• I have tested the changes and verified that they work and don't break anything (as well as I can manage).

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 94.35%. Comparing base (dd8e827) to head (8f8ebb8).

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3561      +/-   ##
==========================================
- Coverage   96.58%   94.35%   -2.23%     
==========================================
  Files         524      520       -4     
  Lines       33632    32484    -1148     
  Branches     5577     3722    -1855     
==========================================
- Hits        32482    30649    -1833     
- Misses        915     1552     +637     
- Partials      235      283      +48     

[codspeed-hq]

CodSpeed Performance Report

Merging #3561 will not alter performance

_{Comparing closeio:field-ordering (8f8ebb8) with main (dd8e827)}

Summary

✅ 13 untouched benchmarks

[erikwrede]

It might be worthwhile re-testing this on python 3.13 once we support it since they overhauled dataclass instantiation quite a bit.