RFC-0039: Types come second

RFC-0039: Types come second
StatusRejected
Areas
  • FIDL
Description

We propose to allow inline declarations of structs, xunions, and tables; flip the order in which field names (or argument and return names) appear with respect to types, specifically to have the type appear after names.

Authors
Date submitted (year-month-day)2019-03-07

"You're Go-ing to love it"

Rejection rationale

When this proposal was drafted, and socialized, there was a strong consensus to consider syntax changes all at once, rather than one at a time (see also RFC-0038). We also wanted one person to be the syntax arbiter, rather than risk designing by committee.

Eventually, this proposal was obsoleted by RFC-0050 which met both conditions sought.

Summary

We propose to:

  • Allow inline declarations of structs, unions, and tables;
  • Flip the order in which field names (or argument/return names) appear with respect to types, specifically to have the type appear after names.

(We stop short of introducing anonymous declarations, since we would likely want improved bindings support to ensure the ergonomics are good.)

Motivation

Quickly:

  • We're starting to see more patterns where combination of various declarations to describe 'one conceptual message' is routine. For instance:
    • Container struct, whose last field is a table (initially empty) to leave the door open to extensions.
    • Container union, where variants are tables to have flexibility.
    • Container table, where fields are grouped in structs, and ordinals loosely match 'version numbers'.
  • Additionally, support for empty struct, unions, and tables offers the low-level pieces to build Algebraic Data Type support from (from a layout standpoint, not bindings).
  • All of these use cases are pushing us towards allowing inline declarations.
  • With inline declarations, it is easier to read the field name first, then have a type description, which could straddle multiple lines. See examples below.

Design

Some examples:

  • Simple struct or table:

    struct Name {
        field int32;
    };
    
    table Name {
        1: field int32;
    };
    
  • Protocols:

    protocol Name {
        Method(arg int32) -> (ret int32);
    };
    
  • Struct with extension:

    struct Name {
        field1 T1;
        field2 T2;
        ...;
        ext table NameExt {};
    };
    
  • Union variants:

    union Name {
        variant1 table NameVariant1 {
            ...
        };
        variant2 table NameVariant2 {
            ...
        };
        ...
    };
    
  • Grouped fields by version:

    table Name {
        1: v1 struct NameGroupV1 {
            ...
        };
        2: v2 struct NameGroupV2 {
            ...
        };
        ...
    };
    

Notes:

  • Scoping wise, while we would consider all declaration names to be top-level (and hence enforce uniqueness on a per-library basis), we would not allow inline declarations from being referenced, i.e. only single use.

Ergonomics

This proposal improves ergonomics by conveying ABI implications to developers through syntax.

Documentation and examples

At least:

Backwards Compatibility

This is not source level backwards compatible.