By changing the indexer, we change the layout, without touching existing user code, and without having to modify structs and (consequently) struct accesses across an entire codebase.

My current implementation is presumably similar to any struct parser / compiler setup. We start to crystallise the struct, in order to compact bitfields, pad and align to key widths in memory. (This begins to take away some control from indexers, which may be part of my problem.)

By changing the indexer, we change the layout, without touching existing user code, and without having to modify structs and (consequently) struct accesses across an entire codebase, since our accessors / indexers are type-agnostic.

My current implementation is presumably similar to any struct parser / compiler setup. We start to crystallise the struct, compacting bitfields, padding and aligning to key memory widths. (This begins to take away some control from indexers, which may be part of my problem.)

This is very similar presumably to any struct parser / compiler implementation.

My current implementation is presumably similar to any struct parser / compiler setup. We start to crystallise the struct, in order to compact bitfields, pad and align to key widths in memory. (This begins to take away some control from indexers, which may be part of my problem.)