Allow literal dots in instruction mnemonics?

[CSharperMantle]

Background and motivation. I am working on contributing a LoongArch64¹ decoder to this project. Some instructions in LA64 contain underscores (_) in their mnemonics, such as the AMSWAP_DB.W atomic memory access instruction². In the current approach, underscores are an "escaped" version of dots (.) and are replaced with dots before display. This workaround will break mnemonics with underscores.

instruction-decoder/src/lib.rs

Line 845 in 8894eaa

let formatted_name = self.name.replace('_', ".");

Possible solution. I can see these possible solutions to this problem.

• Introduce yet another escape sequence (perhaps double underscore __) to represent a single underscore.
  • Pros
    • No changes to current decoding tables are needed.
    • No visible API/output changes.
  • Cons
    • More complexity in parsing.
    • Not utilizing TOML's quoted key feature³.
• Quote mnemonics when necessary so that literal dots can be represented. See Allow dot symbols in insn mnemonics by quoting #32 for an implementation.
  • Pros
    • Better scalability.
    • Improved parser simplicity.
    • Conforms to TOML specs.
    • No visible API/output changes.
  • Cons
    • Major changes to existing decoding tables.
• Do nothing.
  • Pros
    • Zero changes.
  • Cons
    • Please, don't do this. 😉

Which approach can benefit this project most?

Footnotes

1. https://docs.kernel.org/arch/loongarch/introduction.html ↩

2. https://github.com/loongson/LoongArch-Documentation/blob/e0d6592229d9e00e512bd28b688a7f20b171714f/docs/LoongArch-Vol1-EN/basic-integer-instructions/overview-of-basic-integer-instructions/atomic-memory-access-instructions.adoc ↩

3. https://toml.io/en/v1.0.0#keys ↩

[FRoith]

Great idea!

When I originally created the concept, I was not yet aware of quoted keys, and just hacked something together, this clearly is the way to do it, and the existing tables are relatively easily adjusted.