Also:
- make sure to apply diversions when determining offsets for code
relaxation.
- select the smallest encoding when selecting a relaxed branch
instruction.
There might be a silent bug in the Python module which directly reads
from this type_set field; in particular, it does so when reading all
controlling type variables, which all seem to be free variables (i.e. no
parent typevar). So imitate this behavior here, until we're sure there
are no other meta generators that rely on this.
Fixes two issues:
- name of derived type variables wasn't correct. (This is used during
code generation.)
- the TypeVar::derived() function wasn't creating the correct type set
(and would instead propagate the parent's one).
Avoid needlessly copying data from wasm into the host for output
parameters, and factor out the `.unwrap()` for translating pointers
when writing to output parameters.
Whether the downsides in POSIX and existing application compatibility
outweigh the benefits of thread safety remains an open question.
Right now, this note is just documenting the current behavior.
- Adds a compiler warning when the fmtln! macro isn't correctly used.
- Allow to add an empty line.
- Make the output of generated matches more beautiful, by having one
struct per line on the clause.
- Add a method to add match with doesn't read any data from fields.
- Make sure that the placeholder clause of a match is put at the end.
- Both the `wasm` and `compile` commands get this new subcommand, and it defaults to false. This means that test runs with `wasm` can request disassembly (the main reason I am doing this) while test runs with `compile` now must request it, this changes current behavior.
- Switch to using context.compile_and_emit directly, and make the reloc and trap printers just accumulate output, not print it. This allows us to factor the printing code into the disasm module.
