msggen: fix walk through nested json schemas

[daywalker90]

I finally found (what i suspected in the first place) the bug in msggen where it wasn't walking through the if's correctly (you may have noticed the weird "x doesn't have a type" logs.

Fixes: #7185 #5345 #5354
Replaces: #7210

[cdecker]

Looks good, but I'd like the if statement to be clarified a bit, as the precedence of the conditions is not clear, and left-to-right order is even a tautology.

[daywalker90]

need #7217 merged first

I had to merge some arrays so it got more complicated :/

[daywalker90]

I WILL BRING ORDER TO THIS AAAAHHH

I'm honestly not sure why i lost the deterministic ordering of the fields of all the generated code but now its nice and sorted.

[daywalker90]

bip340sig is now String type in rust and .proto, same as signature

[daywalker90]

We derived the i32 from rust enums implicitly by their order in the enum. I added explicit numbers in #7217 so the sorting doesn't mess anything up.

Btw if we ever delete an "open" enum field in *.proto with 0 it's going to be trouble. They must start with a field that is 0. But that's for a future PR.

[cdecker]

Btw if we ever delete an "open" enum field in *.proto with 0 it's going to be trouble. They must start with a field that is 0. But that's for a future PR.

We might want to keep dummy placeholders in there once we do. But we'll cross that bridge once we get to it.

I don't fully understand why sorting is required here, does it have to do with the iteration through nested fields? Otherwise it feels strange to have enums not be in numeric order, but no real issue, I'm just curious :-)

[daywalker90]

Yes when merging the arrays that were previously not collected we would get enums in node.proto that were out of order. And protoc wants them to start at 0. The ordering of enums in the rust files could be removed eagain after the explicit enum values were added. I don't remember exactly why i sorted those to be honest. Want me to check that out?

[daywalker90]

rebased and rust model enums sorted by number not name

[cdecker]

Rebased on top of master after merging #7217, and merging as soon as it passes CI (the previous CI failure was a flake).

[daywalker90]

i have a bunch more coming but i wait until this and #7230 is merged

[cdecker]

ACK c93d955