But isn't this going to make parity-wasm incompatible with v1 wasm?

I believe the intention is that if the section is missing that just means there are no passive data segments which should be compatible with v1 wasm, but as with all new features if used it's incompatible.

I think that compatibility story is the same here, although I could be wrong! I believe though that this PR retains the same level of compatibility as before

I believe that the issue is that your patch introduces DataCount as section 10 and moves Code and Data from 10 to 11 and 11 to 12.

The proposal introduces DataCount as section 12, which does not cause this conflict, but does introduce a new ordering rule about putting DataCount (if present) before the code and data sections in the binary.

The intention was to change the order to have DataCount as section 12, but not the ids. I've run this through wasm-bindgen's test suite and it doesn't break tests, which exercises wasm implementations of Node/Firefox/Chrome, so I'm relatively confident at least that this is not a breaking change