[anxolin]

What

SOL and WSOL resolved to Token-2022's native mint (9pan9bMn…) rather than So111…112. Nothing quotes that mint, so every order selling SOL was unfillable. Both symbols now resolve to the legacy wrapped SOL mint.

Wrapping was also gated on the user typing the literal string sol, so naming the sell token by mint address — the only workaround for the above — skipped the wrap, and the CLI never creates the sell-side account either. That produced a successful transaction and an order over an empty account, which just never filled.

The gate is now what the token names:

Sell token	Behaviour
SOL, or 11111111111111111111111111111111	wraps the shortfall, creating the account if needed
WSOL, or a mint address	spends the balance already held; errors if it is short

The native SOL marker is accepted as a sell token because lamports have no mint on the wire — the sell side can only carry them wrapped, so the CLI does the conversion and reports it. The summary now prints sellMint, plus wrappedLamports when it wrapped.

Test plan

just test && just lint && just fmt-check

I run the following:

❯ cow sell 0.01 SOL for 0.5 EPjFWdd5AufqSSqeM2qN1xzybapC8G4wEGGkZwyTDt1v
      signature = GVm8DP4LvXMjyknjrwN6CWXzdD5aXMtu4wJWEAEWbTzvFC5VQed1tvyHpZqtmodbr7vb8jowBo7LH4gb6ntd8bm
       orderPda = 4GFrNnhXUCczgzevaAfnmC3JqJZ91mK8PPt6oDDexikY
       orderUid = 6639024830e332980b04cb8a845eea4859a62a2f8a21eec3fc1c1ff9503b598d
       sellMint = So11111111111111111111111111111111111111112
wrappedLamports = 10000000

❯ cow buy 0.5 EPjFWdd5AufqSSqeM2qN1xzybapC8G4wEGGkZwyTDt1v for 0.01 SOL
      signature = 2VuxHVAjy14qPtFmJqXp7DnnPcJ3qbDcGQyWirM379LegQrhPRXUFtuiDfGwg3SE3umPrMJXdzygY2sedhgC9vda
       orderPda = B3FaARVjsm1fePUeJLsNuVxC4kVtVuZzPDzrxRBspnGg
       orderUid = 8f166ab4a92ee165b109308b100810ba8d86bf867ff986c7344e7fdbd18e409b
       sellMint = So11111111111111111111111111111111111111112
wrappedLamports = 10000000

[kaze-cow]

overall the changes seem a lot more verbose than actually necessary.

as the most simple solution, couldn't we just update the address of WSOL? example PR

Also, the native sol marker is not universally or low level recognized as 11111111111111111111111111111111, so it seems wierd to explicitly code for it in the test CLI. if we do want to do such a thing

[kaze-cow]

too verbose

also I think this function should be named is_native_sol for clarity that this is effectively a yes/no check. Also, I believe we have an applicable function

[anxolin]

Closing in favour of #216