Testing infrastructure to ensure data-availability

[NiloCK]

Why: Problem description :confused:

From #3739:

Previous claim workflow (circa May 19, asset-holder era)

(4f0def9fe640dceeb778dc66bb03a9d491077dbc, chosen randomly)

• call claim using the guarantor channel Id, guarantee bytes, allocationBytes from the target channel, and target payee indices from the target channel
• claim would:
  • calculate a new outcome for the target channel by applying its holdings and priority order against the target channel outcome in order to construct an array of payouts
  • update the target channel outcome (or delete it if the channel outcome is all zeros)
  • execute these payouts (either as transfers to addresses or as holdings bumps to channels
  • emit AllocationUpdated(guarantorChannelId, guarantorChannelInitialHoldigs)

This is a slight abuse of notation! The guarantor channel's allocation (ie, the guarantee) was not actually changed along the way - only its holdings. In any case, the emitted event provides a listening wallet with:

• the id and previous holdings of the guarantor channel
• the target outcomeBytes at the time of the claim operation (from calldata)
• the payout indices (from calldata)

... which is enough for the wallet to calculate the updated outcome at the target channel using the same logic as in _claim.

:exclamation: :question: If Alice is calling claim via the Alice-Ingrid guarantor channel, how is Bob notified? Catching this emitted event requires that Bob can recognize the channelId of the A-I guarantor channel, but he cannot compute this id independently. Is there a prefund step that involves exchange of guarantor channel ids?

What: Describe the solution you'd like :sparkles:

This particular issue is out-of-date, as the current claim implementation emits from both the source and target channels. Broader question is about what sort of automated setup could / should exist to ensure that less-than-fully-cooperative parties all have the information they need to progress their channels & execute funds recovery on chain.

Generally, our current tests orchestrate actions performed by different participants / wallets. A more realistic approach is for wallets to interact from some form of information silo, so we can be assured that individual wallets have (or can get) all the information they need!

[andrewgordstewart]

❗ ❓ If Alice is calling claim via the Alice-Ingrid guarantor channel, how is Bob notified? Catching this emitted event requires that Bob can recognize the channelId of the A-I guarantor channel, but he cannot compute this id independently. Is there a prefund step that involves exchange of guarantor channel ids?

There is not. (You could exchange the guarantor channel ids in the protocol, but how does Bob know that Alice & Ingrid are telling the truth?)