update to filecoin-ffi with shared types #127
Conversation
There was a problem hiding this comment.
LGTM
I do have a non-blocking question: I see we now take a proof type that is hard coded on the client side but configurable no the provider side. What happens if these diverge?
I was also under the impression that part of the goal in having actual CIDs for PieceCIDs was to be able to communicate proof types in the CID itself -- with the hope that we could support additional constructions in the future. However, I dug into the FFI code and it definitely doesn't seem to be doing that. There is a configurable call in go-fil-commcid with a specific hash code that was intended to support multiple proofs. (https://github.com/filecoin-project/go-fil-commcid/blob/master/commcid.go#L95) but that's not being used. Anyway this has very little to do with this PR and is probably a @laser question more than anything.
|
oh probably run mod tidy as well. |
storagemarket/impl/client_utils.go
Outdated
| ssb := builder.NewSelectorSpecBuilder(ipldfree.NodeBuilder()) | ||
|
|
||
| // entire DAG selector | ||
| allSelector := ssb.ExploreRecursive(selector.RecursionLimitNone(), | ||
| ssb.ExploreAll(ssb.ExploreRecursiveEdge())).Node() | ||
|
|
||
| commp, paddedSize, err := c.pio.GeneratePieceCommitment(root, allSelector) | ||
| commp, paddedSize, err := c.pio.GeneratePieceCommitment(abi.RegisteredProof_StackedDRG2KiBPoSt, root, allSelector) |
There was a problem hiding this comment.
@why this is where I see the hard coding in the storage client
No description provided.