Integrate specs-actors v3 #5363
Conversation
chain/stmgr/utils.go
Outdated
| mid, err := address.IDFromAddress(maddr) | ||
| if err != nil { | ||
| return nil, xerrors.Errorf("getting miner ID: %w", err) | ||
| } | ||
|
|
||
| ids, err := pv.GenerateWinningPoStSectorChallenge(ctx, wpt, abi.ActorID(mid), rand, numProvSect) | ||
| // TODO: move this to somewhere in specs-actors or the state types. |
There was a problem hiding this comment.
cc @anorth. Will the winning proof always be paired 1:1 with the window post proof? If so, we should have a helper function to convert one to the other (well, ideally, we'd just get rid of the difference...).
The tricky part is that we're using the same type for winning and window proofs.
There was a problem hiding this comment.
As far as I know they will be paired 1:1, but I don't know the scope of possible futures. You need @porcuquine.
Winning PoSt has nothing to do with actors. The WinningPoSt proof types are declared there only for proximity to the other types, but arguably should be removed. It's unfortunate they use the same type, yes. We could probably make a change in Lotus in the near future to stop referencing the WinningPoSt constants altogether (by duplicating locally), opening up a path to removing them from there.
I think a mapping function in Lotus is the best immediate solution.
|
Devnet now runs past the upgrade epoch |
| @@ -704,6 +701,73 @@ func UpgradeCalico(ctx context.Context, sm *StateManager, cb ExecCallback, root | |||
| return newRoot, nil | |||
| } | |||
|
|
|||
| func UpgradeActorsV3(ctx context.Context, sm *StateManager, cb ExecCallback, root cid.Cid, epoch abi.ChainEpoch, ts *types.TipSet) (cid.Cid, error) { | |||
| buf := bufbstore.NewTieredBstore(sm.cs.Blockstore(), bstore.NewTemporarySync()) | |||
There was a problem hiding this comment.
Need to check how much memory this uses on mainnet
There was a problem hiding this comment.
What's the best way to test this? compute-state on the live network? Or just set a node to run the upgrade early?
Stebalien
force-pushed
the
asr/specs-update
branch
from
5eeeef5
to
6cbc978
Compare
Different maps have different parameters now so we just construct/load them manually where needed.
Well, not fully, but this is enough to ensure we use the correct bitwidth and such.
Stebalien
force-pushed
the
asr/specs-update
branch
from
b9c3778
to
e8ecfed
Compare
Stebalien
force-pushed
the
asr/specs-update
branch
2 times, most recently
from
d0413f7
to
032ebdc
Compare
It's causing the tests to flake.
Stebalien
force-pushed
the
asr/specs-update
branch
from
032ebdc
to
d412ccd
Compare
It's kind of slow.
As in the title. Needs:
nextin a separate PR)Fixes #5363