chore!: unify miner, market and verifreg builtin wrappers to v16#12960
chore!: unify miner, market and verifreg builtin wrappers to v16#12960
Conversation
chore: bump ffi to v1.32.0-rc2
chore: update actors bundle to v16.0.0-rc2
Changes some APIs exposed in the miner actor wrapper.
| market9 "github.com/filecoin-project/go-state-types/builtin/v16/market" | ||
| miner9 "github.com/filecoin-project/go-state-types/builtin/v16/miner" | ||
| verifreg9 "github.com/filecoin-project/go-state-types/builtin/v16/verifreg" |
There was a problem hiding this comment.
I didn't rename these because they need to be checked first, this is for nv17 invariant checks and I'm not sure it's OK to do this
3bf4c3f to
e0c7149
Compare
|
Yeah, let's punt this for now. But I think the right way to do this is to alias the state types in the abstraction (chain/actors/builtin/...) and avoid importing |
|
Agreed, but last time I proposed doing something like that it involved pulling in a lot more types than we reference and I got push-back. But I'm happy to go further than this. |
eadab96 to
286bb13
Compare
72f5d6f to
e432ffb
Compare
Just flagging that this will also cause pain in boost, curio, cidgravity-gateway, and probably many other projects which use the current aliases. Really that situation is quite unfortunate, and I'd also like to one day fix this, but I wouldn't do this without very strong motivation. |
#12932 (comment)
This is the cost of unifying the miner types, it has flow-on effects for market and verifreg. It cleans some things up and there's probably a lot of opportunities in here to do more disentanglement (I haven't, this is just a quick rework of all the broken bits).
Is it worth it?