Closed ZenGround0 closed 2 years ago
i'm surprised that the butterflynet was able to handle a PRU message if that's the case (unless it's just in the API or some such)
I am also surprised, I was assuming this would break butterfly. Then again if it was launched with the feat/snap-deals branch of lotus then this change was getting pulled in. feat/snap-deals updated actors v7 to point to here.
Where is this actually getting used? I didn't see any of the FFI calls changing around WinningPoSt...
Yes no changes to the ffi call. The problem is the layer above. It needs to be able to select between sealed copy and updated replica when reading on disk files to start satisfying post challenges for generating both winning and window post. This new sector info data type provides the info needed to select the correct file.
Can you link me to where the assumption about pointerness lives in Lotus? I think butterfly was updated before this commit existed, so we have some figuring out to do.
Here it is: https://github.com/filecoin-project/lotus/blob/master/chain/stmgr/utils.go#L37. I think the butterfly mystery can be resolved by either 1) we built butterfly off of the feat/snap-deals branch which linked to a pointer return value, or 2) this message never succeeded on chain in butterfly.
Merging #1547 (167118a) into master (2d5d491) will not change coverage. The diff coverage is
0.0%
.:exclamation: Current head 167118a differs from pull request most recent head 6148c0e. Consider uploading reports for the commit 6148c0e to get more accurate results
@@ Coverage Diff @@ ## master #1547 +/- ## ====================================== Coverage 69.6% 69.6% ====================================== Files 72 72 Lines 8700 8700 ====================================== Hits 6060 6060 Misses 1780 1780 Partials 860 860
Needed for snap deals integration