Prefer Operation instead of SyncOp in tests
#431
Conversation
This is a test-only change: all modified code is `#[cfg(test)]` Lots of tests used `SyncOp`, including some proptests. One of those is moved out to `tests/syncing-proptest.rs` and updated to use Replicas intead of TaskDb's. `op.rs` remains in terms of `SyncOp`, since it defines that type, but converts to `Operation` for tests. With their uses in tests removed, two `apply_..` methods can finally be deleted.
|
The one failing action here is #432 and not specific to this PR. |
| assert_eq!(t0, t1); | ||
| assert_eq!(t1, t2); | ||
| } | ||
| } |
There was a problem hiding this comment.
Just out of curiosity, am I right to interpret the word "strategy" as "random choice" in the context of proptest? They use it pervasively in their docs and I'm not sure I know what they mean by it.
There was a problem hiding this comment.
Yeah, but it's not quite random - proptest has a bunch of mechanisms to help it focus down on problematic test cases, so it's more like a transformation from a bit string to a test case, where that bit string it built in "smart" ways.
|
Yeah, but it's not quite random - proptest has a bunch of mechanisms to help it focus down on problematic test cases, so it's more like a transformation from a bit string to a test case, where that bit string it built in "smart" ways. |
This is a test-only change: all modified code is
#[cfg(test)]Lots of tests used
SyncOp, including some proptests. One of those is moved out totests/syncing-proptest.rsand updated to use Replicas intead of TaskDb's.op.rsremains in terms ofSyncOp, since it defines that type, but converts toOperationfor tests.With their uses in tests removed, two
apply_..methods can finally be deleted.This is the last "boring" change in #372. The next two will be breaking changes and have some open questions!