Feature record update#337
Conversation
| it "works with RecordUpdate" $ do | ||
| uses (named "f") (RecordUpdate (Reference "f") [("g", Reference "h")]) `shouldBe` True | ||
| uses (named "h") (RecordUpdate (Reference "f") [("g", Reference "h")]) `shouldBe` True | ||
| uses (named "g") (RecordUpdate (Reference "f") [("g", Reference "h")]) `shouldBe` False |
There was a problem hiding this comment.
Not married to this. See description.
There was a problem hiding this comment.
I am unsure, too.
I would take the same path as with assignments. If we are considering assignment as usage, then should we consider record update's assignee as a usage too
flbulgarelli
left a comment
flbulgarelli
left a comment
There was a problem hiding this comment.
It looks quite good. My main concerns are about the visitor module: we should at least update the documentation of the visit pattern.
However, I am unsure about the motivation of this feature. Are we actually going to introduce new expectations for this RecordUpdate? Or do we just need the inner contents of the record update to be parsed? May an Other here also solve the problem?
| it "works with RecordUpdate" $ do | ||
| uses (named "f") (RecordUpdate (Reference "f") [("g", Reference "h")]) `shouldBe` True | ||
| uses (named "h") (RecordUpdate (Reference "f") [("g", Reference "h")]) `shouldBe` True | ||
| uses (named "g") (RecordUpdate (Reference "f") [("g", Reference "h")]) `shouldBe` False |
There was a problem hiding this comment.
I am unsure, too.
I would take the same path as with assignments. If we are considering assignment as usage, then should we consider record update's assignee as a usage too
| normalizeSwitchCases ops = map (\(e1, e2) -> (normalize ops e1, normalize ops e2)) | ||
| normalizeRecordUpdates ops = map (\(i, e) -> (i, normalize ops e)) | ||
| normalizeTryCases ops = map (\(p, e) -> (p, normalize ops e)) | ||
| normalizeSwitchCases ops = map (\(e1, e2) -> (normalize ops e1, normalize ops e2)) |
There was a problem hiding this comment.
Perhaps we could generalize those tree functions now. The first two are the same, and are mostly the same in all files.
Agree |
I am unsure about this one, too. Do we have any other examples in other languages? In any case, rename and replace don't currently target record's field names. |
I... I forgot |
2 participants
Fixes #336.
Thought of using
FieldAssignmentat first but got dissuaded by the fact that you can update multiple fields per record update.I didn't use an expression for the update's field name as it's not quite the same as other
References (i.e. it cannot be parameterized). That being said, in retrospective, that might be too strongly tied to haskell's way of doing it? How is this supposed to work with rename and replace? Not so sure about this one!