Add Data.Constr casing to Case - #7914
Open
SeungheonOh wants to merge 12 commits into
Open
Conversation
SeungheonOh
force-pushed
the
sho/caseDataConstr
branch
from
August 20, 2026 10:04
67382f3 to
d20120e
Compare
Contributor
SeungheonOh
force-pushed
the
sho/caseDataConstr
branch
from
August 20, 2026 10:07
d20120e to
89ea27d
Compare
zliu41
reviewed
Aug 20, 2026
| {-# OPAQUE wrapUnsafeDataAsConstr #-} | ||
|
|
||
| wrapCaseData :: BuiltinData -> [BuiltinList BuiltinData -> a] -> a | ||
| wrapCaseData = BI.caseData |
Member
There was a problem hiding this comment.
Are you sure we need this? It's possible that neither this nor wrapUnsafeDataAsConstr is needed any more.
If we do, then this needs to be mentioned in the Note below.
Contributor
Author
There was a problem hiding this comment.
If we don't have this, inliner won't inline the bindings it makes for each fields which prevents DeadCase and CollapseCase pass from running. Can't really think of other ways to make this work better
Contributor
Author
There was a problem hiding this comment.
On the side note, this method only works on quite specific case where each handler of the match must be lambda directly like so:
matchRich ...
(\a b c -> ...)
(\x y z k -> ...)
...
if we have non-lambda handler, this doesn't work at all!
let f = \a b c -> ...
in matchRich ...
f
(\x y z k -> ...)
...
All fields gets bounded for f.
Member
There was a problem hiding this comment.
We could retire the matcher and try to optimize pattern matching in the plugin. The matcher is already inconvenient to begin with, compared to direct pattern matching. May not be easy to do though.
zliu41
requested changes
Aug 20, 2026
zliu41
left a comment
zliu41
left a comment
Member
There was a problem hiding this comment.
It otherwise looks good, but this new feature must be gated by dijkstraPV, and I don't think it is gated properly.
Contributor
Author
|
Do we need this to be gated by dijkstra? I don't think adding this to older PV with builtin casing would have any issue. What do you think? |
Member
Yes otherwise old node version rejects casing on Constr, new node version accepts it. |
zliu41
requested changes
Aug 25, 2026
| -> HeadSpine Text term (Some (ValueOf DefaultUni)) | ||
| caseBuiltinDataUnavailable someVal@(Some (ValueOf uni x)) branches = case uni of | ||
| DefaultUniUnit | ||
| | 1 == len -> HeadOnly $ branches Vector.! 0 |
Member
There was a problem hiding this comment.
I don't think you need to duplicate the code for each branch. Can you not call caseBuiltin here?
Contributor
Author
There was a problem hiding this comment.
This would make it slower! aruond 5~10%
Member
There was a problem hiding this comment.
Can you try some refactoring, such as defining a caseBuiltinForPV that takes PV as an argument? Explicit code duplication should rarely be needed for efficiency, and should be the last resort.
Member
|
You may also want to check if any of the conformance tests need to be updated. |
Member
|
Do we have a benchmark that covers casing on constants? If so can you run it on this PR to ensure there isn't regression. |
Contributor
Author
|
/benchmark validation |
Contributor
Author
|
/benchmark casing |
Contributor
|
Click here to check the status of your benchmark. |
Contributor
|
Comparing benchmark results of 'validation' on 'f424a79ad' (base) and '894db473c' (PR) Results table
|
Contributor
|
Click here to check the status of your benchmark. |
Contributor
|
Comparing benchmark results of 'casing' on 'f424a79ad' (base) and '894db473c' (PR) Results table
|
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
I discarded this idea originally as most of the cost would've been in the deconstruction of fields(
xshere). However, introduction ofDropListmade deconstruction ofxsreasonable fast so this approach actually gives a decent performance boost with minimal change!