1lab/mikan
13
52
Fork
You've already forked mikan
9

Move to diagnostic codes #99

Open
opened 2026年05月15日 16:18:49 +02:00 by amyliao · 0 comments

The current diagnostic reporting infrastructure requires maintaining by hand a mapping from the TypeError and Warning types into another pair of monster ErrorName and WarningName types. This is used both in reporting (it is shown in brackets) and to control what options should be enabled/disabled/made fatal. While the mapping doesn't have to be injective, this basically blesses the names of the Warning constructors as they were when the Name infrastructure was added as part of the backwards-compatibility surface.

I think that instead, diagnostics should have unique, randomly-assigned codes. These should be what is presented to the user, and should be indexed in the documentation, but we should never ingest them. We can look at how GHC assigns codes for inspiration on how to keep them unique across versions.

To control warnings, we should invent a new WarningFlag type that actually controls what (group of) warnings should be enabled, and which warnings should be errors. These should be coarser than the Warning constructors and ideally also have shorter names. We can then (again looking at GHC for inspiration) write functions TypeError -> DiagnosticReason and Warning -> DiagnosticReason into a type

data DiagnosticReason
 = Error -- ^ Always an error, including "error-warnings".
 | WarningWithFlags (List1 WarningFlag) -- ^ A warning affected by the given flags.

which can then be post-processed by a function PragmaOptions -> DiagnosticReason -> Severity to e.g. set a warning's severity to "ignored" or "error" depending.

The current diagnostic reporting infrastructure requires maintaining *by hand* a mapping from the `TypeError` and `Warning` types into another pair of monster `ErrorName` and `WarningName` types. This is used both in reporting (it is shown in brackets) and to control what options should be enabled/disabled/made fatal. While the mapping doesn't have to be injective, this basically blesses the names of the `Warning` constructors as they were when the `Name` infrastructure was added as part of the backwards-compatibility surface. I think that instead, diagnostics should have unique, randomly-assigned *codes*. These should be what is *presented* to the user, and should be indexed in the documentation, but we should never *ingest* them. [We can look at how GHC assigns codes for inspiration](https://github.com/ghc/ghc/blob/master/compiler/GHC/Types/Error/Codes.hs) on how to keep them unique across versions. To control warnings, we should invent a new `WarningFlag` type that actually controls what (group of) warnings should be enabled, and which warnings should be errors. These should be coarser than the `Warning` constructors and ideally also have shorter names. We can then (again looking at GHC for inspiration) write functions `TypeError -> DiagnosticReason` and `Warning -> DiagnosticReason` into a type ```haskell data DiagnosticReason = Error -- ^ Always an error, including "error-warnings". | WarningWithFlags (List1 WarningFlag) -- ^ A warning affected by the given flags. ``` which can then be post-processed by a function `PragmaOptions -> DiagnosticReason -> Severity` to e.g. set a warning's severity to "ignored" or "error" depending.
Sign in to join this conversation.
No Branch/Tag specified
main
aliao/occurs-opt
aliao/rec-con
fix-license
aliao/oopsie
aliao/strengthen-iapply
No results found.
Labels
Clear labels
1 - scope
language
Backwards-compatible changes to the surface language, such as adding new features.
1 - scope
language change
Breaking changes to a documented part of the surface language. These should be documented and specifically called out in the changelog.
1 - scope
performance
Semantics-preserving performance work.
1 - scope
task
Changes that are not apparent to the user, e.g. developer documentation and refactors, which do not need to be documented.
A: dead-code
Dead code elimination and --save-metas.
A: errors
Reporting of error and warning messages.
A: highlighting
Highlighting during interaction & literate compilation
A: imports
Handling of imports and interface file de/serialization.
A: infra
Iinfrastructure (e.g. CI, developer documentation, the build system, or the repository itself)
A: instance
Instance declarations, arguments, and resolution
A: lhs
LHS checking, coverage checking, and clause compilation
A: modules
Declaration and instantiation of parametrised modules.
A: parallel
Parallel typechecking.
A: positivity
Positivity checking
A: printing
Pretty-printing, display forms, and the wording of diagnostic messages
A: records
Record declarations, record expressions and copattern matching
A: scoping
Scope checking (ConcreteToAbstract)
A: termination
Termination checking
0 - type
bug
Issues describing an incorrect behaviour in the project, typos in the documentation, etc.
0 - type
discussion
An issue raised to gather opinions on a proposed change, or a suggested language change.
0 - type
question
An issue raised to ask a question.
9 - status
duplicate
This issue or pull request already exists.
9 - status
info needed
More information is needed from the submitter to decide how to proceed with the issue report.
9 - status
invalid
The issue report or pull request does not describe an actual bug, or does otherwise does not meet the contributing guidelines.
good first issue
Interested in contributing? Get started here.
upstream
Related to an upstream repository, already reported there
Milestone
Clear milestone
No items
No milestone
Projects
Clear projects
No items
Assignees
Clear assignees
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
1lab/mikan#99
Reference in a new issue
1lab/mikan
No description provided.
Delete branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?