[NFC][Diagnostics]: remove a number of unused diagnostics #85282
Conversation
|
@swift-ci please smoke test |
| NOTE(scanner_cycle_source_target_shadow_module, none, | ||
| "source target '%0' shadowing a%select{ |n SDK }2Swift module with the same name at: '%1'", (StringRef, StringRef, bool)) | ||
|
|
||
| ERROR(scanner_arguments_invalid, none, |
| "unknown library level '%0', " | ||
| "expected one of 'api', 'spi', 'ipi', or 'other'", (StringRef)) | ||
|
|
||
| ERROR(error_old_spi_only_import_unsupported, none, |
| ERROR(error_create_symbolic_interfaces_dir,none, | ||
| "creating symbolic interfaces directory: %0", (StringRef)) | ||
| ERROR(error_symbolic_interfaces_failed_status_check,none, | ||
| "failed file status check: %0", (StringRef)) | ||
| ERROR(error_write_symbolic_interface,none, | ||
| "writing symbolic interface file: %0", (StringRef)) | ||
| REMARK(remark_emitting_symbolic_interface_module,none, |
| ERROR(error_formatting_multiple_file_ranges,none, | ||
| "file ranges don't support multiple input files", ()) | ||
|
|
||
| ERROR(error_formatting_invalid_range,none, | ||
| "file range is invalid", ()) |
| // Caching related diagnostics | ||
| GROUPED_ERROR(error_caching_no_cas_fs, CompilationCaching, none, | ||
| "caching is enabled without CAS file-system options, input is not immutable", ()) | ||
| ERROR(error_prefix_mapping, none, "cannot create scanner prefix mapping: '%0'", (StringRef)) |
There was a problem hiding this comment.
last use: 0384347
- TODO: verify removal – cc @sina-mahdavi could you confirm whether it is expected that this is unused please?
There was a problem hiding this comment.
edit: looks like the error was introduced here, but the error handling function it was for (llvm::MappedPrefix::transformJoined) is no longer used, so seems reasonable to remove.
| "transitive dependency on '%0'", | ||
| (StringRef, Identifier, unsigned)) | ||
|
|
||
| REMARK(explicit_interface_build_skipped,none, |
| "protocol %0 can not yet be represented in C++", (ValueDecl *)) | ||
| ERROR(expose_move_only_to_cxx,none, | ||
| "noncopyable %kind0 can not yet be represented in C++", (ValueDecl *)) | ||
| ERROR(expose_nested_type_to_cxx,none, |
| "actor boundary", | ||
| (Type, const ValueDecl *, ActorIsolation)) | ||
|
|
||
| ERROR(non_sendable_call_result_type,none, |
| ERROR(invalid_ownership_not_optional,none, | ||
| "%0 variable should have optional type %1", | ||
| (ReferenceOwnership, Type)) | ||
| ERROR(invalid_ownership_is_let,none, |
There was a problem hiding this comment.
last use: 89f8f8b
- TODO: verify removal – cc @nickolas-pohilets i see this was recently(ish) removed for the 'weak let' feature, rather than becoming conditional on the presence of that feature itself; is that expected/intentional?
There was a problem hiding this comment.
edit: upon further inspection, i think this was the desired behavior – weak let is now unconditionally supported, but the mutability of weak captures is the thing controlled via the upcoming feature flag. thus, the error is no longer applicable. do correct me if i'm mistaken.
| none, "'@_noImplicitCopy' attribute can only be applied to local lets and params", ()) | ||
| ERROR(noimplicitcopy_attr_not_allowed_on_moveonlytype,none, | ||
| "'@_noImplicitCopy' has no effect when applied to a noncopyable type", ()) | ||
| ERROR(noncopyable_types_cannot_be_resilient, none, |
jamieQ
requested review from
CodaFi,
DougGregor,
artemcm,
bnbarham,
hamishknight,
hborla,
rintaro,
slavapestov,
tshortli and
xedin
as code owners
|
@swift-ci please clean smoke test macos |
|
@artemcm thank you for the review! i've now convinced myself all of these seem reasonable to remove – idk what the protocol is for merging something like this, though i'd prefer it be merged sooner rather than later to reduce the likelihood of conflicts if possible. if that seems alright to you, please merge at your convenience. 🙇 |
2 participants
did a pass through the diagnostics to identify potentially-unused ones. criteria for inclusion was roughly: