Support for [@error_message] attribute on Pexp_constraint #1954

alanechang · 2023-10-20T12:55:18Z

This PR follows #1943 and adds support for the [@error_message] attribute on Pexp_constraint.

It enables custom error messages on type mismatches:

let f (x : bool) = (x : int)[@error_message "custom message"]
[%%expect{|
Line 1, characters 20-21:
1 | let f (x : bool) = (x : int)[@error_message "custom message"]
                        ^
Error: This expression has type bool but an expression was expected of type
         int
       because [@error_message]: custom message
|}]

It's implemented using the explanation system already in place within typecore.ml. See added tests for its exact behavior.

ncik-roberts · 2023-10-20T17:07:34Z

I volunteered to review.

ncik-roberts · 2023-10-30T18:10:08Z

Reassigning review to Richard as discussed.

ocaml/testsuite/tests/typing-layouts/error_message_attr.ml

ocaml/typing/typecore.ml

goldfirere · 2023-11-21T18:39:30Z

This is ready to merge. With the unusual merge target, I'm leaving it for you to merge @alanechang -- don't want to cause git trouble.

* enable display_histories * accept test changes and mark them as unreviewed * improve CR messages for unreviewed tests --------- Co-authored-by: Alan Chang <alanechang@janestreet.com> Remove duplicated creation_reasons from layout history error messages (ocaml-flambda#1825) * stop storing duplicates in creation_reasons * accept test changes * reduce to only displaying one creation reason * accept test changes * prefer other reasons over Concrete_creation * accept test changes * stop building history tree if flatten is enabled * remove CR version Co-authored-by: Richard Eisenberg <reisenberg@janestreet.com> * Add CR for assumption in combine_histories Co-authored-by: Richard Eisenberg <reisenberg@janestreet.com> --------- Co-authored-by: Alan Chang <alanechang@janestreet.com> Co-authored-by: Richard Eisenberg <reisenberg@janestreet.com> Improve layout history error message formatting and wording (ocaml-flambda#1832) * remove extra whitespace * fix Unequal_var_layouts error fmt * improve message wording * fix spacing * accept test changes * grammar and phrasing * test changes --------- Co-authored-by: Alan Chang <alanechang@janestreet.com> Improve layout history error message: add path field to Type_argument value_creation_reason (ocaml-flambda#1834) * add Path.t to Type_argument * test changes * add Imported_type_argument * Imported_type_argument for imported type params * fix float64 error format * test changes * add more tests * update layout reason not just in failure case * tests with more than one type param * combine Imported_type_argument with Type_argument * add option_argument_layout to predef * track position and arity for Type_argument * test changes * more tests around typa argument arity * have Imported_type_argument as a creation_reason * test changes * add note about performance benchmark * extract position format logic into function * add comments about 1-indexed * change equals to is * test changes * define list_argument_layout * format comments --------- Co-authored-by: Alan Chang <alanechang@janestreet.com> Improve layout history error message: additional info on value layout defaulting (ocaml-flambda#1855) * add message when defaulting to layout value * test changes * relax to all const defaulting * test changes * improve wording * test changes --------- Co-authored-by: alanechang <alanechang@janestreet.com> Improve layout history error message: change message wording and add more tests (ocaml-flambda#1887) * change imported reason wording * test changes * lowercase * test changes * add more tests * change wording for match * update Immediate_polymorphic_variant message * test changes Improve layout history error message: hide layout history reason of various generalized vars (ocaml-flambda#1889) * add generalized creation reason * use change layout reason for value decl * change layout reason for decl type params * test changes * additional tests * add "the" * test changes * improve generalize on type decl * test changes * update err msg test case * add print_loc_in_lowercase * test changes * update_decls_layout_reason code cleanup Improve layout history error message: better test coverage (ocaml-flambda#1924) * Annotation * concrete layout * value layout * immediate layout * float64 layout * the rest * fix probe test * fix tests * more changes * more value test * add note * remove unused With_constraint constructor * fix format * remove unused void_creation_reason * remove unused Value_kind * add Missing_cmi test * consistent names for Type_variable&Univar * add param to Unannotated_type_parameter * add CRs * fix wording * add warning to all reasons we can't produce * more CRs * fix formatting * add new tests * enhance creation reason when missing cmi * fix Unannotated_type_parameter * test changes * code formatting (only) * Record missing cmi in the layout history * Add a CR * add broken test * fix is_missing_cmi check --------- Co-authored-by: Richard Eisenberg <reisenberg@janestreet.com> Support for [@error_message] attribute on layout annotations (ocaml-flambda#1943) support for error_message attr on layout annots add error_message to the builtin_attrs table change error_message_attr to return first refactor - layout_of_annotation_or_attr to layout_of_annotation - Err_msg_attr to With_error_message fix tests improve formatting Support for [@error_message] attribute on Pexp_constraint (ocaml-flambda#1954) * implement error_message on Pexp_constraint * add tests * force message to be on new line * fix typo Improve layout history error message: omit history for signature mismatch (ocaml-flambda#2089) strip away layout history for signature mismatch Layout history: mark `typing-immediate/immediate.ml` as reviewed (ocaml-flambda#2144) mark test file as reviewed Improve layout history error message: review new messages (ocaml-flambda#2091) * remove layouts v2.9 CRs * "must be representable" instead of var name * catch missing history generalization * add history to Bad_univar_layout * remove crs * Toplevel_nonvalue refer to types * change Non_value_in_sig to refer to types * fix typo of Polymorphic * type argument to a class constructor * change message of Not_a_value * remove stale test file * refer to types in all messages * update unannotated universal variable message * split into Class_type_argument&Class_term_argument * add comment on update_generalized_ty_layout_reason * undo change to Not_a_value * update Not_a_value message

* enable display_histories * accept test changes and mark them as unreviewed * improve CR messages for unreviewed tests --------- Co-authored-by: Alan Chang <alanechang@janestreet.com> Remove duplicated creation_reasons from layout history error messages (ocaml-flambda#1825) * stop storing duplicates in creation_reasons * accept test changes * reduce to only displaying one creation reason * accept test changes * prefer other reasons over Concrete_creation * accept test changes * stop building history tree if flatten is enabled * remove CR version Co-authored-by: Richard Eisenberg <reisenberg@janestreet.com> * Add CR for assumption in combine_histories Co-authored-by: Richard Eisenberg <reisenberg@janestreet.com> --------- Co-authored-by: Alan Chang <alanechang@janestreet.com> Co-authored-by: Richard Eisenberg <reisenberg@janestreet.com> Improve layout history error message formatting and wording (ocaml-flambda#1832) * remove extra whitespace * fix Unequal_var_layouts error fmt * improve message wording * fix spacing * accept test changes * grammar and phrasing * test changes --------- Co-authored-by: Alan Chang <alanechang@janestreet.com> Improve layout history error message: add path field to Type_argument value_creation_reason (ocaml-flambda#1834) * add Path.t to Type_argument * test changes * add Imported_type_argument * Imported_type_argument for imported type params * fix float64 error format * test changes * add more tests * update layout reason not just in failure case * tests with more than one type param * combine Imported_type_argument with Type_argument * add option_argument_layout to predef * track position and arity for Type_argument * test changes * more tests around typa argument arity * have Imported_type_argument as a creation_reason * test changes * add note about performance benchmark * extract position format logic into function * add comments about 1-indexed * change equals to is * test changes * define list_argument_layout * format comments --------- Co-authored-by: Alan Chang <alanechang@janestreet.com> Improve layout history error message: additional info on value layout defaulting (ocaml-flambda#1855) * add message when defaulting to layout value * test changes * relax to all const defaulting * test changes * improve wording * test changes --------- Co-authored-by: alanechang <alanechang@janestreet.com> Improve layout history error message: change message wording and add more tests (ocaml-flambda#1887) * change imported reason wording * test changes * lowercase * test changes * add more tests * change wording for match * update Immediate_polymorphic_variant message * test changes Improve layout history error message: hide layout history reason of various generalized vars (ocaml-flambda#1889) * add generalized creation reason * use change layout reason for value decl * change layout reason for decl type params * test changes * additional tests * add "the" * test changes * improve generalize on type decl * test changes * update err msg test case * add print_loc_in_lowercase * test changes * update_decls_layout_reason code cleanup Improve layout history error message: better test coverage (ocaml-flambda#1924) * Annotation * concrete layout * value layout * immediate layout * float64 layout * the rest * fix probe test * fix tests * more changes * more value test * add note * remove unused With_constraint constructor * fix format * remove unused void_creation_reason * remove unused Value_kind * add Missing_cmi test * consistent names for Type_variable&Univar * add param to Unannotated_type_parameter * add CRs * fix wording * add warning to all reasons we can't produce * more CRs * fix formatting * add new tests * enhance creation reason when missing cmi * fix Unannotated_type_parameter * test changes * code formatting (only) * Record missing cmi in the layout history * Add a CR * add broken test * fix is_missing_cmi check --------- Co-authored-by: Richard Eisenberg <reisenberg@janestreet.com> Support for [@error_message] attribute on layout annotations (ocaml-flambda#1943) support for error_message attr on layout annots add error_message to the builtin_attrs table change error_message_attr to return first refactor - layout_of_annotation_or_attr to layout_of_annotation - Err_msg_attr to With_error_message fix tests improve formatting Support for [@error_message] attribute on Pexp_constraint (ocaml-flambda#1954) * implement error_message on Pexp_constraint * add tests * force message to be on new line * fix typo Improve layout history error message: omit history for signature mismatch (ocaml-flambda#2089) strip away layout history for signature mismatch Layout history: mark `typing-immediate/immediate.ml` as reviewed (ocaml-flambda#2144) mark test file as reviewed Improve layout history error message: review new messages (ocaml-flambda#2091) * remove layouts v2.9 CRs * "must be representable" instead of var name * catch missing history generalization * add history to Bad_univar_layout * remove crs * Toplevel_nonvalue refer to types * change Non_value_in_sig to refer to types * fix typo of Polymorphic * type argument to a class constructor * change message of Not_a_value * remove stale test file * refer to types in all messages * update unannotated universal variable message * split into Class_type_argument&Class_term_argument * add comment on update_generalized_ty_layout_reason * undo change to Not_a_value * update Not_a_value message fix merge conflicts

* enable display_histories * accept test changes and mark them as unreviewed * improve CR messages for unreviewed tests --------- Co-authored-by: Alan Chang <alanechang@janestreet.com> Remove duplicated creation_reasons from layout history error messages (#1825) * stop storing duplicates in creation_reasons * accept test changes * reduce to only displaying one creation reason * accept test changes * prefer other reasons over Concrete_creation * accept test changes * stop building history tree if flatten is enabled * remove CR version Co-authored-by: Richard Eisenberg <reisenberg@janestreet.com> * Add CR for assumption in combine_histories Co-authored-by: Richard Eisenberg <reisenberg@janestreet.com> --------- Co-authored-by: Alan Chang <alanechang@janestreet.com> Co-authored-by: Richard Eisenberg <reisenberg@janestreet.com> Improve layout history error message formatting and wording (#1832) * remove extra whitespace * fix Unequal_var_layouts error fmt * improve message wording * fix spacing * accept test changes * grammar and phrasing * test changes --------- Co-authored-by: Alan Chang <alanechang@janestreet.com> Improve layout history error message: add path field to Type_argument value_creation_reason (#1834) * add Path.t to Type_argument * test changes * add Imported_type_argument * Imported_type_argument for imported type params * fix float64 error format * test changes * add more tests * update layout reason not just in failure case * tests with more than one type param * combine Imported_type_argument with Type_argument * add option_argument_layout to predef * track position and arity for Type_argument * test changes * more tests around typa argument arity * have Imported_type_argument as a creation_reason * test changes * add note about performance benchmark * extract position format logic into function * add comments about 1-indexed * change equals to is * test changes * define list_argument_layout * format comments --------- Co-authored-by: Alan Chang <alanechang@janestreet.com> Improve layout history error message: additional info on value layout defaulting (#1855) * add message when defaulting to layout value * test changes * relax to all const defaulting * test changes * improve wording * test changes --------- Co-authored-by: alanechang <alanechang@janestreet.com> Improve layout history error message: change message wording and add more tests (#1887) * change imported reason wording * test changes * lowercase * test changes * add more tests * change wording for match * update Immediate_polymorphic_variant message * test changes Improve layout history error message: hide layout history reason of various generalized vars (#1889) * add generalized creation reason * use change layout reason for value decl * change layout reason for decl type params * test changes * additional tests * add "the" * test changes * improve generalize on type decl * test changes * update err msg test case * add print_loc_in_lowercase * test changes * update_decls_layout_reason code cleanup Improve layout history error message: better test coverage (#1924) * Annotation * concrete layout * value layout * immediate layout * float64 layout * the rest * fix probe test * fix tests * more changes * more value test * add note * remove unused With_constraint constructor * fix format * remove unused void_creation_reason * remove unused Value_kind * add Missing_cmi test * consistent names for Type_variable&Univar * add param to Unannotated_type_parameter * add CRs * fix wording * add warning to all reasons we can't produce * more CRs * fix formatting * add new tests * enhance creation reason when missing cmi * fix Unannotated_type_parameter * test changes * code formatting (only) * Record missing cmi in the layout history * Add a CR * add broken test * fix is_missing_cmi check --------- Co-authored-by: Richard Eisenberg <reisenberg@janestreet.com> Support for [@error_message] attribute on layout annotations (#1943) support for error_message attr on layout annots add error_message to the builtin_attrs table change error_message_attr to return first refactor - layout_of_annotation_or_attr to layout_of_annotation - Err_msg_attr to With_error_message fix tests improve formatting Support for [@error_message] attribute on Pexp_constraint (#1954) * implement error_message on Pexp_constraint * add tests * force message to be on new line * fix typo Improve layout history error message: omit history for signature mismatch (#2089) strip away layout history for signature mismatch Layout history: mark `typing-immediate/immediate.ml` as reviewed (#2144) mark test file as reviewed Improve layout history error message: review new messages (#2091) * remove layouts v2.9 CRs * "must be representable" instead of var name * catch missing history generalization * add history to Bad_univar_layout * remove crs * Toplevel_nonvalue refer to types * change Non_value_in_sig to refer to types * fix typo of Polymorphic * type argument to a class constructor * change message of Not_a_value * remove stale test file * refer to types in all messages * update unannotated universal variable message * split into Class_type_argument&Class_term_argument * add comment on update_generalized_ty_layout_reason * undo change to Not_a_value * update Not_a_value message fix merge conflicts Layout histories: resolve rebase conflicts (#2197) * get things to build * test changes * remove unused jkind reason * fix formatting

* Enable layout histories (#1823) * enable display_histories * accept test changes and mark them as unreviewed * improve CR messages for unreviewed tests --------- Co-authored-by: Alan Chang <alanechang@janestreet.com> Remove duplicated creation_reasons from layout history error messages (#1825) * stop storing duplicates in creation_reasons * accept test changes * reduce to only displaying one creation reason * accept test changes * prefer other reasons over Concrete_creation * accept test changes * stop building history tree if flatten is enabled * remove CR version Co-authored-by: Richard Eisenberg <reisenberg@janestreet.com> * Add CR for assumption in combine_histories Co-authored-by: Richard Eisenberg <reisenberg@janestreet.com> --------- Co-authored-by: Alan Chang <alanechang@janestreet.com> Co-authored-by: Richard Eisenberg <reisenberg@janestreet.com> Improve layout history error message formatting and wording (#1832) * remove extra whitespace * fix Unequal_var_layouts error fmt * improve message wording * fix spacing * accept test changes * grammar and phrasing * test changes --------- Co-authored-by: Alan Chang <alanechang@janestreet.com> Improve layout history error message: add path field to Type_argument value_creation_reason (#1834) * add Path.t to Type_argument * test changes * add Imported_type_argument * Imported_type_argument for imported type params * fix float64 error format * test changes * add more tests * update layout reason not just in failure case * tests with more than one type param * combine Imported_type_argument with Type_argument * add option_argument_layout to predef * track position and arity for Type_argument * test changes * more tests around typa argument arity * have Imported_type_argument as a creation_reason * test changes * add note about performance benchmark * extract position format logic into function * add comments about 1-indexed * change equals to is * test changes * define list_argument_layout * format comments --------- Co-authored-by: Alan Chang <alanechang@janestreet.com> Improve layout history error message: additional info on value layout defaulting (#1855) * add message when defaulting to layout value * test changes * relax to all const defaulting * test changes * improve wording * test changes --------- Co-authored-by: alanechang <alanechang@janestreet.com> Improve layout history error message: change message wording and add more tests (#1887) * change imported reason wording * test changes * lowercase * test changes * add more tests * change wording for match * update Immediate_polymorphic_variant message * test changes Improve layout history error message: hide layout history reason of various generalized vars (#1889) * add generalized creation reason * use change layout reason for value decl * change layout reason for decl type params * test changes * additional tests * add "the" * test changes * improve generalize on type decl * test changes * update err msg test case * add print_loc_in_lowercase * test changes * update_decls_layout_reason code cleanup Improve layout history error message: better test coverage (#1924) * Annotation * concrete layout * value layout * immediate layout * float64 layout * the rest * fix probe test * fix tests * more changes * more value test * add note * remove unused With_constraint constructor * fix format * remove unused void_creation_reason * remove unused Value_kind * add Missing_cmi test * consistent names for Type_variable&Univar * add param to Unannotated_type_parameter * add CRs * fix wording * add warning to all reasons we can't produce * more CRs * fix formatting * add new tests * enhance creation reason when missing cmi * fix Unannotated_type_parameter * test changes * code formatting (only) * Record missing cmi in the layout history * Add a CR * add broken test * fix is_missing_cmi check --------- Co-authored-by: Richard Eisenberg <reisenberg@janestreet.com> Support for [@error_message] attribute on layout annotations (#1943) support for error_message attr on layout annots add error_message to the builtin_attrs table change error_message_attr to return first refactor - layout_of_annotation_or_attr to layout_of_annotation - Err_msg_attr to With_error_message fix tests improve formatting Support for [@error_message] attribute on Pexp_constraint (#1954) * implement error_message on Pexp_constraint * add tests * force message to be on new line * fix typo Improve layout history error message: omit history for signature mismatch (#2089) strip away layout history for signature mismatch Layout history: mark `typing-immediate/immediate.ml` as reviewed (#2144) mark test file as reviewed Improve layout history error message: review new messages (#2091) * remove layouts v2.9 CRs * "must be representable" instead of var name * catch missing history generalization * add history to Bad_univar_layout * remove crs * Toplevel_nonvalue refer to types * change Non_value_in_sig to refer to types * fix typo of Polymorphic * type argument to a class constructor * change message of Not_a_value * remove stale test file * refer to types in all messages * update unannotated universal variable message * split into Class_type_argument&Class_term_argument * add comment on update_generalized_ty_layout_reason * undo change to Not_a_value * update Not_a_value message fix merge conflicts Layout histories: resolve rebase conflicts (#2197) * get things to build * test changes * remove unused jkind reason * fix formatting * fix rebase conflicts * fix format

* Enable layout histories (ocaml-flambda#1823) * enable display_histories * accept test changes and mark them as unreviewed * improve CR messages for unreviewed tests --------- Co-authored-by: Alan Chang <alanechang@janestreet.com> Remove duplicated creation_reasons from layout history error messages (ocaml-flambda#1825) * stop storing duplicates in creation_reasons * accept test changes * reduce to only displaying one creation reason * accept test changes * prefer other reasons over Concrete_creation * accept test changes * stop building history tree if flatten is enabled * remove CR version Co-authored-by: Richard Eisenberg <reisenberg@janestreet.com> * Add CR for assumption in combine_histories Co-authored-by: Richard Eisenberg <reisenberg@janestreet.com> --------- Co-authored-by: Alan Chang <alanechang@janestreet.com> Co-authored-by: Richard Eisenberg <reisenberg@janestreet.com> Improve layout history error message formatting and wording (ocaml-flambda#1832) * remove extra whitespace * fix Unequal_var_layouts error fmt * improve message wording * fix spacing * accept test changes * grammar and phrasing * test changes --------- Co-authored-by: Alan Chang <alanechang@janestreet.com> Improve layout history error message: add path field to Type_argument value_creation_reason (ocaml-flambda#1834) * add Path.t to Type_argument * test changes * add Imported_type_argument * Imported_type_argument for imported type params * fix float64 error format * test changes * add more tests * update layout reason not just in failure case * tests with more than one type param * combine Imported_type_argument with Type_argument * add option_argument_layout to predef * track position and arity for Type_argument * test changes * more tests around typa argument arity * have Imported_type_argument as a creation_reason * test changes * add note about performance benchmark * extract position format logic into function * add comments about 1-indexed * change equals to is * test changes * define list_argument_layout * format comments --------- Co-authored-by: Alan Chang <alanechang@janestreet.com> Improve layout history error message: additional info on value layout defaulting (ocaml-flambda#1855) * add message when defaulting to layout value * test changes * relax to all const defaulting * test changes * improve wording * test changes --------- Co-authored-by: alanechang <alanechang@janestreet.com> Improve layout history error message: change message wording and add more tests (ocaml-flambda#1887) * change imported reason wording * test changes * lowercase * test changes * add more tests * change wording for match * update Immediate_polymorphic_variant message * test changes Improve layout history error message: hide layout history reason of various generalized vars (ocaml-flambda#1889) * add generalized creation reason * use change layout reason for value decl * change layout reason for decl type params * test changes * additional tests * add "the" * test changes * improve generalize on type decl * test changes * update err msg test case * add print_loc_in_lowercase * test changes * update_decls_layout_reason code cleanup Improve layout history error message: better test coverage (ocaml-flambda#1924) * Annotation * concrete layout * value layout * immediate layout * float64 layout * the rest * fix probe test * fix tests * more changes * more value test * add note * remove unused With_constraint constructor * fix format * remove unused void_creation_reason * remove unused Value_kind * add Missing_cmi test * consistent names for Type_variable&Univar * add param to Unannotated_type_parameter * add CRs * fix wording * add warning to all reasons we can't produce * more CRs * fix formatting * add new tests * enhance creation reason when missing cmi * fix Unannotated_type_parameter * test changes * code formatting (only) * Record missing cmi in the layout history * Add a CR * add broken test * fix is_missing_cmi check --------- Co-authored-by: Richard Eisenberg <reisenberg@janestreet.com> Support for [@error_message] attribute on layout annotations (ocaml-flambda#1943) support for error_message attr on layout annots add error_message to the builtin_attrs table change error_message_attr to return first refactor - layout_of_annotation_or_attr to layout_of_annotation - Err_msg_attr to With_error_message fix tests improve formatting Support for [@error_message] attribute on Pexp_constraint (ocaml-flambda#1954) * implement error_message on Pexp_constraint * add tests * force message to be on new line * fix typo Improve layout history error message: omit history for signature mismatch (ocaml-flambda#2089) strip away layout history for signature mismatch Layout history: mark `typing-immediate/immediate.ml` as reviewed (ocaml-flambda#2144) mark test file as reviewed Improve layout history error message: review new messages (ocaml-flambda#2091) * remove layouts v2.9 CRs * "must be representable" instead of var name * catch missing history generalization * add history to Bad_univar_layout * remove crs * Toplevel_nonvalue refer to types * change Non_value_in_sig to refer to types * fix typo of Polymorphic * type argument to a class constructor * change message of Not_a_value * remove stale test file * refer to types in all messages * update unannotated universal variable message * split into Class_type_argument&Class_term_argument * add comment on update_generalized_ty_layout_reason * undo change to Not_a_value * update Not_a_value message fix merge conflicts Layout histories: resolve rebase conflicts (ocaml-flambda#2197) * get things to build * test changes * remove unused jkind reason * fix formatting * fix rebase conflicts * fix format

alanechang requested review from lpw25, ccasin, goldfirere, riaqn, ncik-roberts, stedolan, mshinwell, poechsel, gretay-js, xclerc and lukemaurer as code owners October 20, 2023 12:55

ncik-roberts self-assigned this Oct 20, 2023

ncik-roberts assigned goldfirere and unassigned ncik-roberts Oct 30, 2023

goldfirere approved these changes Oct 30, 2023

View reviewed changes

alanechang force-pushed the aec/error-message-attr-on-layouts branch from dfcce22 to cefa5ff Compare October 31, 2023 13:22

alanechang force-pushed the aec/error-message-attr-on-terms branch from 093e403 to 8a9bcd3 Compare November 2, 2023 18:37

alanechang force-pushed the aec/error-message-attr-on-layouts branch from cefa5ff to cf252f1 Compare November 21, 2023 21:26

Base automatically changed from aec/error-message-attr-on-layouts to layout-histories November 21, 2023 21:54

alanechang added 4 commits November 21, 2023 16:59

implement error_message on Pexp_constraint

6cf15ad

add tests

3630621

force message to be on new line

c6278ce

fix typo

01bea87

alanechang force-pushed the aec/error-message-attr-on-terms branch from 8a9bcd3 to 01bea87 Compare November 21, 2023 22:03

alanechang merged commit 3a2b229 into layout-histories Nov 21, 2023

alanechang deleted the aec/error-message-attr-on-terms branch November 21, 2023 22:22

ncik-roberts mentioned this pull request Jul 1, 2024

Honor [@error_message] attribute even when its location is ghost #2750

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Support for [@error_message] attribute on Pexp_constraint #1954

Support for [@error_message] attribute on Pexp_constraint #1954

alanechang commented Oct 20, 2023

ncik-roberts commented Oct 20, 2023

ncik-roberts commented Oct 30, 2023

goldfirere commented Nov 21, 2023

Support for [@error_message] attribute on Pexp_constraint #1954

Support for [@error_message] attribute on Pexp_constraint #1954

Conversation

alanechang commented Oct 20, 2023

ncik-roberts commented Oct 20, 2023

ncik-roberts commented Oct 30, 2023

goldfirere commented Nov 21, 2023