Non-nilable type checking meta-issue #13598
Assignees
Comments
vasslitvinov
changed the title
|
By pinning this issue, my concern is more w.r.t. the tests that are requiring |
vasslitvinov
added a commit
to vasslitvinov/chapel
that referenced
this issue
Oct 5, 2019
For 12 of these tests, I added `?` where the code was using non-nilable types incorrectly. The remaining fixes are: * classes/errors/newValueNotType.chpl - I merely adjusted .bad * deprecated/arrayAsVec/push-back-owned.chpl - no changes, just removed .compopts Here is my tally of the remaining compopts with `--legacy-classes`, as of 92fc605 - chapel-lang#13598 needs updating: * The compile-time control-flow-based nil checking works differently in the presence of nilable types: classes/delete-free/nil-checking/nil-check-control.chpl classes/delete-free/nil-checking/nilcheck-empty-refs.chpl classes/delete-free/nil-checking/nilcheck.chpl * Compile-time control-flow-based nil checking, this is a wish item: classes/vass/if-object-2.chpl * Related to owned/shared - I suspect straightforward to update: classes/ferguson/class-double-modifier.chpl * Old-style runtime nil checking chapel-lang#13625, intentionally legacy: classes/diten/callNilClassesMethod.chpl classes/diten/nilDynamicDispatch.chpl * Related to nilability; probably intentionally legacy: classes/delete-free/default-arg-typeless-error.chpl classes/delete-free/default-arg-typeless-legacy.chpl * Intentional legacy checking: classes/errors/newValueNotType2.chpl compflags/ferguson/unstable-class.chpl compflags/ferguson/unstable-owned-shared-intent2.chpl
vasslitvinov
added a commit
that referenced
this issue
Oct 7, 2019
Fix some tests to work without --legacy-classes For 12 of these tests, I added `?` where the code was using non-nilable types incorrectly. The remaining fixes are: * classes/errors/newValueNotType.chpl - I merely adjusted .bad * deprecated/arrayAsVec/push-back-owned.chpl - no changes, just removed .compopts Here is my tally of the remaining compopts with `--legacy-classes`, as of 92fc605 - #13598 needs updating: * The compile-time control-flow-based nil checking works differently in the presence of nilable types: classes/delete-free/nil-checking/nil-check-control.chpl classes/delete-free/nil-checking/nilcheck-empty-refs.chpl classes/delete-free/nil-checking/nilcheck.chpl * Compile-time control-flow-based nil checking, this is a wish item: classes/vass/if-object-2.chpl * Related to owned/shared - I suspect straightforward to update: classes/ferguson/class-double-modifier.chpl * Old-style runtime nil checking #13625, intentionally legacy: classes/diten/callNilClassesMethod.chpl classes/diten/nilDynamicDispatch.chpl * Related to nilability; probably intentionally legacy: classes/delete-free/default-arg-typeless-error.chpl classes/delete-free/default-arg-typeless-legacy.chpl * Intentional legacy checking: classes/errors/newValueNotType2.chpl compflags/ferguson/unstable-class.chpl compflags/ferguson/unstable-owned-shared-intent2.chpl While there, remove .compopts with `--no-legacy-classes`. r: @mppf
|
@bradcray - perhaps unpin this issue now? |
|
This issue has served its purpose. Closing. The individual issues linked from here are unaffected by closing this issue. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
This is a summary of nilability / nil checking issues.
Starting point: nilable types #12614
Design-ish
const refversion of owned.borrow() ? should owned.borrow() have a const ref version? #13638!w.r.t.--fastand performance impact Should runtime check in '!' operator run upon --fast ? #13603Convenience features
!for accessing param fields and similar Should not need '!' for accessing param fields and similar #13617 (resolved), Should not need '!' for accessing compile-time properties of a field #14705 (open)Bugs, would-be-nice
Resolved issues
--legacy-classesoption--legacy-classes!Prototype modules to no longer enable implicit '!' #14559!to be applied to typesnew owned (value)(?) How to create ownership for an unmanaged class? #14300?-- postfix-? leads to "error: cannot convert to a pointer type" #13759--no-legacy-classesby default--no-legacy-classesnew [flavor] C()?mean? What should 'new [flavor] C()?' mean? #13619x!be an alias tox? Design: should the result of 'x!' be an alias to 'x' ? #13621The text was updated successfully, but these errors were encountered: