David Blass

Validated completions available for string props are not available for shallow strings

1

# Bug Report I encountered this bug while adding autocomplete support for ArkType. The goal is for completions to behave the same way whether they are parsed at an object...

Feature request: Distinguish safely-hoistable circular imports from those that will fail at runtime

3

If you're building a recursive structure like a DOM or a type system (in this case [the latter](https://github.com/arktypeio/arktype)), at some point you'll have to choose to either maintain a massive...

typeof with identically named type and value causes invalid .d.ts output

7

### 🔎 Search Terms type, value, .d.ts, circular, circularity, invalid, output, declaration ### 🕗 Version & Regression Information - This changed between versions 5.4 and 5.5 ### ⏯ Playground Link...

2.0 Tasks

3

This is for personal tracking purposes for the major tasks left for the beta release. - [x] Finish migrating Problems API - [x] Global/scope/type configs - [ ] Finish integrating...

Currently, since we only populate the snapshots from a single tsVersion, we also only check the primary version for an assertion like `attest({}).type.toString.snap("{}")`. It would be ideal to only populate...

Use the result with the shortest path when finalizing errors for a cyclic value

1

A [Boston TS](https://boston-ts-website.vercel.app/) attendee (please take credit if you see this) suggested this could be optimized using a breadth-first approach to traversal. While changing the traversal pattern would likely incur...

Especially relevant for "===" check for symbols with the same name or objects with the same structure. Not sure if there are other cases where this would come up, but...

Currently, because JSON for a cyclic type references only its alias like `$a`, if two scopes had differing definitions for the same cyclic alias, caching could break. Really this should...

Currently unions don't use their discriminators during traversal. Reintegrate these, and ensure that unions of morphs that are not discriminable are errors.

The current transform behavior for cyclic nodes is not robust. I slapped together a ctx with seen and lazy resolutions, but currently `lazilyResolve` is limited to root nodes, whereas to...