Add virtual network access in templates for MarkdownField#5131
Add virtual network access in templates for MarkdownField#5131backspace wants to merge 1 commit into
MarkdownField#5131Conversation
Preview deploymentsHost Test Results 1 files 1 suites 1h 51m 27s ⏱️ Results for commit 83d2c13. Realm Server Test Results 1 files ±0 1 suites ±0 8m 50s ⏱️ - 6m 9s Results for commit 83d2c13. ± Comparison against earlier commit 00f5249. |
| return trimJsonExtension( | ||
| virtualNetwork.resolveURL(raw, baseUrl || undefined).href, | ||
| ); |
backspace
force-pushed
the
require-virtual-network-cs-11374
branch
from
0eba1c7 to
70424d7
Compare
MarkdownField.embedded / MarkdownField.atom render BFM markdown with `<MarkdownTemplate @content={{@model}} />`. The field templates only receive the field's primitive @model (a string), with no path to the owning card — so before this change they couldn't supply a VN to MarkdownTemplate, and prefix-form BFM pill refs (e.g. `@cardstack/catalog/foo`) couldn't be resolved at render time. Pick design option 2 from CS-11375: wrap MarkdownField's defaults in `<CardContextConsumer as |context|>` and pass `@cardReferenceVirtualNetwork={{context.store.virtualNetwork}}`. Adding `virtualNetwork: VirtualNetwork` to the Store interface (and exposing it as a getter on host StoreService that delegates to `this.network.virtualNetwork`) makes the context.store path supply VN without changing the field-render protocol — touches MarkdownField specifically rather than every field template. With every MarkdownTemplate call site now threading VN, `cardReferenceVirtualNetwork` becomes required on the component arg and the `resolveUrl` helper's no-VN branch deletes. markdown-file-def and rich-markdown's `virtualNetwork` getters tighten to `VirtualNetwork` with a `?? new VirtualNetwork()` fallback for detached models (preserves the no-throw behavior we already accept for static parsers). Adds a new `markdown-field-test.gts` covering MarkdownField.embedded and .atom rendering BFM pills, including a prefix-form ref that exercises the consumer plumbing end-to-end. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
backspace
force-pushed
the
network-in-template-cs-11375
branch
from
00f5249 to
83d2c13
Compare
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 83d2c13ddb
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
lukemelia
left a comment
lukemelia
left a comment
There was a problem hiding this comment.
Can we come up with a design that does not initiate a pattern of cards and/or card components interacting directly with VirtualNetwork instances? The possibilities for abuse are myriad. What capabilities do cards need that they don't have, and can we narrowly expose that ability?
3 participants
This is a small continuation from 5109, which changed all
VirtualNetworkparameters to be required but one:MarkdownField, which needs one to resolve@cardstack/whatever-style card references in markdown.The CI failure about the title is because of rebasing, this doesn’t have anything to do with Boxel CLI.