[docs] Wave 3 guide: taking playground-ui off antd - #5700
Conversation
…f antd We are not doing this wave ourselves, so the guide has to carry what the last two waves learned rather than assume the reader was there. Measured, not estimated: 78 files, 31 on antd, zero Storybook coverage today, and 213 atom hooks across 31 jotai files. That last number is the point of the chunking advice — the fixtures are the work here, not the antd swaps, so the six suggested chunks are grouped by data coupling rather than file count. Records what is specific to this package: the PlaygroundUIContext seam (prefer providing a context over seeding atoms), the entity-ui composites already available to reuse, antd Upload having no primitive but a working precedent, and the shared ExecutionItems helpers that make three of the chunks unsafe to run in parallel. Adds standards for contributors using coding agents. A frontier model is asked for on migration and gate triage because every expensive mistake in waves 1 and 2 was silent: a token that type-checks, an accessible name that lands on the wrong element, a prop a wrapper drops. Small models are fine for the mechanical parts once the mapping is decided. The agent rules are the ones that were actually violated, and the review checklist is what a maintainer can grep.
dosubot
Bot
added
the
size:L
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
|
Important Review skippedAuto reviews are disabled on base/target branches other than the default branch. Please check the settings in the CodeRabbit UI or the ⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Plus Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
3 participants
Context
Wave 3 of the antd migration is
@agenta/playground-ui: 31 antd files, and no Storybook coverage at all today. We are not doing this wave ourselves, so it needs a guide that carries what waves 1 and 2 learned rather than assuming the reader was there for them.Changes
Adds
antd-inventory/playground-ui.mdand links it fromSTATUS.md.The numbers in it are measured, not estimated: 78 files, 31 on antd, about 8k lines, and 213 atom hooks across 31 jotai files. That last figure is the highest data-coupling density of any package so far, and it drives the advice. The six suggested chunks are grouped by data coupling rather than file count, because a 1,084-line presentational file is a smaller job than a 250-line container with nine atom reads.
It records what is specific to this package and would otherwise be learned the hard way:
PlaygroundUIContextis a second data seam alongside jotai. Eight files read it, and providing a context in a story beats seeding atoms.@agenta/entity-ui, so the composites built in waves 1 and 2 are already available. Check before building a second version.Uploadhas no primitive, butSkillUploadZonefrom wave 2 is a working precedent.ExecutionItemshelpers. Waves 1 and 2 ran six agents in parallel safely only because the chunks touched disjoint files, so this one carries an explicit warning against doing that blindly.It also adds standards for contributors working with coding agents. A frontier model is asked for on migration and gate triage, justified with evidence rather than as policy: every expensive mistake in the first two waves was silent.
Typography.Text type="secondary"maps tocolorTextDescription, not the identically namedcolorTextSecondary, and 26 sites were wrong where the pixel gate could not see it. An accessible name passed to a Slider's Root instead of its thumb looks correct at the call site and names nothing. Lexical takesaria-label, and anariaLabelprop type-checks and is dropped. Small models are explicitly fine for the mechanical parts once the mapping is decided.The six agent rules are the ones that were actually violated during our waves, and the maintainer review checklist is written so it can be grepped, for example "does the parity story contain an antd half".
Notes
Docs only, no code. Stacked on #5694 because the guide builds on what wave 2 produced, including the corrections now recorded in
entity-ui.md. Split out of #5694 so contributors can pick wave 3 up without waiting for a 256-file PR to clear review.