[sapphi-red]

It seems we can align this behavior by tweaking some mappings.

fixes #19625

/ecosystem-ci run

Open in StackBlitz

@vitejs/plugin-legacy

pnpm add https://pkg.pr.new/@vitejs/plugin-legacy@22829 -D

npm i https://pkg.pr.new/@vitejs/plugin-legacy@22829 -D

yarn add https://pkg.pr.new/@vitejs/plugin-legacy@22829.tgz -D

vite

pnpm add https://pkg.pr.new/vite@22829 -D

npm i https://pkg.pr.new/vite@22829 -D

yarn add https://pkg.pr.new/vite@22829.tgz -D

commit: 2388d44

📝 Ran ecosystem CI on 0e21011: Open

suite	result	latest scheduled
astro	❌ failure	❌ failure
react-router	❌ failure	❌ failure
nuxt	❌ failure	✅ success
storybook	✅ success	❌ failure
tanstack-start	⏹️ cancelled	⏹️ cancelled
vitest	❌ failure	✅ success

✅ marko, module-federation, analogjs, vite-environment-examples, quasar, qwik, laravel, vite-plugin-svelte, vite-plugin-vue, vite-plugin-cloudflare, sveltekit, vite-plugin-react, vite-plugin-pwa, vike, unocss, vite-setup-catalogue, vitepress, vite-plugin-rsc, vuepress, waku

Nice trick 👍

So cool!
Had struggled to understand the idea, so I had AI visualize the mapping 😃 https://artifacts.hiro18181.workers.dev/vite-pr-22829

Not sure how much trusted but AI said this adjustment happens here https://github.com/v8/v8/blob/0c7a9d0d8c05491e8053d7259a1a6b11e75c1907/src/parsing/parser-base.h#L4037-L4043

The comment calls out the case of IIFE function(){}(), but parenthesized case seems to also hit this case.

Also, it looks like bun isn't affected (and already correct) since they already have different callee mapping approach that doesn't rely on argument list (.

Had struggled to understand the idea, so I had AI visualize the mapping 😃 https://artifacts.hiro18181.workers.dev/vite-pr-22829

Cool! What prompt did you use to generate this page?

Had struggled to understand the idea, so I had AI visualize the mapping 😃 https://artifacts.hiro18181.workers.dev/vite-pr-22829

Cool! What prompt did you use to generate this page?

I have some some personal skills in https://github.com/hi-ogawa/dotfiles/tree/main/skills and this was mostly a combination of ho-review then ho-artifact. I used opencode with anthropic/claude-opus-4-8.

---

Full session transcript is https://gist.github.com/hi-ogawa/8f306058811eab7bd85244aeaab3fc6e and I prompted this after review:

can you visualize the mapping fix idea? perhas before and after which how column maps?

I confirmed the change on Vitest vitest-dev/vitest#10722. It looks like CI is failing for other reason between 8.0.11 to 8.1.3 though vitest-dev/vitest#10723

#22484 and #22695 had some sourcemap changes so that might be related

Turns out Vitest CI fails is because of a Vitest bug. It's being fixed in vitest-dev/vitest#10724.
This PR should be good to merge.

/ecosystem-ci run

📝 Ran ecosystem CI on cc7b640: Open

suite	result	latest scheduled
astro	❌ failure	❌ failure
react-router	❌ failure	❌ failure
vitest	❌ failure	✅ success
tanstack-start	⏹️ cancelled	⏹️ cancelled

✅ nuxt, sveltekit, laravel, quasar, vike, vite-environment-examples, module-federation, analogjs, marko, storybook, qwik, unocss, vuepress, vite-plugin-react, vite-plugin-vue, vite-plugin-cloudflare, vite-plugin-svelte, vite-plugin-pwa, vite-setup-catalogue, waku, vite-plugin-rsc, vitepress