Reporters data is bloated compared to v3.1.x

[idanen]

Describe the bug

Hey.
We found that using a newer version than 3.1.4 creates larger reporters data which fails some reporters.
On 3.2.4, this created a situation for us that the html reporter failed (RangeError: Invalid string length), and on 4.0.15 it failed with the blob reporter (on the same reason).
In the repro we only see small increase, but in large projects (we have ~10,000 tests in ~15 projects), this can go as high as 20 times (this is not a mistake, 20 times larger) difference.
When sharding, blob reporter did finish successfully, but merging failed silently (we found that the html reporter swallows the exception).

Reproduction

Blob size with v3.1.4 (stackblitz):

Blob size with v4.0.15 (stackblitz):

System Info

System:
    OS: Linux 5.0 undefined
    CPU: (8) x64 Intel(R) Core(TM) i9-9880H CPU @ 2.30GHz
    Memory: 0 Bytes / 0 Bytes
    Shell: 1.0 - /bin/jsh
  Binaries:
    Node: 20.19.1 - /usr/local/bin/node
    Yarn: 1.22.19 - /usr/local/bin/yarn
    npm: 10.8.2 - /usr/local/bin/npm
    pnpm: 8.15.6 - /usr/local/bin/pnpm
  npmPackages:
    @vitest/ui: 4.0.15 => 4.0.15 
    vite: latest => 6.4.1 
    vitest: 4.0.15 => 4.0.15

Used Package Manager

npm

Validations

• Follow our Code of Conduct
• Read the Contributing Guidelines.
• Read the docs.
• Check that there isn't already an issue that reports the same bug to avoid creating a duplicate.
• Check that this is a concrete bug. For Q&A open a GitHub Discussion or join our Discord Chat Server.
• The provided reproduction is a minimal reproducible example of the bug.

[AriPerkkio]

Comparing the reports on the minimal repro, I think the only difference is the metrics from import durations:

• Ability to report on collect time #8026

Those are collected always. There is no way to opt-out even if you don't use that data for anything.

Can you create a reproduction that demonstrates the reporter crash? I guess that can happen in v3 too if the results are just too large.

[idanen]

Thanks for getting back on this.

Those are collected always. There is no way to opt-out even if you don't use that data for anything.

But can they be omitted when creating the report (I tried editing the reporter inside node_modules which helped)?

Can you create a reproduction that demonstrates the reporter crash? I guess that can happen in v3 too if the results are just too large.

That would be tough, since it requires a giant suite of tests... Not sure how to approach. The failure is due to large data coming in to reporters that want to stringify it.
We noticed this fails similarly in v3.2.4, but there it only fails in the html reporter, not the blob one

[hi-ogawa]

I made artificial test suites to show blob json size increase between v3.1.4 and v4.0.15 https://github.com/hi-ogawa/vitest-9216-repro The increase seems significant when there are many dependencies.

I'm not sure if there's a room for optimization. Otherwise, we need a way to opt-out certain reports at certain stage, either don't track some data at all internally, or just exclude some data at reporter level before serializing raw result to a huge string.

[AriPerkkio]

Marking this as reporter bug due to the crash. But something should be done to the metrics collected by test runners too, if possible.

[idanen]

Amazing repro @hi-ogawa!
I can confirm we have a dependencies problem. We're working on breaking a monolith to packages