Bug 299976 – [WPE] Different WPE Release post-commit layout testers are reporting results with same configuration

• Component: Tools / Tests
• Status: RESOLVED FIXED

[Lauro Moura]

Currently, we have two WPE Release post-commit bots that run the layout tests and upload the results to https://results.webkit.org

- The main `WPE-Linux-64-bit-Release-Tests` at https://build.webkit.org/#/builders/40
- The legacy API `WPE-Linux-64-bit-Release-Legacy-API-Tests` at https://build.webkit.org/#/builders/1618

Currently, both bots share the same configuration, which for the recent set of results is:

```
    "configuration": {
      "architecture": "x86",
      "flavor": "wk2",
      "is_simulator": false,
      "platform": "WPE",
      "style": "release",
      "version": 6016003
    },
```

Note that "version" by default refers to the OS version. In our case, the kernel version.

This setup means that both bots are viewed as part of the same "result queue" when the resultsdb processes an incoming test upload. And, as a result, when we try to fetch the latest aggregated data for the wpe-release bots with something like https://results.webkit.org/api/results/layout-tests?platform=WPE&style=release&limit=10, we get results from both bots, interleaved by timestamp. Ditto for the Web UI.

Maybe we could use an attribute like "version_name" or "flavor" to separate both queues.

[Lauro Moura]

Pull request: https://github.com/WebKit/WebKit/pull/51680

[EWS]

Committed 300891@main (4c0d8ce0b00b): <https://commits.webkit.org/300891@main>

Reviewed commits have been landed. Closing PR #51680 and removing active labels.

[Radar WebKit Bug Importer]

<rdar://problem/161836435>