chore(NA): run del with force option on ref_output_cache

[mistic]

Looks like deleting the build_refs caches is failing on Windows and that is fixed by running del with force option.

[elasticmachine]

Pinging @elastic/kibana-operations (Team:Operations)

[peteharverson]

Tested and LGTM

[kibanamachine]

💚 Build Succeeded

• continuous-integration/kibana-ci/pull-request
• Commit: 045d0d7
• Storybooks Preview

Metrics [docs]

✅ unchanged

To update your PR or re-run it, just comment with:
@elasticmachine merge upstream

[spalger]

The outDir values are all expected to be relative to the repo, in what case would an outDir be outside of the repo and require force?

[mistic]

Okay I think I found the root cause for this. On the current Node.js version on Windows (probably also happening in previous ones), as soon as we require some file/module that is contained inside a symlink it is treated by Windows like it was on a literally different but still equivalent path. In our case as we require @kbn/optimizer as part of the setup_node_env the problem starts right here. For example, before requiring the optimizer __dirname returns C:\Users\IEUser\Downloads\kibana and after the require it while we are inside of it , it starts returning c:\users\ieuser\downlo~1\kibana. That will also affect the result of REPO_ROOT that we use all over. When running node with --preserve-symlinks it stops happening so I think we are on the right path. Aside from that, I was not able to fully bootstrap on windows as the build for kbn/ui-shared-deps complains about a node_module that is being loaded from 2 different paths by the same reason listed above and I was in fact able to found old issues about this problem nodejs/node#6624 webpack/webpack#2362.

For our Bazel work I was already working on #94046 which might also solve that issue. @peteharverson could you please try that other PR I have open to see if that alternative solution also works for you on Windows?

[peteharverson]

@mistic I gave #94046 a test, but see errors - see #94046 (comment). Let me know if you want me to try anything else.

[mistic]

Closing in favor of #94046