Pedro
Pedro
No sorry, no time this quarter.
Closing this PR. If still interested, please file an issue instead so that this is tackled in a different way. It should pass Yannic's test and internal tests.
Hey Fabian, are you sure about this? I see [here](https://cs.opensource.google/bazel/bazel/+/master:src/main/java/com/google/devtools/build/lib/sandbox/SandboxStash.java;drc=f799eb14e747ed95fe79fafedec7ccd0b2e50a1f;l=124) that we are only stashing the `execroot` from the `sandboxBase`. The `_hermetic_tmp` directory is a sibling of the `execroot`. The...
Oh I can still fix the _tmp one. I think it makes total sense not to stash that in tests.
Having to fix every output symlink seems costly. It will be costly for every build since we have to scan outputs of every action to see if they are symlinks...
Perfect, thank you Fabian. I'd go with the overlayfs solution when we detect sourceroot and execroot are under `/tmp` but would take a simpler code path when we don't detect...
I'd say that fixing symlinks being simple in most cases is not a strong argument. The current implementation of hermetic tmp also worked most of the time. What if we...
Ok, so let's try `overlayfs` for both cases then. I will not be working the next 3 days. Can this wait till next week?
This can wait till 7.2.0.
Does everyone affected by this NOT have dynamic execution enabled?