Ergonomic, hermetic, user-friendly software build system for Python, Go, Java, Kotlin, Protobuf, Scala, Shell. Pants lets you fearlessly scale up your codebase!

Pants

...mm, it wasn't precisely master. i think this might be fixed. sorry for the runaround

curious to take a look at how thats happening even on older shas

on Source master, you get something like `PID 57078 waiting for a file lock (/Users/stuhood/src/source/.pids/.lock.pantsd) held by: 56924 (with no owner: stu's paraphrasing)`

after Daniel's pid file termination change, you do see the full process path

and then after b95299f0ad800da564282387887b1529b2fc2434, i don't think i'm able to repro at all.

guessing the backoff there may be helping

thanks for #5979 - should’ve thought about that during review, my bad there.

i feel like there might still be some distinction between bootstrap and global that we're missing

<https://github.com/python-tap/pytest-tap>

this looks potentially interesting for streaming test results from rules

i am... not optimistic about streaming of test outputs

...except maybe when there is exactly one ?

going to guess that the best we'll be able to do with multiple processes is dump each as they finish, followed by the frontend task rendering an aggregated summary

exactly one (at a time) is a property that a theoretical `@console_rule` can guarantee I think?

<@U054MQUPL>: yes, but to achieve parallelism, the tests won't be running "in" the `console_rule`, per se

i mean, it isn't possible for them to anyway... to run the test you'd make a process request, at which point, you suspend

and process execution might be via remoting over the gRPC protocol

which... doesn't support streaming, afaik. hence the concern that that won't be a thing.

Technically the API does. We haven’t implemented it client-side, and I don’t think it’s implemented server-side on Scoot.

<https://github.com/googleapis/googleapis/blob/943102cfb5fa8811a15a4c21bd12421aadb423b5/google/devtools/remoteexecution/v1test/remote_execution.proto#L849-L857>