Paul Draper

Right, and my point is that the architecture of the image should default to the Bazel target platform (`--cpu`). Currently, it defaults to amd64 https://github.com/bazelbuild/rules_docker/blob/8e70c6bcb584a15a8fd061ea489b933c0ff344ca/container/image.bzl#L578

A workaround that should work: https://github.com/buchgr/bazel-remote/pull/493#issuecomment-984803190

PnP is pretty neat, as it allows strong flexibility in linking, giving better performance. I actually thing it could be appealing to use in rule_nodejs for everything.

Rediscovered this ticket here. What the heck is Terraform expecting me to do!??!?!?! Compile my plugin for every operating system and platform, just to run it on **my** operating system...

Correct, this request doesn't make sense with the ZIP format. *Writing* a stream makes sense, but *reading* a stream does not.

FYI, AWS only supports DLQ redrive for standard queues, not FIFO queues.

> and will be very small value Okay, thanks. (Personally, I've override the naming to be `_`.)

> there was once a big conversation about the value Surprised 20 was the result....as mentioned, with 40% less, theres 1 in one trillion chance of collision with 100,000 classes...

still relevant

New source has the same issue