Bootstrap and automatically update Buildkite pipelines (#8619)
* Bootstrap and automatically update Buildkite pipelines
This means that:
1. postsubmit runs always use with the version of the pipeline
configuration from that commit
2. presubmit runs use with the version of the pipeline configuration
from that commit if the PR comes from the main repo. This
limitation is for security reasons and is why this PR branch is in
the main repo. It limits potentially bypassing our normal checks to
people who already have write access to the repository. Suggestions
for ways to do this that support third party forks but maintains
this security are welcome.
3. Pipelines that we need to register with Buildkite can all be
checked in to source control and updated as part of the commit to
the main branch.
This required modifications to the smooth-checkout plugin, which I am
upstreaming in
https://github.com/hasura/smooth-checkout-buildkite-plugin/pull/25.
This allows testing changes to the presubmit pipeline itself on
presubmit. Of course, it relies on the previous version of the pipeline
being sufficiently compatible that it can successfully bootstrap, but
that shouldn't be too hard.
We now have enough stuff going on here that I created `pipelines/` and
`scripts/` subdirectories. I did not move any of the legacy files (e.g.
samples.yml) to avoid breaking anything.
It's also looking like I should factor out some of the shared Buildkite
Python API usage, but I'd prefer to defer that till a later PR.
Combines work from obsolete PRs
https://github.com/google/iree/pull/8609 and
https://github.com/google/iree/pull/8606. I tried to have this broken
out into two PRs, but testing the presubmit pipeline in its current form
(as registered with Buildkite) doesn't work very well, which is why this
PR exists. It got all messed up because of the directory restructuring.
I can split that out into a separate PR if that's preferred.
* Lint
* Avoid slashes in keys, which are apparently not allowed
* Maybe avoiding reported issues with duplicate keys
* Actually fix step key conflictsIREE (Intermediate Representation Execution Environment, pronounced as “eerie”) is an MLIR-based end-to-end compiler and runtime that lowers Machine Learning (ML) models to a unified IR that scales up to meet the needs of the datacenter and down to satisfy the constraints and special considerations of mobile and edge deployments.
See our website for project details, user guides, and instructions on building from source.
IREE is still in its early phase. We have settled down on the overarching infrastructure and are actively improving various software components as well as project logistics. It is still quite far from ready for everyday use and is made available without any support at the moment. With that said, we welcome any kind of feedback on any communication channels!
See our website for more information.
IREE is licensed under the terms of the Apache 2.0 License with LLVM Exceptions. See LICENSE for more information.