* Fix publish in release-cli.yml
Using latest stable rust
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix publish in release.yml
Use latest stable rust
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Run release-dry-run in merge queue only
Running it twice on PR pipeline and merge queue pipeline only wastes time and causes rate limit.
For PR that modifies release-dry-run and wants to test it, they can temporarily run release-dry-run on PR pipeline.
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
---------
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Add cargo-binstall installation path to GITHUB_PATH in install-from-binstall-release.sh
Fixed#1714
* Add cargo-binstall installation path to GITHUB_PATH in install-from-binstall-release.ps1
* Update install-script.yml to test
* Fix use of env var in install-from-binstall-release.ps1
* Fix job changed-file in ci.yml
* Do not run job e2e-tests on PR
which might not have access to GitHub Token in secrets, if opened by non-maintainer
* Provide GITHUB_TOKEN fallback if the secrets is not accessible
* Fix regex expression passed to sed
* Run e2e-test on x86_64h-apple-darwin
Since it is a target built with `build-std`, we need to make sure it is
runnable.
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix `upgrade-transitive-deps`
Fill in PR description, so that reviewer can just re-open it to run the CI.
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
---------
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix job `changed-files`
Convert `ALL_CHANGED_FILES` to multiple lines before processing
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Run e2e-tests and unit-tests on aarch64-apple-darwin
Since the CI now uses M1 instead of x86_64, built a native executable
instead of relying on rosetta
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Run unit-tests on Linux-only
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
---------
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix job changed-files in ci.yml
When there are multiple crates, it would be in multiple lines, breaking github output
* Fix job changed-files in ci.yml
When no crate is changed, command cut will fail.
* Fix job unit-tests in ci.yml
If noo crate is changed, do not run cargo-nextest.
* Fix ci.yml syntax
* Fix unit test for `GhApiClient`
Set client-side rate limit to 1 request per 200 ms
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Only run `detect-targets` test if changed
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Only run unit tests for crates that are changed
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix job `changed-files`
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix job `changed-files`
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix shell error in job `changed-files`
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* FIx separator for `changed-files`
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix passning of craetes to run
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix passing of `CARGO_NEXTEST_ADDITIONAL_ARGS`
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix deciding which crates to test on windows
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix just recipe `unit-tests`
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Split test into two jobs
so that they will have they own cache (since different feature flags are used).
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
---------
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Optimize binstalk-git-repo-api
Use a dedicated github token in CI
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Speedup CI using `cargo-nextest`
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix release profile override on Windows
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Optimize unit test in binstalk-registry
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Use `secrets.CI_RELEASE_TEST_GITHUB_TOKEN` for just-setup
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix unit testing in justfile
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Add retry on rate limit in unit testing
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Only use `CI_UNIT_TEST_GITHUB_TOKEN` in unit testing
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix `test_get_repo_info`: Retry on rate limit
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix `test_has_release_artifact_and_download_artifacts`
Retry on rate limit
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
---------
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Use secrets,CI_TEST_GITHUB_TOKEN for testing in ci.yml
* Run e2e-test in release-packages.yml using secrets.CI_TEST_GITHUB_TOKEN
* Use CI_RELEASE_TEST_GITHUB_TOKEN in e2e-test in release-packages.yml
* Refactor: Mv leon and leon-macros into another repo
It's moved to https://github.com/cargo-bins/leon
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix CI: Rm `cargo-hack` check for `leon`
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
---------
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* CI: Use runner `macos-14` which uses M1 machine
`macos-14` provides 3-core M1, 7G mem and 14G stroage, while `macos-latest`
(`macos-13` as of writing) provides 4-core intel, 14G mem and 14G
storage.
While the memory is cut by half and core count reduce by 1, I believe
that it would still speedup the CI since M1 is much more powerful than
previous Intel CPU used in `macos-13`.
It would also allow us to run tests on aarch64-apple-darwin and on
universal-apple-darwin, if necessary.
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* CI: Fix `e2e-test-subcrate`: Bump `cargo-audit` ver to v0.18.3
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
---------
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix detect-targets on android targets
By enabling linux fallback on Android.
Also add CI regression test for aarch64-linux-android target.
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix CI: Add job `detect-targets-android-check` as required for merging
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
---------
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix CI: Dry-run `cargo-publish` on crate to be published
If there's no crate to publish, do not dry-run `cargo-publish`.
Only dry-run `cargo-publish` on crate to be published to avoid
`cargo-publish` error on `cargo-binstall` when one of its direct
dependencies is bumped.
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix passing of crate to `release-cli.yml`
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix passing of `needs.pr-info.outpus.crate`
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix `release-dry-run`: Only run if is release or is not PR event
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix passing of crate
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Rm `just-setup` from `release-cli.yml` since it cannot be cached
release is trigged by the closing of PR, so the cache is saved under the
branch `release/*` which cannot be reused.
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix dry-run publish
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
---------
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix `release-cli.yml`: Do not dry run `cargo-publish`
If dependencies of `carog-binstall` is bumped, then `cargo-publish` dry
run will fail since the dependency has not published yet
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Speedup step publish in `release-cli.yml`
Use `just-setup` to cache rust/zig compilation, also
disable all features to reduce dependencies pulled in.
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* `release-cli.yml`: Restore rust cache before ephermal-crate.sh is run
which writes the key to `Cargo.toml` and invalidates the cache.
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
---------
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* CI: Cache zig global_cache to speedup build
This can be helpful in two situations:
- If `cc` is bumped to a new release, then all the `*-sys` crates will
be rebuilt even if their source doesn't change.
Caching zig global_cache would avoid expensive rebuilds.
- For target `x86_64h-apple-darwin`, it uses build-std which means the
rust cache is quite ineffective (only build-dependencies are cached),
so we would also need zig global_cache to avoid rebuilding c
dependencies in CI.
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix missing `shell: bash`
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix CI: Create symlink to `zig`
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* CI: Fix getting output from prev step in just-setup
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* CI: Include job_id `github.job` in zig cache key
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
---------
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix ci: Install deps before caching and include `JUST*` env values in caching key
To make sure caching is done properly and does not accidentally include
(old) pre-built binaries that we don't want.
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix missing `shell`
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix justfile target `check`: Call `cargo hack check` directly
without using the `CARGO=cargo-zigbuild` env since `leon` does not use
external C/C++ compiler anyway.
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* CI: Dowgrade cargo-hack to 0.6.10
This is the last version working
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
---------
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
For PRs opened by contributors who are not part of the cargo-bins team and for
PRs opened by dependabot, they cannot access the secrets, thus signing will
always fail.
Skipping release-dry-run on pull_request would fix that for them, while
retaining the signing stage on merge_queue and on main to ensure that the
release workflow is working.
It will also speedup CI for PR.
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix build again
* Recommend (r)age over minisign password
* Dry-run the entire release process
* Reorg a bit so dry-run works
* Fix secret name
* Add check on age key
* Pass secrets down
* Use a cross-platform "date"
* Delete signing key artifact to be extra safe
* Last little bits
* Revert "release: cargo-binstall v1.4.0 (#1397)"
This reverts commit 99e8256018.
* --allow-dirty on publish
* Remove private key file after generation
* Write public key file to bin crate and to package
* Upload public key file to release
* Sign our releases
* Use secrets instead of artifacts
* And the universal
* Apparently we can’t use secrets like that?
* Minor fixes to doc
* Private key requires untrusted comment
* Dogfood one deeper
* Fix `detect-targets` on ubuntu 20.04, glibc 2.31
Fixed#1375 and fixed#1378
glibc 2.31 does not support `--version`, so we need to detect and
fallback to passing an executable linked with that glibc to check if it
is indeed glibc.
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
* Fix compilation faillure on ubuntu-20.04
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
---------
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
Fixed#1353
Merge all dependabot configuration for cargo into one, with directory
setting to `/` so that it will check for dep updates for entire
workspace while also updating lockfile in PR.
To prevent it from opening too many PRs, it's configured to run on
every Saturday, after `upgrade-transitive-deps.yml` is run.
It will also group all dependencies into one PR to prevent opening too
many PRs.
Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>
