RAJA Portability Suite Coordinated Releases¶
The RAJA Portability Suite does coordinated releases, meaning that when a non-patch release is done for one, a new version release is done for all Suite projects at the same time. When a coordinated release is done, common dependencies, such as BLT and camp, are set to the same versions in all Suite projects.
Individual projects in the RAJA Portability Suite may do patch releases (to fix bugs, etc.) independently of other Suite projects.
Release Candidate Branch¶
A release candidate branch is a temporary branch used to finalize a release.
When the features, documentation, bug fixes, etc. to include in a release are
complete and merged into the develop branch, a release candidate branch is made
from the develop branch. Typically, a release candidate branch is named
rc-<release name>, or similar. Please see RAJA Release Process for a
description of how a release candidate branch is used in the release process.
Finalizing a release on a release candidate branch involves the following steps:
Complete the release notes describing the release in the
RELEASE_NOTES.mdfile. Describe all API changes, notable new features, bug fixes, improvements, build changes, etc. included in the release in appropriately labeled sections of the file.
Please follow the pattern established in the release notes file used for previous releases.
All changes that users need to be aware of should be documented in the release notes. Hopefully, the release notes file has been updated along with the corresponding changes in PRs that are merged into the develop branch. Regardless, it is good practice to look over the commit history since the last release to ensure all important changes are documented in the release notes.
Update the version number entries in the code. The top-level
CMakeLists.txtfile must be changed, where the entries:
RAJA_VERSION_PATCHLEVELare defined. These items are used to define corresponding macro values in the
include/RAJA/config.hppfile when the code is configured so that users can access and check the RAJA version in their code by including that header file. The
releasefields in the
RAJA/docs/conf.pyfile must also be changed to the new release number. This information is used in the online RAJA documentation.
No feature development is done on a release branch. Only bug fixes, release documentation, and other release-oriented changes are made on a release candidate branch.
RAJA Release Process¶
The RAJA release process includes the following sequence of steps:
Identify all work to be to be included in the release, such as new features, improvements, and bug fixes.
Merge each PR containing work to be included in the release into the develop branch after it is reviewed and approved by the team, and passes all the CI checks.
Make a release candidate branch from the develop branch.
Creation of the release candidate branch begins the next release cycle. While the release candidate branch is being finalized, work can continue on the develop branch.
Finalize the release on the release candidate branch by completing remaining release tasks on it. See Release Candidate Branch for typical tasks to complete.
Make a PR to merge the release candidate branch into the main branch when the release candidate branch is ready for the release.
Since the main branch only changes when a release is made, the release candidate PR will likely contain many modifications. Fortunately, the vast majority of those changes will have been reviewed, approved, and merged into the develop branch.
While the release candidate PR targets main, opening a companion draft PR that targets develop will make it easier for team members to review the changes not yet merged into develop. Cross-reference the two PRs (release candidate and draft) in their descriptions, and tell reviewers to review the draft PR into develop, but approve the release candidate PR to merge into main. We will merge main into develop after the release to get the missing changes from the release.
Merge the release candidate branch into the RAJA main branch when it is approved and all CI checks pass.
Create the release on GitHub.
Choose the RAJA main branch as the release target and the option to create the release tag when the release is published. The release tag is the name of the release.
Set the release name (and associated git tag name) following the convention established for prior releases. Specifically, the tag label should have the format
vYYYY.mm.pp. See RAJA Release Version Naming for more description of the version naming scheme we use.
Fill in the release description. Note key features, bugfixes, etc. included in the release. The description should summarize the relevant items in the
RELEASE_NOTES.mdfile in the release candidate branch that was merged. Also, add a note to the release description to remind users to download the gzipped tarfile named for the release (see below) instead of the assets GitHub creates for a release. The assets created by GitHub do not contain the RAJA submodules and may cause issues for users as a result.
For consistency, please follow a similar release description pattern for all RAJA releases.
Publish the release when it is ready by clicking the button on GitHub.
Generate a release tarfile. Check out the main branch locally and make sure it is up-to-date. Then, run the script:./scripts/make_release_tarball.sh
from the top-level RAJA directory. The script strips out the Git files from the code and generates a tarfile whose name contains the release tag name in the top-level RAJA directory of your local repository. If this is successful, the name of the generated gzipped tarfile will not contain extraneous SHA-1 hash information. If it does, you need to make sure that your local repo checkout is at the same commit as the release tag. To do this, run the command:git checkout <release tag name>
in your local clone of the repository.
Edit the release in GitHub and upload the tarfile to the release.
Lastly, make a PR to merge the main branch into the develop branch. After it passes all CI checks and is approved, merge the PR. This will ensure that all changes done to finalize the release are included in the develop branch.
After a RAJA release is done, there are other tasks that typically need to be performed to update content in other projects. These tasks are described in Post-release Activities.
A Hotfix branch is used in the (hopefully!) rare event that a bug is found shortly after a release that may negatively impact RAJA users. A hotfix branch will address the issue in both the develop and main branches.
A hotfix branch is treated like a release candidate branch and it is used to generate a patch release following the same basic process that is described in RAJA Release Process.
For completeness, the key steps for performing a hotfix (patch) release are:
- Make a hotfix branch from main at the buggy release tag (hotfix/<issue>), fix the issue on the branch and verify, testing against user code if necessary. Update the release notes and RAJA patch version number as described in Release Candidate Branch.
- When the hotfix branch is ready, make a PR for it to be merged into the main branch. When that is approved and all CI checks pass, merge it into the RAJA main branch.
- On GitHub, make a new release with a tag for the release. Following our convention, the tag label should have the format
YYYY.mm.pp, where only the patch portion of the release tag should differ from the last release. In the GitHub release description, note that the release is a bugfix release and describe the issue(s) that it resolves. Also, add a note to the release description to download the gzipped tarfile for the release rather than the assets GitHub creates as part of the release.
- Check out the main branch locally and make sure it is up-to-date. Then, generate the tarfile for the release by running the script
./scripts/make_release_tarball.shfrom the top-level RAJA directory. If this is successful, a gzipped tarfile whose name includes the release tag with no extraneous SHA-1 hash information will be in the top-level RAJA directory.
- Make a PR to merge the main branch back into the develop branch. After it passes all CI checks and is approved, merge the PR. This will ensure that changes for the bugfix will be included in future development on develop.
After a RAJA release is complete, other tasks are performed to update content in other repositories, typically. These tasks include:
- Update the RAJAProxies project to the new RAJA Portability Suite project release. This typically consists of updating the submodules to the new RAJA Portability Suite project versions, making sure the proxy-apps build and run correctly. When this is done, tag a release for proxy-app project.
- Update the RAJA Template Project project to the new RAJA release.
- Update the RAJA Spack package in the Spack repository. This requires some knowledge of Spack and attention to details and Spack conventions. Please see Spack Package Update for details.
Typically, we also do a new release of the RAJA Performance Suite project after completing a RAJA release. This involves updating the RAJA and BLT submodules to match the RAJA release and follows the same process as RAJA Release Process.
Spack Package Update¶
The Spack package is used in RAJA GitLab CI testing and also by RAJA users who use Spack to manage their third party library installations. The RAJA Spack package that we use in our GitLab CI resides in the RADIUSS Spack Configs submodule. Typically, users will use the Spack package in the Spack repo. We maintain the RAJA Spack package in the Spack project to be as close as possible to the one in the RADIUSS Spack Configs project, which may contain minor modifications specific to our GitLab CI testing.
Like all Spack packages, the RAJA package is a file containing a Python class. The following list contains a description of items to update.
Add a new RAJA version when a release is made. Near the beginning of the
Raja classdefinition, you will find a list of versions that identify RAJA releases as well as items for the
mainbranches. Adding a new RAJA version is done by adding a line, such as:version("2022.10.3", tag="v2022.10.3", submodules=False)
The last entry indicates whether Spack will use RAJA’s submodules when it builds RAJA. Currently, we do not use the submodules by default and allow Spack to manage the installation of RAJA dependencies.
Add new (build) variants as needed. The
variantitems identify how to specify RAJA build variations in a
Spack spec. For example, the RAJA build variant to enable desul atomics is defined by the line:variant("desul", default=False, description="Build Desul Atomics backend")
For each variant, there may be an entry in the file to enable the corresponding CMake option in the CMake cache, such as:entries.append(cmake_cache_option("RAJA_ENABLE_DESUL_ATOMICS", "+desul" in spec))
There may also be additional options needed. For example, desul also requires that C++ 14 (at least) is enabled for the build. Such information may appear as:if "+desul" in spec: entries.append(cmake_cache_string("BLT_CXX_STD","c++14")) if "+cuda" in spec: entries.append(cmake_cache_string("CMAKE_CUDA_STANDARD", "14"))
When a variant is defined properly, it can be enabled in a Spack spec using the shorthand name in the
variantline. For example, to enable desul atomics in a Spack build of RAJA, one can include:+desul
in the Spack spec.
Add new TPL version constraints and package entries as needed. For example, RAJA depends on BLT to configure a build and the 0.5.2 version of BLT is used for all RAJA versions greater than 2022.10.0. This dependency and version constraint is expressed in the package file as:depends_on("firstname.lastname@example.org:", type="build", when="@2022.10.0:")
In the Spack package file, you will see similar version constraint specifications for RAJA camp and CMake dependencies as well as others.
Add or update configuration package entries as needed. In addition to the TPL version constraints, there are lines in the package file that specify which CMake variables are used to pass options to a CMake configuration. For example, the CMake variables that indicate the location of BLT and camp to use for a RAJA build are specified on the lines:entries.append(cmake_cache_path("BLT_SOURCE_DIR", spec["blt"].prefix))
and:if "camp" in self.spec: entries.append(cmake_cache_path("camp_DIR", spec["camp"].prefix))
Information that applies to specific build variants, CMake variables, etc. should be specified in the appropriate Python class function implementation in the package file. Specifically,
initconfig_compiler_entriesfunction contains compiler options
initconfig_hardware_entriesfunction contains options hardware-based RAJA back-end support
initconfig_package_entriesfunction contains options for TPLs and build variants that are not specific to a compiler or hardware
One final point is worth noting. We try to add known conflicts to our Spack package as early as we can. For example, enabling OpenMP in a ROCm compiler build for HIP is only allowed in recent ROCm releases. So we include this conflict in our Spack package:
depends_on("rocprim", when="+rocm") with when("+rocm @0.12.0:"): .... conflicts("+openmp", when="@:2022.03")
This helps users avoid unknown conflicts and potential build or runtime failures.
It is good practice to add known conflicts to the Spack package as soon as we know about them.