Thank you for your interest in contributing to Tendermint! Before contributing, it may be helpful to understand the goal of the project. The goal of Tendermint is to develop a BFT consensus engine robust enough to support permissionless value-carrying networks. While all contributions are welcome, contributors should bear this goal in mind in deciding if they should target the main tendermint project or a potential fork. When targeting the main Tendermint project, the following process leads to the best chance of landing changes in master.
All work on the code base should be motivated by a Github Issue. Search is a good place start when looking for places to contribute. If you would like to work on an issue which already exists, please indicate so by leaving a comment.
All new contributions should start with a Github Issue. The issue helps capture the problem you're trying to solve and allows for early feedback. Once the issue is created the process can proceed in different directions depending on how well defined the problem and potential solution are. If the change is simple and well understood, maintainers will indicate their support with a heartfelt emoji.
If the issue would benefit from thorough discussion, maintainers may request that you create a Request For Comment. Discussion at the RFC stage will build collective understanding of the dimensions of the problems and help structure conversations around trade-offs.
When the problem is well understood but the solution leads to large structural changes to the code base, these changes should be proposed in the form of an Architectural Decision Record (ADR). The ADR will help build consensus on an overall strategy to ensure the code base maintains coherence in the larger context. If you are not comfortable with writing an ADR, you can open a less-formal issue and the maintainers will help you turn it into an ADR. ADR numbers can be registered here.
When the problem as well as proposed solution are well understood, changes should start with a draft pull request against master. The draft signals that work is underway. When the work is ready for feedback, hitting "Ready for Review" will signal to the maintainers to take a look.
Each stage of the process is aimed at creating feedback cycles which align contributors and maintainers to make sure:
Please note that Go requires code to live under absolute paths, which complicates forking.
While my fork lives at https://github.com/ebuchman/tendermint
,
the code should never exist at $GOPATH/src/github.com/ebuchman/tendermint
.
Instead, we use git remote
to add the fork as a new remote for the original repo,
$GOPATH/src/github.com/tendermint/tendermint
, and do all the work there.
For instance, to create a fork and work on a branch of it, I would:
$GOPATH/src/github.com/tendermint/tendermint
)git remote rename origin upstream
git remote add origin git@github.com:ebuchman/basecoin.git
Now origin
refers to my fork and upstream
refers to the tendermint version.
So I can git push -u origin master
to update my fork, and make pull requests to tendermint from there.
Of course, replace ebuchman
with your git handle.
To pull in updates from the origin repo, run
git fetch upstream
git rebase upstream/master
(or whatever branch you want)We use go modules to manage dependencies.
That said, the master branch of every Tendermint repository should just build
with go get
, which means they should be kept up-to-date with their
dependencies so we can get away with telling people they can just go get
our
software.
Since some dependencies are not under our control, a third party may break our
build, in which case we can fall back on go mod tidy
. Even for dependencies under our control, go helps us to
keep multiple repos in sync as they evolve. Anything with an executable, such
as apps, tools, and the core, should use dep.
Run go list -u -m all
to get a list of dependencies that may not be
up-to-date.
When updating dependencies, please only update the particular dependencies you
need. Instead of running go get -u=patch
, which will update anything,
specify exactly the dependency you want to update, eg.
GO111MODULE=on go get -u github.com/tendermint/go-amino@master
.
We use Protocol Buffers along with gogoproto to generate code for use across Tendermint Core.
For linting and checking breaking changes, we use buf. If you would like to run linting and check if the changes you have made are breaking then you will need to have docker running locally. Then the linting cmd will be make proto-lint
and the breaking changes check will be make proto-check-breaking
.
To generate new stubs based off of your changes you can run make proto-gen
after installing protoc
and gogoproto.
To install protoc
, download an appropriate release (https://github.com/protocolbuffers/protobuf) and then move the provided binaries into your PATH (follow instructions in README included with the download).
To install gogoproto
, do the following:
$ go get github.com/gogo/protobuf/gogoproto
$ cd $GOPATH/pkg/mod/github.com/gogo/protobuf@v1.3.1 # or wherever go get installs things
$ make install
You should now be able to run make proto-gen
from inside the root Tendermint directory to generate new files from proto files.
If you are a Vagrant user, you can get started hacking Tendermint with the commands below.
NOTE: In case you installed Vagrant in 2017, you might need to run
vagrant box update
to upgrade to the latest ubuntu/xenial64
.
vagrant up
vagrant ssh
make test
Every fix, improvement, feature, or breaking change should be made in a
pull-request that includes an update to the CHANGELOG_PENDING.md
file.
Changelog entries should be formatted as follows:
- [module] \#xxx Some description about the change (@contributor)
Here, module
is the part of the code that changed (typically a
top-level Go package), xxx
is the pull-request number, and contributor
is the author/s of the change.
It's also acceptable for xxx
to refer to the relevent issue number, but pull-request
numbers are preferred.
Note this means pull-requests should be opened first so the changelog can then
be updated with the pull-request's number.
There is no need to include the full link, as this will be added
automatically during release. But please include the backslash and pound, eg. \#2313
.
Changelog entries should be ordered alphabetically according to the
module
, and numerically according to the pull-request number.
Changes with multiple classifications should be doubly included (eg. a bug fix that is also a breaking change should be recorded under both).
Breaking changes are further subdivided according to the APIs/users they impact.
Any change that effects multiple APIs/users should be recorded multiply - for
instance, a change to the Blockchain Protocol
that removes a field from the
header should also be recorded under CLI/RPC/Config
since the field will be
removed from the header in rpc responses as well.
The main development branch is master.
Every release is maintained in a release branch named vX.Y.Z
.
Note all pull requests should be squash merged except for merging to a release branch (named vX.Y
). This keeps the commit history clean and makes it
easy to reference the pull request where a change was introduced.
master
master
must never fail make test
master
(except when reverting a broken commit, which should seldom happen)git remote add origin
)CHANGELOG_PENDING.md
to record your changegit rebase
on top of the latest master
master
make test
to ensure that all tests passunstable
branch may be used to aggregate pull merges before fixing testsWe follow the Go style guide on commit messages. Write concise commits that start with the package name and have a description that finishes the sentence "This change modifies Tendermint to...". For example,
``` cmd/debug: execute p.Signal only when p is not nil
[potentially longer description in the body]
Fixes #nnnn ```
Each PR should have one commit once it lands on master
; this can be accomplished by using the "squash and merge" button on Github. Be sure to edit your commit message, though!
master
test_integrations
in Makefile)master
(to be squash merged):
CHANGELOG_PENDING.md
to top of CHANGELOG.md
python ./scripts/linkify_changelog.py CHANGELOG.md
to add links for
all issuesbash ./scripts/authors.sh
to get a list of authors since the latest
release, and add the github aliases of external contributors to the top of
the changelog. To lookup an alias from an email, try bash ./scripts/authors.sh <email>
CHANGELOG_PENDING.md
vX.X
(this will trigger the release vX.X.0
)If there were no breaking changes and you need to create a release nonetheless, the procedure is almost exactly like with a new release above.
The only difference is that in the end you create a pull request against the existing X.X
branch.
The branch name should match the release number you want to create.
Merging this PR will trigger the next release.
For example, if the PR is against an existing 0.34 branch which already contains a v0.34.0 release/tag,
the patch version will be incremented and the created release will be v0.34.1.
All repos should be hooked up to CircleCI.
If they have .go
files in the root directory, they will be automatically
tested by circle using go test -v -race ./...
. If not, they will need a
circle.yml
. Ideally, every repo has a Makefile
that defines make test
and
includes its continuous integration status using a badge in the README.md
.
If you contribute to the RPC endpoints it's important to document your changes in the Swagger file
To test your changes you should install nodejs
and run:
npm i -g dredd
make build-linux build-contract-tests-hooks
make contract-tests
This command will popup a network and check every endpoint against what has been documented