A prdoc is like a changelog but for a Pull Request. We use this approach to record changes on a crate level. This information is then processed by the release team to apply the correct crate version bumps and to generate the CHANGELOG of the next release.
When creating a PR, the author needs to decide with the R0-silent
label whether the PR has to contain a prdoc. The
R0
label should only be placed for No-OP changes like correcting a typo in a comment or CI stuff. If unsure, ping
the CODEOWNERS for advice.
A .prdoc
file is a YAML file with a defined structure (ie JSON Schema). Please follow these steps to generate one:
- Install the
prdoc
CLI by runningcargo install parity-prdoc
. - Open a Pull Request and get the PR number.
- Generate the file with
prdoc generate <PR_NUMBER>
. The output filename will be printed. - Optional: Install the
prdoc/schema_user.json
schema in your editor, for example VsCode. - Edit your
.prdoc
file according to the Audience and SemVer sections. - Check your prdoc with
prdoc check -n <PR_NUMBER>
. This is optional since the CI will also check it.
Tip: GitHub CLI and jq can be used to provide the number of your PR to generate the correct file:
prdoc generate $(gh pr view --json number | jq '.number') -o prdoc
Alternatively you can call the prdoc from PR via /cmd prdoc
(see args with /cmd prdoc --help
)
in a comment to PR to trigger it from CI.
Options:
pr
: The PR number to generate the PrDoc for.audience
: The audience of whom the changes may concern.bump
: A default bump level for all crates. The PrDoc will likely need to be edited to reflect the actual changes after generation.force
: Whether to overwrite any existing PrDoc.
While describing a PR, the author needs to consider which audience(s) need to be addressed. The list of valid audiences is described and documented in the JSON schema as follow:
-
Node Dev
: Those who build around the client side code. Alternative client builders, SMOLDOT, those who consume RPCs. These are people who are oblivious to the runtime changes. They only care about the meta-protocol, not the protocol itself. -
Runtime Dev
: All of those who rely on the runtime. A parachain team that is using a pallet. A DApp that is using a pallet. These are people who care about the protocol (WASM), not the meta-protocol (client). -
Node Operator
: Those who don't write any code and only run code. -
Runtime User
: Anyone using the runtime. This can be a token holder or a dev writing a front end for a chain.
If you have a change that affects multiple audiences, you can either list them all, or write multiple sections and re-phrase the changes for each audience.
All published crates that got modified need to have an entry in the crates
section of your PRDoc
. This entry tells
the release team how to bump the crate version prior to the next release. It is very important that this information is
correct, otherwise it could break the code of downstream teams.
The bump can either be major
, minor
, patch
or none
. The three first options are defined by
rust-lang.org, whereas None
should be picked if no other
applies. The None
option is equivalent to the R0-silent
label, but on a crate level. Experimental and private APIs
are exempt from bumping and can be broken at any time. Please read the Crate Section of the RELEASE doc
about them.
Note: There is currently no CI in place to sanity check this information, but should be added soon.
For example when you modified two crates and record the changes:
crates:
- name: frame-example
bump: major
- name: frame-example-pallet
bump: minor
It means that downstream code using frame-example-pallet
is still guaranteed to work as before, while code using
frame-example
might break.
A crate that depends on another crate will automatically inherit its major
bumps. This means that you do not need to
bump a crate that had a SemVer breaking change only from re-exporting another crate with a breaking change.
minor
an patch
bumps do not need to be inherited, since cargo
will automatically update them to the latest
compatible version.