-
Notifications
You must be signed in to change notification settings - Fork 3.6k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
docs(adr): ADR-044 Guidelines for updating proto defs #9613
Conversation
|
||
## Abstract | ||
|
||
This ADR provides guidelines and recommended practices when updating Protobuf definitions. These guidelines are targeted at module developers. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
this is a good example where it's halfway between ADR and docs.
Talking with some people, we decided to go with an ADR, as it may provide more visibility for other chain & client developers
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
we decided to go with an ADR, as it may provide more visibility for other chain & client developers
I think this makes sense. I'm wondering if we should also be pointing module developers to this ADR somewhere within the building modules documentation. Adding a line or two that mentions best practices when making updates to Protobuf definitions that points to these guidelines might provide a bit more visibility. For example, I think a simple reference to these guidelines could be added in the Msg Service documentation.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Definitely makes sense to me!
|
||
On the other hand, module developers MAY add new fields to Protobuf definitions related to the `Query` service, as the unknown field rejection does not apply to queries. | ||
|
||
#### 2. Fields MAY be marked as `deprecated`, and nodes MAY implement a protocol-breaking change for handling these fields. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
In v0.43, we deprecated a field, added some proto comments around it, and introduced a protocol-breaking change related to this field. Do we want to continue doing so in the future?
An argument could be made that protocol-breaking change => proto pkg bump. Or more generally, what's a "breaking change" which would warrant a proto pkg bump?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Do we want to continue doing so in the future?
Ideally, I think we want to avoid this in the future and a protocol-breaking change should warrant a proto pkg bump but this guideline might be worth keeping as a recovery solution should there be a need again.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I think this is a solid initial draft. Nice work! A few suggestions and comments.
|
||
## Abstract | ||
|
||
This ADR provides guidelines and recommended practices when updating Protobuf definitions. These guidelines are targeted at module developers. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
we decided to go with an ADR, as it may provide more visibility for other chain & client developers
I think this makes sense. I'm wondering if we should also be pointing module developers to this ADR somewhere within the building modules documentation. Adding a line or two that mentions best practices when making updates to Protobuf definitions that points to these guidelines might provide a bit more visibility. For example, I think a simple reference to these guidelines could be added in the Msg Service documentation.
- `UNARY_RPC`: the SDK currently does not support streaming RPCs. | ||
- `COMMENT_FIELD`: the SDK allows fields with no comments. | ||
- `SERVICE_SUFFIX`: we use the `Query` and `Msg` services convention, which don't have the `-Service` suffix. | ||
- `PACKAGE_VERSION_SUFFIX`: some packages, such as `cosmos.crypto.ed25519`, don't use version suffix. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
- `PACKAGE_VERSION_SUFFIX`: some packages, such as `cosmos.crypto.ed25519`, don't use version suffix. | |
- `PACKAGE_VERSION_SUFFIX`: some packages, such as `cosmos.crypto.ed25519`, don't use a version suffix. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Would it make sense to use the version suffix and include the crypto packages in v1? Why was it decided that these packages should not be versioned?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
IIRC we thought these defs should never change. But tbh i'm not opposed to have everything version-suffixed and put crypto as v1.
|
||
On the other hand, module developers MAY add new fields to Protobuf definitions related to the `Query` service, as the unknown field rejection does not apply to queries. | ||
|
||
#### 2. Fields MAY be marked as `deprecated`, and nodes MAY implement a protocol-breaking change for handling these fields. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Do we want to continue doing so in the future?
Ideally, I think we want to avoid this in the future and a protocol-breaking change should warrant a proto pkg bump but this guideline might be worth keeping as a recovery solution should there be a need again.
Co-authored-by: Ryan Christoffersen <[email protected]>
Co-authored-by: Ryan Christoffersen <[email protected]>
Co-authored-by: Ryan Christoffersen <[email protected]>
Co-authored-by: Ryan Christoffersen <[email protected]>
Co-authored-by: Ryan Christoffersen <[email protected]>
Co-authored-by: Ryan Christoffersen <[email protected]>
Co-authored-by: Ryan Christoffersen <[email protected]>
Co-authored-by: Ryan Christoffersen <[email protected]>
Co-authored-by: Ryan Christoffersen <[email protected]>
Co-authored-by: Ryan Christoffersen <[email protected]>
Visit https://dashboard.github.orijtech.com?pr=9613&repo=cosmos%2Fcosmos-sdk to see benchmark details. |
Thanks @ryanchristo for the review! |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Pre approving
- let's use more MUST instead of SHALL
- Let's add "Draft" section to the README and add this ADR there
Also left few suggestions
|
||
#### 1. `Msg`s SHALL NOT have new fields. | ||
|
||
When processing `Msg`s, the SDK's antehandlers are strict and don't allow unknown fields in `Msg`s. This is checked by the unknown field rejection in the [`codec/unknownproto` package](https://github.com/cosmos/cosmos-sdk/blob/master/codec/unknownproto). |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Shall we be more specific here:
- we allow new fields for objects which are saved in the store
- we don't allow new fields for the
Msg
service arguments
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
It's a bit related to #9613 (comment), so I added a new section
Co-authored-by: Robert Zaremba <[email protected]>
Co-authored-by: Robert Zaremba <[email protected]>
Co-authored-by: Robert Zaremba <[email protected]>
Co-authored-by: Robert Zaremba <[email protected]>
Co-authored-by: Robert Zaremba <[email protected]>
Co-authored-by: Robert Zaremba <[email protected]>
Co-authored-by: Robert Zaremba <[email protected]>
Co-authored-by: Robert Zaremba <[email protected]>
Co-authored-by: Robert Zaremba <[email protected]>
Co-authored-by: Robert Zaremba <[email protected]>
Co-authored-by: Robert Zaremba <[email protected]>
I'm going to put automerge on. It's been 1+ month, and have 2 approvals from relevant people, so as a draft it should be okay. Can have another round of review with external parties once this goes into "proposed" |
Description
This ADR is to be merged as "DRAFT" status for now, as the details for the "Bumping Protobuf Package Version" section have not been sorted out yet.
This ADR comes from discussion with @webmaster128 and @robert-zaremba about proto updates strategy. We decided to go for an ADR to document our decision for v0.43, and for visibility for other chains doing proto upgrades.
rendered
Closes: #9477
ref: #9446
ref: #9445
Author Checklist
All items are required. Please add a note to the item if the item is not applicable and
please add links to any relevant follow up issues.
I have...
!
to the type prefix if API or client breaking changeCHANGELOG.md
Reviewers Checklist
All items are required. Please add a note if the item is not applicable and please add
your handle next to the items reviewed if you only reviewed selected items.
I have...
!
in the type prefix if API or client breaking change