You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@seatunnel.apache.org by "davidzollo (via GitHub)" <gi...@apache.org> on 2026/07/11 10:54:56 UTC
[I] [Discussion][Connector-V2] Introduce a minimal connector capability contract [seatunnel]
davidzollo opened a new issue, #11403:
URL: https://github.com/apache/seatunnel/issues/11403
## Background
SeaTunnel can already describe connector configuration rules, but it still cannot clearly describe a more important production question:
**What does a connector actually promise at runtime?**
Examples include whether a connector supports CDC, schema evolution, checkpoint restore, multi-table behavior, or only partial/best-effort semantics.
## Problem to solve
As the connector ecosystem grows, the long-term problem is no longer connector count alone. The harder problem is connector governance.
Without a structured capability contract:
- users cannot tell what a connector truly supports
- pre-run validation cannot detect capability mismatch early
- quality classification becomes subjective
- documentation can drift away from real behavior
## Proposal
Introduce a minimal connector capability contract that can be consumed by validation, documentation, and later connector quality governance.
Suggested first-delivery scope:
- keep the model intentionally small
- start from a small set of high-value capabilities
- allow gradual connector adoption instead of all-at-once migration
Candidate capability groups for v1:
- execution mode support
- CDC support
- schema evolution support
- checkpoint/restore support
- multi-table support
- support-level or exactness boundary where needed
## Why this should be discussion-first
This topic touches:
- connector API expectations
- validation flow
- documentation generation possibilities
- long-term connector governance and certification
If the capability model is too large, the community will not maintain it. If it is too weak, it will not solve the real problem.
## Non-goals for v1
- do not design a huge YAML-first system
- do not require every historical connector to implement everything immediately
- do not turn this into a docs-only label system
## Suggested validation
- validate that at least a small group of trial connectors can express the model cleanly
- validate that capability mismatch can be detected before job execution
- validate that the contract is small enough for connector authors to maintain honestly
## Acceptance criteria
- a minimal capability model is clearly defined
- at least a few trial connectors can adopt it without excessive boilerplate
- the model can be consumed by validation or governance logic instead of being documentation-only
- the community agrees on what belongs in v1 and what should be deferred
## Related issues
- #10247
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: commits-unsubscribe@seatunnel.apache.org.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [I] [Discussion][Connector-V2] Introduce a minimal connector capability contract [seatunnel]
Posted by "davidzollo (via GitHub)" <gi...@apache.org>.
davidzollo closed issue #11403: [Discussion][Connector-V2] Introduce a minimal connector capability contract
URL: https://github.com/apache/seatunnel/issues/11403
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: commits-unsubscribe@seatunnel.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [I] [Discussion][Connector-V2] Introduce a minimal connector capability contract [seatunnel]
Posted by "DanielLeens (via GitHub)" <gi...@apache.org>.
DanielLeens commented on issue #11403:
URL: https://github.com/apache/seatunnel/issues/11403#issuecomment-4949417312
This is an important design topic, and I agree it is a better foundation than trying to solve connector governance only through documentation or ad hoc labels.
The most important thing for v1 is to keep the capability contract both small and honest. If the model becomes too large, connector authors will not maintain it. If it is too vague, validation and governance will not gain anything useful from it.
For a first pass, I would keep two boundaries especially clear:
1. which capabilities are stable runtime promises versus best-effort hints
2. which capability groups are important enough to make early mismatch detection materially better for users
A good v1 scope would be a small set such as execution mode support, CDC support, schema evolution support, checkpoint/restore support, and multi-table support, plus an explicit support-level boundary where exact semantics differ.
I would also treat gradual adoption as a hard requirement. The design should let a few trial connectors prove the model first, and only then expand, instead of forcing a big-bang migration across the full connector ecosystem.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: commits-unsubscribe@seatunnel.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [I] [Discussion][Connector-V2] Introduce a minimal connector capability contract [seatunnel]
Posted by "davidzollo (via GitHub)" <gi...@apache.org>.
davidzollo commented on issue #11403:
URL: https://github.com/apache/seatunnel/issues/11403#issuecomment-5078707386
Meaningless for now
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: commits-unsubscribe@seatunnel.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org