Parent: #4894
Motivation
Serves "two client families by persona; ship a decision guide instead of unifying specs." Without this, users will default to whichever client they find first (likely the hand-written Java grpc-client, since it's the only mature one today) rather than the one that actually fits their use case.
Proposed design
- New page in
arcadedb-docs (location TBD — verify the docs repo/site structure during implementation, likely alongside existing driver/client documentation) answering: "HTTP (universal, admin+data+observability, works everywhere fetch works) vs. gRPC (performance data-plane, streaming/bulk ingest, backend languages + Connect-ES for JS)."
- Per-client README for each package produced by this Epic, each following a consistent template (install, auth, quickstart, "which client is this" cross-link).
- Capability/version matrix: a table of {client, language, transport, generated-from, current version, server-release compatibility, streaming support} kept up to date by the release automation issue (ideally auto-generated as part of its release pipeline rather than hand-maintained, to avoid the matrix itself drifting).
Acceptance criteria
Dependencies
Depends on the thin TS client, REST fan-out, Connect-ES, and per-language gRPC client issues existing (needs real clients to document) and benefits from the release automation issue (auto-generating the matrix rather than hand-maintaining it).
Effort estimate
S — primarily writing, but blocked until enough clients exist to compare.
Parent: #4894
Motivation
Serves "two client families by persona; ship a decision guide instead of unifying specs." Without this, users will default to whichever client they find first (likely the hand-written Java
grpc-client, since it's the only mature one today) rather than the one that actually fits their use case.Proposed design
arcadedb-docs(location TBD — verify the docs repo/site structure during implementation, likely alongside existing driver/client documentation) answering: "HTTP (universal, admin+data+observability, works everywherefetchworks) vs. gRPC (performance data-plane, streaming/bulk ingest, backend languages + Connect-ES for JS)."Acceptance criteria
Dependencies
Depends on the thin TS client, REST fan-out, Connect-ES, and per-language gRPC client issues existing (needs real clients to document) and benefits from the release automation issue (auto-generating the matrix rather than hand-maintaining it).
Effort estimate
S — primarily writing, but blocked until enough clients exist to compare.