ci: hydra-engine PR validation (build+ctest, server pytest, C# parity) - #81
Open
ddvnguyen wants to merge 1 commit into
Open
ci: hydra-engine PR validation (build+ctest, server pytest, C# parity)#81ddvnguyen wants to merge 1 commit into
ddvnguyen wants to merge 1 commit into
Conversation
…rity) Adds a Hydra-specific PR/push gate for hydra-fork, the test companion to hydra-build.yml. Upstream auto-triggers are disabled fork-wide (ba525ea), so this is the single entry point that validates llama-engine changes on every PR: - build-unit: cmake with GGML_RPC=ON (required for llama-engine to link the ggml-RPC backend), builds llama-engine + llama-server + rpc-server, runs the llama.cpp unit suite (ctest -L main). - server-tests: upstream HTTP pytest suite run against the llama-engine binary. Marked continue-on-error pending a known-good test set. - parity: checks out the public ddvnguyen/hydra_vortex repo (no secrets), builds Tests.EngineParity, drives the engine over HTTP + Hydra RPC. PrefillDecode_HttpAndRpc_ProduceIdenticalKvState gates (ggml-org#469 guard); Decode_HttpAndRpc_ProduceIdenticalLogits is excluded until the framed DECODE async-result path is fixed (traces to #64). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What & why
Adds the missing test/validation CI for the Hydra llama-engine on
hydra-fork. Upstream auto-triggers are disabled fork-wide (ba525ea), so no CI ran on fork PRs — this is the single Hydra-specific validation entry point, the test companion to the existinghydra-build.yml.Jobs (all on GitHub-hosted CPU runners, no GPU)
build-unitGGML_RPC=ON(required for llama-engine to link the ggml-RPC backend), buildsllama-engine+llama-server+rpc-server, runsctest -L mainserver-testsllama-enginebinary (continue-on-error pending known-good set)parityddvnguyen/hydra_vortex(no secrets), buildsTests.EngineParity, drives the engine over HTTP + Hydra RPC;PrefillDecode_HttpAndRpc_ProduceIdenticalKvStategates (ggml-org#469 guard)Notes / follow-ups
Decode_HttpAndRpc_ProduceIdenticalLogitsis excluded from the parity filter until the framed-DECODE async-result path is fixed (traces to feat(rpc): #470 Phase 2 — framed DECODE (0x43) rewrite + HTTP /v1/decode/{id} #64 — invertedllama_state_seq_set_datacheck +/v1/decode/{id}entry storage).server-testsis intentionally not a hard gate yet (upstream suite targets llama-server).Test plan
-DGGML_RPC=ONPrefillDecode_HttpAndRpc_ProduceIdenticalKvStatepasses against a locally-built engine