Repository navigation
Release/0.7 - #2897
Closed
tisnik wants to merge 73 commits into
Closed
Release/0.7#2897tisnik wants to merge 73 commits into
tisnik wants to merge 73 commits into
Conversation
Bump version to 0.7.0rc1
Fixed 1 file(s) based on 1 unresolved review comment. Co-authored-by: CodeRabbit <noreply@coderabbit.ai>
Bump version to 0.7.0rc1
LCORE-2945: regenerated openapi schema
LCORE-3396: (for release/0.7) fix token usage in query
…anch LCORE-2277: Point the Tekton pipeline triggers to the release/0.7 branch [release/0.7]
LCORE-2953: updated supported versions
its broken and should not be used. fixed .konflux/requirements
LCORE-2944: Bump CLA version to 0.7.0rc1
Fix Konflux Issues [release/0.7]
…rc1-ec-violations LCORE-2934: Lightspeed Core 0.7.0rc1 - resolve EC violations [release/0.7]
Signed-off-by: red-hat-konflux-kflux-prd-rh02 <190377777+red-hat-konflux-kflux-prd-rh02[bot]@users.noreply.github.com>
…ase/0.7 Update Konflux references
Signed-off-by: red-hat-konflux-kflux-prd-rh02 <190377777+red-hat-konflux-kflux-prd-rh02[bot]@users.noreply.github.com>
…ase/0.7 Update Konflux references
Signed-off-by: red-hat-konflux-kflux-prd-rh02 <190377777+red-hat-konflux-kflux-prd-rh02[bot]@users.noreply.github.com>
Signed-off-by: red-hat-konflux-kflux-prd-rh02 <190377777+red-hat-konflux-kflux-prd-rh02[bot]@users.noreply.github.com>
…ase/0.7 Update Konflux references
LCORE-3523: Lightspeed Core 0.7.0rc2 RPM CVEs
…se/0.7/quay.io-redhat-developer-mapt-0.x Update quay.io/redhat-developer/mapt Docker tag to v0.14.4
Signed-off-by: red-hat-konflux-kflux-prd-rh02 <190377777+red-hat-konflux-kflux-prd-rh02[bot]@users.noreply.github.com>
LCORE-3994: Deprecate Solr vector provider in LCORE 0.7
LCORE-4167: Merge `main` into `release/0.7`
Signed-off-by: red-hat-konflux-kflux-prd-rh02 <190377777+red-hat-konflux-kflux-prd-rh02[bot]@users.noreply.github.com>
…ase/0.7 Update Konflux references
Signed-off-by: red-hat-konflux-kflux-prd-rh02 <190377777+red-hat-konflux-kflux-prd-rh02[bot]@users.noreply.github.com>
…ase/0.7 Update Konflux references
LCORE-3694 moved the removal of the legacy two-file configuration (ogx.library_client_config_path plus an external run.yaml) from 0.7 to 0.8, and the startup warning in src/lightspeed_stack.py says 0.8. Three user-facing places still said 0.7, so the 0.7.0 release would tell users the legacy path is already gone while it keeps working: - README.md, the deprecation warning box - docs/user_doc/deployment_guide.md, the "Do nothing" row of the migration paths table - examples/lightspeed-stack-lls-library.yaml, the comment above ogx.config The spike document keeps its 0.7 wording: it records the schedule as it was decided at the time.
The first commit on this branch corrected three user-facing statements that the legacy two-file configuration is removed in release 0.7. Four more statements were missed because the release number sits on a different line than the word 'removed', so a single-line search did not find them: - README.md: comment in the library-mode container example - docs/user_doc/deployment_guide.md: 'Configuration modes' list - docs/user_doc/deployment_guide.md: 'Deprecation schedule' paragraph - docs/user_doc/deployment_guide.md: warning box in the library-mode container section All four now say 0.8, which matches the startup deprecation warning in src/lightspeed_stack.py, the library_client_config_path field description in src/models/config.py and the design document. Documentation only; no code or behaviour change.
…ion notes The three YAML blocks that illustrate the llama_stack -> ogx key rename carried 'config: profile: default'. 'profile' is a path to a run.yaml-shaped file; 'default' is a value of 'baseline', not of 'profile'. A reader copying the block would get a lookup for a file named 'default' at synthesis time. The section is about the key rename only, so the config block is removed from all three examples.
The startup deprecation warning for the built-in OpenAI provider in 'baseline: default' tells operators to set 'baseline: byo-llm' and to declare their LLM providers under 'inference.providers'. Neither was shown anywhere in the user documentation or in examples/. - deployment_guide.md: new section 'High-level inference providers' with a working snippet, a table of the three baseline values and the deprecation notice (deprecated in 0.7, removed in 0.8). - examples/lightspeed-stack-unified-byo-llm.yaml: a complete single-file configuration that needs no run.yaml. The example was loaded with 'lightspeed_stack.py --dump-configuration' and synthesized with 'ogx_configuration.py'; both succeed and no deprecation warning is logged.
These examples used the deprecated library_client_config_path and logged the deprecation warning at startup: - lightspeed-stack-a2a-state-pg.yaml - lightspeed-stack-a2a-state-sqlite.yaml - lightspeed-stack-azure-entraid-lib.yaml - lightspeed-stack-conversation-cache-pg.yaml - lightspeed-stack-conversation-cache-sqlite.yaml - lightspeed-stack-skills.yaml They now reference the same run.yaml as the synthesis profile through ogx.config.profile. The Azure example pointed at 'examples/azure-run.yaml'. A profile path resolves against the directory of the configuration file, not the working directory, so it is now 'azure-run.yaml'. Each file was synthesized with ogx_configuration.py. The two a2a examples still fail to load for an unrelated, pre-existing reason: they reference agent_card.yaml, which is not shipped in examples/.
… code The design document had drifted from what was merged: - Synthesis inputs: a non-empty vector_store.providers also counts as a synthesis input (since the top-level vector store configuration was added); R3, R11, 'Trigger mechanism' and 'Mode detection' now say so. - Server mode: the entrypoint writes /tmp/generated-run.yaml, not /opt/app-root/run.yaml, and passes -i for the legacy input. - Synthesis steps: the documented order had enrichment before the high-level sections and the native_override merge. The implemented order, since LCORE-3370, is baseline, dedupe, high-level inference, MCP tool runtime (skipped for the empty baseline), native_override, enrichment, dedupe. The step list and the architecture diagram are corrected, and the byo-llm baseline is named in the diagram. - tests/e2e/features/steps/unified_mode.py: the module docstring said five unified-mode feature files; there are four. Not changed: the 'Acceptance test surface' table still names e2e for some requirements that are verified at the integration level.
The example used the deprecated library_client_config_path and logged the deprecation warning at startup. It now references run.yaml as the synthesis profile through ogx.config.profile.
…se-0.7 LCORE-4229: say the legacy two-file configuration is removed in 0.8 (backport to 0.7)
Backport of #2791 from main. A compaction summarizes the older turns and keeps the most recent ones verbatim (buffer_turns). The summary marker is appended to the end of the conversation, so it lands after those kept turns, while every later request rebuilt its context as "summaries + items following the last marker". The kept turns sit before the marker and are in no summary, so from the next request on the model never saw them again. Each marker now records how many leading conversation items its summary covers ("[lightspeed:compaction-summary] [covers:2] <text>") and the boundary comes from that count instead of the marker's position. Marker items are filtered out of the recent turns explicitly. Markers already written carry no count and keep the old behavior, so conversations created before this are unaffected. Scope differs from main: the integration tests and the e2e scenario the original carries do not exist on this branch (LCORE-1574 and the compaction feature file landed after the branch was cut), so this backport is the fix plus its unit tests.
…(backport to 0.7) Backport of #2794 from main. Compaction writes each summary into the conversation as a synthetic user message carrying the [lightspeed:compaction-summary] sentinel. Nothing filtered those on the way back, so GET /v1/conversations/{id} returned them as though the user had typed them: a UI rendering the history shows an internal string attributed to the user, and consumers counting messages or replaying history include a message nobody sent. The conversations API now drops marker items on read. Storage is untouched, because the markers remain the fallback source of truth for deployments without a persisting cache. Scope differs from main: the e2e scenario and the integration tests the original carries do not exist on this branch (they landed after the branch was cut), so this backport is the fix plus its unit tests. Stacked on the LCORE-4219 backport, which touches the same module.
LCORE-3909: leave compaction markers out of the conversation history (backport to 0.7)
LCORE-2976: Sync main to release 0.7
Signed-off-by: red-hat-konflux-kflux-prd-rh02 <190377777+red-hat-konflux-kflux-prd-rh02[bot]@users.noreply.github.com>
…ase/0.7 Update Konflux references
EaaS is deprecated and being replaced by OpenShift-CI. Increase AWS instance size to avoid InsufficientMemory errors when deploying test resources. Assisted-by: Codex GPT-5.6 (cherry picked from commit 267a24f) Signed-off-by: amisstea <amisstea@redhat.com>
Signed-off-by: red-hat-konflux-kflux-prd-rh02 <190377777+red-hat-konflux-kflux-prd-rh02[bot]@users.noreply.github.com>
…ase/0.7 Update Konflux references
Contributor
|
Note Currently processing new changes in this PR. This may take a few minutes, please wait... ⚙️ Run configuration
📒 Files selected for processing (36)
✨ Finishing Touches
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
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
8 participants
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.
Description
Type of change
pyproject.toml+uv.lock]requirements.*.txtfor Konflux]Tools used to create PR
Identify any AI code assistants used in this PR (for transparency and review context)
Related Tickets & Documents
Checklist before requesting a review
Testing