Repository navigation
Merge dev into main - #1329
Merged
Merged
Merge dev into main#1329
Conversation
Merge main into dev
* Keep completed job work from repeating after a writeback failure Signed-off-by: Floating-Y <118035379+Floating-Y@users.noreply.github.com> * Link writeback recovery to its implementation review and validation Signed-off-by: Floating-Y <118035379+Floating-Y@users.noreply.github.com> * Remove duplicated writeback checks while preserving regression coverage Signed-off-by: Floating-Y <118035379+Floating-Y@users.noreply.github.com> * Keep writeback recovery in CI while trimming duplicated acceptance evidence Signed-off-by: Floating-Y <118035379+Floating-Y@users.noreply.github.com> --------- Signed-off-by: Floating-Y <118035379+Floating-Y@users.noreply.github.com>
Signed-off-by: Floating-Y <118035379+Floating-Y@users.noreply.github.com>
…clared ends (#1206) Signed-off-by: Virtue039 <virtue039@qq.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
…essing job together (#1204) Signed-off-by: Virtue039 <virtue039@qq.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
…back (#1211) #1204 gave multipart uploads their first version row. The JSON push (POST /kbs/{id}/ingest) creates its document through documents::create, which still wrote none, so GET /documents/{id}/content answered 500 for every pushed document. The document and version 1 are now written in one transaction there too. Documents pushed before this are repaired by migration 0111. Signed-off-by: Wayland Yang <wayland0916@gmail.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Floating-Y <118035379+Floating-Y@users.noreply.github.com>
…sion (#1217) Signed-off-by: Floating-Y <118035379+Floating-Y@users.noreply.github.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
… cases (#1202) * Measure both namesake failure paths with fixed model replies in CI Signed-off-by: Virtue039 <virtue039@qq.com> * Give namesake fixtures distinguishing clues and require their named failures to remain observable Signed-off-by: Virtue039 <virtue039@qq.com> --------- Signed-off-by: Virtue039 <virtue039@qq.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
…f a question mark (#1220) * A literal value reaches the adjudicator and the review card instead of a question mark A fact whose object is a value (a title, an employee number, an amount) has no entity on the other end, so its line read "phrase → ?". The adjudicator and the review card both lost it; on the identity bench about half of all live facts are values. The line now shows the value, cut to 120 characters. The two adjudicator rows of the scripted namesake cases (#1202) that measured this leave the known-failure list. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Signed-off-by: Wayland Yang <wayland0916@gmail.com> * The review page test expects the literal value on its line Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Signed-off-by: Wayland Yang <wayland0916@gmail.com> --------- Signed-off-by: Wayland Yang <wayland0916@gmail.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
#1221) * Tests: Snowflake and Databricks error / lifecycle / edge paths via wiremock Signed-off-by: Royce Lam <roycelam@umich.edu> * Format the Snowflake and Databricks error-path tests Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Signed-off-by: Wayland Yang <wayland0916@gmail.com> --------- Signed-off-by: Royce Lam <roycelam@umich.edu> Signed-off-by: Wayland Yang <wayland0916@gmail.com> Co-authored-by: Royce Lam <roycelam@umich.edu> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
The middle tab lists relations only (kind = relation) but used the Properties label, which reads 属性 in Chinese, the same as the Attributes tab beside it: a user saw 属性 9 and 属性 2 and could not tell them apart. It now uses the existing Relations label (关系). Signed-off-by: Wayland Yang <wayland0916@gmail.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…tons say relation (#1225) The attributes tab of the ontology table had no button: the attribute dialog took its class from the class panel that opened it, so the only way in was to open a class first, and nothing on the tab said so. A user looking at it asked whether attributes could be added by hand at all. The tab now has New attribute; opened from there, the dialog asks which class the attribute belongs to and will not save without one. The relations tab, its button and the relation dialog used the Properties strings, which read 属性 in Chinese, the word the attributes tab uses. They now say 关系 / Relations (the edit title already did). Signed-off-by: Wayland Yang <wayland0916@gmail.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…ys at twelve (#1226) Signed-off-by: Wayland Yang <wayland0916@gmail.com> Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
Signed-off-by: Zhoulu <ldazzh@lineadirecta.es> Co-authored-by: Zhoulu <ldazzh@lineadirecta.es>
* A system admin can grant the flag to another user (#1233) * Revoking your own system admin flag is refused (#1233) * Granting a deactivated or unknown user is not found (#1233) * PUT /admin/users/{id} grants or revokes system admin, admins only, audited (#1233) * Member dialog: a system admin can switch another user's system admin flag (#1233) * Route test: only a system admin grants the flag, and the grant is audited (#1233) * cargo fmt
Signed-off-by: Floating-Y <118035379+Floating-Y@users.noreply.github.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
) * Specify cases that expect no property binding before changing their contract Signed-off-by: Virtue039 <virtue039@qq.com> * Expose negative case conflicts with human bindings in a concise contract Signed-off-by: Virtue039 <virtue039@qq.com> --------- Signed-off-by: Virtue039 <virtue039@qq.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
…1240) Signed-off-by: dada-yan <BinjunYann@gmail.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
…1241) Signed-off-by: dada-yan <BinjunYann@gmail.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
) * Add an export button to the ontology sidebar for the existing Turtle endpoint Signed-off-by: Zhoulu <ldazzh@lineadirecta.es> * Move the export button from Ontology to KbSettings and clarify that it exports the whole knowledge base Signed-off-by: Zhoulu <ldazzh@lineadirecta.es> --------- Signed-off-by: Zhoulu <ldazzh@lineadirecta.es> Co-authored-by: Zhoulu <ldazzh@lineadirecta.es> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
* A write token may set up a base Five routes take a personal token with write scope as well as a session, acting as the token's owner: creating a base in a workspace, creating a source, reading a source's push key, importing an ontology, and deciding a kind word. Listing a workspace's bases takes any token. The person's role is still the ceiling, a read token is refused with token_read_only, and a token limited by kb_ids reaches only those bases and creates none (token_kb_out_of_scope). One extractor, SessionOrToken, takes either credential; AuthUser still reads every bearer as a JWT. 0014 gains decision 6. Closes #1243 Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> * A token creates only statements sources, and audit rows name the token A write token may create only a statements source; any other kind is refused with token_source_kind, since url, rss and custom sources make the server fetch on a schedule. source.created, alignment.kind_word_decided and ontology.imported carry detail.token_id when a token acted, so an admin can see what a token did before revoking it. Decision 6 of 0014 says so, and that a system admin's write token acts as Owner in every base. Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> --------- Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
* The ontology decides a pushed phrase it declares After a pushed document's extraction, a signature whose phrase equals the key or label of exactly one declared relation, with the subject's class in its domain and the object's class in its range, binds without the model, recorded as decided_by 'ontology'. Materialize is enqueued unless queued, so the typed graph follows the push in a base with no chat model. A person's decision is never overwritten, a near match or an ambiguous one goes to the aligner, and the aligner leaves an ontology decision alone while the relation still declares it. Migration 0113 admits 'ontology' in phrase_bindings.decided_by; 0044 gains a revision under decision 3. Closes #1242 Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> * The ontology leaves state relations to the aligner A state relation bound by the ontology carried no marks, and neither the aligner nor the review queue would ever reach it. declared() now skips temporal = 'state', so the aligner binds it and asks for marks. Tests: a pushed phrase naming a state relation gets no ontology binding, and an ontology binding is re-decided once its relation is narrowed or renamed. 0044's revision corrects the votes it records, says a prose statement sharing a pushed signature is bound with it, and says how a binding whose relation stops matching is re-decided. Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> --------- Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
…rd sign-in can be turned off (#1237) * A verified email in an allowed domain is the one an SSO login may provision Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> * An SSO login finds its account by verified email, or opens a viewer account with no password Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> * SSO logins in an allowed domain link or open their account (UTOPIA_OIDC_ALLOWED_DOMAINS) Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> * UTOPIA_PASSWORD_LOGIN=false refuses password sign-in and registration Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> * Login page shows only SSO when password sign-in is off Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> * Document UTOPIA_OIDC_ALLOWED_DOMAINS and UTOPIA_PASSWORD_LOGIN Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> * Address review on SSO provisioning (#1236) - refuse provisioning when a deactivated account holds the email; open the account and link the subject in one transaction - the server ignores UTOPIA_PASSWORD_LOGIN=false when OIDC is not configured, and warns at startup when no one unlinked could get in - ADR 0069 records the trust assumption; 0066 points to it - email_verified accepts true or "true" only - unparseable password hashes still run argon2 against the dummy - name claim trimmed, capped at 64, falls back to the local part - exchange no longer returns the internal refusal reason - password_login_disabled and an SSO-only unlinked message in en/zh Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> --------- Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
…'s chat draft, conversation or answers (#1238) * The next account signed in on a tab does not see the previous account's chat draft, conversation or answers Signed-off-by: dada-yan <BinjunYann@gmail.com> * A tab clears its chat state when another account signs in, not on every sign-in Signed-off-by: dada-yan <BinjunYann@gmail.com> --------- Signed-off-by: dada-yan <BinjunYann@gmail.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
…base no longer reads another base's quotes (#1261) Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com>
…irst documents (#1256) Signed-off-by: dada-yan <BinjunYann@gmail.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
…ecision together (#1257) * Adopting a proposal commits the element, its follow-up work and the decision together Signed-off-by: dada-yan <BinjunYann@gmail.com> * Name the pull request in the 0061 revision Signed-off-by: dada-yan <BinjunYann@gmail.com> --------- Signed-off-by: dada-yan <BinjunYann@gmail.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
Signed-off-by: dada-yan <BinjunYann@gmail.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
…ations (#1254) Signed-off-by: dada-yan <BinjunYann@gmail.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
…overlap (#1259) * Keep an active system administrator when account removals overlap Signed-off-by: Floating-Y <118035379+Floating-Y@users.noreply.github.com> * Serialize administrator removals without blocking foreign key checks Signed-off-by: Floating-Y <118035379+Floating-Y@users.noreply.github.com> --------- Signed-off-by: Floating-Y <118035379+Floating-Y@users.noreply.github.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
* The importer reads utopia:temporal on a property The export has written utopia:temporal on every property that is not a state since 0031, and the importer never read it back: an ontology that said a relation is an event came in as a state, and the rule of #1242, which binds a pushed phrase by the declaration only when the relation is not a state, could never apply to an imported vocabulary. The turtle projection reads urn:utopia:ns:temporal with one of state, event or eternal; any other value is reported under the preview's unprojected predicates and the property stays unset. A new relation is created with the value, a re-import updates an existing relation's temporal to what the file says, and a file that says nothing leaves it alone, so a value a person set on the page survives an unannotated re-import. Attributes stay state. Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> * The import reports a bad utopia:temporal under its IRI, the export writes state too, and an import that changes temporal moves updated_at Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> --------- Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
* An end moved to the end of its period says so on the row Signed-off-by: Maya Kid <BinjunYann@gmail.com> * Name the pull request and the close-by-date issue in the 0053 revision Signed-off-by: Maya Kid <BinjunYann@gmail.com> * Leave the backfill to its own pull request and keep the mark's rule in one helper Signed-off-by: Maya Kid <BinjunYann@gmail.com> --------- Signed-off-by: Maya Kid <BinjunYann@gmail.com>
…s given, and the source cleanup reads only that base's statements. (#1307) Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com>
…choose between candidates (#1309) Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
…ures only for keys not already settled (#1310) Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
… seconds instead of three times a second (#1311) Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
… reaches, so a service mirroring into Utopia can read its own base back without a person's session (#1263) Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
Signed-off-by: Floating-Y <118035379+Floating-Y@users.noreply.github.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
* Propose that a statement's role word follows its newest live evidence instead of the first value written, as of reads included Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> * Narrow 0070 to supersession, back-fill role words onto every evidence chunk, give a role word its chunk's record time, and point 0037 at it Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> * Rename 0070 to say that a role word speaks while its chunk is live, now that no ordering decides between live rows Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> --------- Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
… base switcher, the chat scope chip, the token picker and My knowledge bases (#1250) Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
…ges (#1314) Signed-off-by: Aschenbath <Aschenbath@users.noreply.github.com> Co-authored-by: Aschenbath <Aschenbath@users.noreply.github.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
Signed-off-by: Aschenbath <Aschenbath@users.noreply.github.com> Co-authored-by: Aschenbath <Aschenbath@users.noreply.github.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
* A push that names its external_id keeps that id's document, so identical content under another id is reported as a duplicate instead of renaming the first, vanishing as unchanged, or failing the push Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> * Tombstone then push renames a declared id, and a duplicate answer leaves no restore or missing-flag change behind Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> --------- Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
…n words such as line ranges and amounts are no longer stored but never shown (#1272) Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
…, as containment and type drift already do (#1308) Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…aiting (#1304) * The inference scheduler queues a base again only when it has no job waiting Signed-off-by: Quintin Botes <quintinbotes@outlook.com> * The inference scheduling test runs in a binary of its own, since it queues every due base in the database Signed-off-by: Quintin Botes <quintinbotes@outlook.com> --------- Signed-off-by: Quintin Botes <quintinbotes@outlook.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
…on it (#1305) * Raising a model's concurrency limit admits the tasks already waiting on it Signed-off-by: Quintin Botes <quintinbotes@outlook.com> * A changed model limit applies when it is saved, and raising one only adds permits Signed-off-by: Quintin Botes <quintinbotes@outlook.com> --------- Signed-off-by: Quintin Botes <quintinbotes@outlook.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
…#1318) Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
#1320) Signed-off-by: Bogdan Szabo <bogdan@re-cinq.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
Signed-off-by: Aschenbath <Aschenbath@users.noreply.github.com> Co-authored-by: Aschenbath <Aschenbath@users.noreply.github.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
Signed-off-by: Aschenbath <Aschenbath@users.noreply.github.com> Co-authored-by: Aschenbath <Aschenbath@users.noreply.github.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
… to the token ceiling (#1306) * A reply that keeps repeating itself comes back cut instead of running to the token ceiling Signed-off-by: Quintin Botes <quintinbotes@outlook.com> * Only a whole-row loop is cut, and the cut is reported as repetition rather than the token ceiling Signed-off-by: Quintin Botes <quintinbotes@outlook.com> --------- Signed-off-by: Quintin Botes <quintinbotes@outlook.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
…behind the extraction backlog (#1324) Signed-off-by: Quintin Botes <quintinbotes@outlook.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
… the files that changed since the last sync (#1325) Signed-off-by: Quintin Botes <quintinbotes@outlook.com> Co-authored-by: WaylandYang <145302500+WaylandYang@users.noreply.github.com>
Merged
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.
Sync: everything landed on
devsince the last sync (#1197), the last being #1325.🤖 Generated with Claude Code