Repository navigation
Conversation
Bumps the `lore` submodule to v0.12.0, sets `LORE_VERSION` to match so `lore-bin` downloads the 0.12.0 release assets, regenerates the bindings and adapts the safe wrappers to the C API changes of 0.10.1 and 0.11.0; 0.12.0 changes nothing in the C API. 0.10.1: every storage `*_ITEM_COMPLETE` event carries an `error` detail where it carried an `error_code`, so the five storage completion events read the code from `error.error_code` and gain a `message`. The code is the failure's own FFI code, so `ErrorCode` names the two codes a store read now runs into that `lore.h` does not: a missing payload (81), which `LoreError::is_not_found` covers alongside a missing address, and the server's plain miss (101), which `Store::resolve` folds to `None` as it does the local store's. `lore_revision_sync_args_t` appends `view`, which `RevisionSyncArgs` passes through. An empty string Lore emits is a null pointer from 0.10.1 on, which `try_to_str` already read as empty; `Event::StorageGetData` now guards its bytes the same way. 0.11.0: `lore_global_args_t` gains `auth_mode` and `trust_remote`, which `GlobalArgs` exposes as `AuthMode` and a flag; `skip_verify` on a storage open is now an alias of `trust_remote` for that handle, which the docs say. `lore_auth_login_with_token_args_t` gains `client_id` and `client_secret`, redacted from `Debug` like the token. `lore_storage_get_resolved_item_t` gains `fragments`, and `lore_auth_url_event_data_t` a `user_code`. `lore_branch_merge_start_args_t` gains two arrays, but nothing here wraps it. The bindings were generated with libclang on Windows and the enum typedefs bindgen emits as signed there were put back to unsigned, which reproduces the committed 0.10.0 file byte for byte from its header. The 0.12.0 release's `lore.h` differs from the tag's only in its version stamp, and the released `lore.dll` loads through these bindings, reports `0.12.0+1967` and round-trips a put/get on an in-memory store. Co-Authored-By: Claude Fable 5.1 <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
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.
Bumps the
loresubmodule to v0.12.0, setsLORE_VERSION = "0.12.0"solore-bindownloads the 0.12.0 release assets (sameliblore-v0.12.0-<triple>.<ext>naming as before), regenerateslore-sysand adapts thelore-rswrappers. breda needs 0.11+ because from there the Lore background service relays storage and revision-tree calls, which is how several breda processes will share one store.C API changes since 0.10.0
0.12.0 changes nothing in the C API; the changes come from 0.10.1 and 0.11.0.
0.10.1
*_ITEM_COMPLETEevent replaceserror_codewith anerrordetail struct (code, message, trace). The five storage completionEventvariants readerror.error_codeand gain amessage.PayloadNotFound(81) instead ofAddressNotFound(80), a short bufferOversized(118), and a mutable-key miss from the serverNotFound(101).lore.hnames only five codes, soErrorCodegainsPayloadNotFoundandNotFoundwith the registry values spelled out;LoreError::is_not_foundcovers both not-found codes andStore::resolvefolds the server's miss toNonelike the local one.lore_string_tnow has aNULLpointer.try_to_stralready handled that;Event::StorageGetDatanow guards a null byte pointer the same way instead of building a slice from it.lore_revision_sync_args_tgainsview, exposed onRevisionSyncArgs.lore-capipackage; the header still lives atlore-capi/lore.h, which is what the generator reads.0.11.0
lore_global_args_tgainsauth_modeandtrust_remote, exposed asGlobalArgs::auth_mode: AuthModeandGlobalArgs::trust_remote. The raw struct is still filled field by field, so every field is set.skip_verifyonlore_storage_open_args_tis now an alias oftrust_remotefor the handle; the docs onStorageOpenArgsandStoreOptionssay so.lore_auth_login_with_token_args_tgainsclient_id/client_secret(secret redacted fromDebug, like the token).lore_storage_get_resolved_item_tgainsfragments;lore_auth_url_event_data_tgainsuser_code;lore_event_tgainsLORE_EVENT_FILE_AUDITand other new events, which reach a callback asEvent::Other.lore_branch_merge_start_args_tgainslink_paths/link_messages; nothing here wraps it.Bindings
Generated on Windows (no Linux at hand) with the repo's generator, then the enum typedefs that libclang emits as signed on MSVC were put back to
c_uint. That procedure reproduces the committed 0.10.0bindings.rsbyte for byte from the 0.10.0 header, so the 0.12.0 output should match what thegenerate-bindingsCI job produces on Linux. 311 functions now (263 before).Verified beyond CI
cargo fmt --check,cargo clippy --workspace --all-targets -- -D warnings,cargo test --workspace(478 + 30 + 1 tests) all pass locally on Windows; the repo has nocargo denyorshearstep.cargo test -p lore-bin -- --ignoreddownloadsliblore-v0.12.0-x86_64-pc-windows-msvc.zipfrom the GitHub release.lore.hshipped in that release archive is identical to the submodule's atv0.12.0apart fromLORE_INTERFACE_VERSION(0.12.0vs0.11.1-nightly).lore.dllthroughlore_rs::load(which requires all 311 symbols to resolve), reports0.12.0+1967, opens an in-memoryStore, round-trips aput/get, and a read of an absent address comes back asLoreError::Failedwithis_not_found() == trueand the message now carried by the error detail.Not covered: no run against a real server, so the
NotFound(101) mapping for a remote mutable-key miss andtrust_remoteare taken from the release notes and header rather than observed.🤖 Generated with Claude Code