Syncing main branch with the upstream repo - #33
Closed
AryanSuvarna wants to merge 1169 commits into
Closed
AryanSuvarna wants to merge 1169 commits into
AryanSuvarna wants to merge 1169 commits into
Conversation
…n. It does compile, but it does not work. FossilOrigin-Name: fa06977b6db7fa745720561ec0b10570cf7e71598dc7a7c5ee650640e5bdf6f5
FossilOrigin-Name: 80461e0d724963aaf2646005298f1194c5f1c4c9ae41c1085d4d137ed485bd9f
…turns a BLOB. FossilOrigin-Name: fe65821a3b912f061026e6fd7174be26897010e6b474e2780350cac60faebaad
…sides an opportunity to negotiate a suitable protocol number, for future compatibility. Send the page size as a power-of-two. FossilOrigin-Name: df0623aae1154281157409f62d6d3fb3ce41829281d53bc55868ce44b3d36883
…mote side, so that if the remote is the replica, it will have access to the origin database name in case the replica name is really a directory. FossilOrigin-Name: 435c30171d3c6073b7aaf5cc11cc4813f6a2d225ae6dce1b0e478f0cd5a0b532
debugging. FossilOrigin-Name: b979d02ffd1370d8840328bce06c76c224f0fc1fb54b47d6c904547580a820a1
FossilOrigin-Name: e385525793c7d74ce8ee139c9d6cfc1248834754598f3fd45b22b9426ff106ee
… make the arg handling more robust. FossilOrigin-Name: 53fb9b11807ff7accd8cd41f9cb6516d2503f161ea976940437a1d3aae868665
FossilOrigin-Name: b2a3497e5525dd33faf70961107a0529f476735fef756953c66e105747271c6d
PTRMAP entries that extend off the end of the database, as long as they appear to be well-formatted. FossilOrigin-Name: a9f95fe5ce90ab9864165e603f3a34013c3c98d03f1db689996f4a32086e2ed6
FossilOrigin-Name: 75d5a8eb3d4ece06900109ad4022ba2a3e82de2f0acb012e3a02bfb4326bfa6d
…invalid entries that are within the range of the database file. Continue to ignore invalid entries beyond the end of the database file. FossilOrigin-Name: 4cad385b90eaca2d90e3375e473472145af4134160b81097a8535d06638c2e4a
of 160. This uses about 1/3rd fewer CPU cycles. FossilOrigin-Name: 96c7f47a8f59e5078bd296979421c1b57fbcb7be261f8a7a0b1d22a4b5914db0
corrects that to just 6 rounds. FossilOrigin-Name: 3c36f5814f25483586c4fd49ef2fe5c7c0ff8c59672b1622c92061ec0ba8547a
FossilOrigin-Name: e55e3e8ec2fe3a9190872d999cee55c85bde92667040cc166233faaa2fa34266
…1 feature, into this branch. FossilOrigin-Name: d2f0d19936222911bc317efecc831007d3aba81f9b32877030ffb29d1728bbdc
…forum post, "sqlite3_analyzer not closing WAL-mode db cleanly" ].) FossilOrigin-Name: 94ceac98845e31249b656dcdb8a58f456b9212dc83968ea333852a66d72a0dae
FossilOrigin-Name: 86e0219d977c493ac19d00c3ddcf560eb317d506c7cf6e4ef17e92daa91e1762
…toring the values of fts5 UNINDEXED column belonging to contentless tables. FossilOrigin-Name: c51dc2a5e75baacbd905cf314e7b1a58a81993ff05ca656739e028d7db25d5b2
ATTACH-ed databases within the same transaction. FossilOrigin-Name: 6aa9c8e79b440c6419e65990d9ceba8f00a6f975455138cf2aa82b113daec825
FossilOrigin-Name: dff76b7a3436031ea5a61b8a44ddfa1d40ea20c983f3d34a8501cd7074db68b8
…s it is with regular rsync. Only show the SSH command with two or more -v options, or if there is an error in popen2(). FossilOrigin-Name: 105ec44b470318fc9ff1773027c4064343f224068c9b6e71c5618f18f7dfcc3f
FossilOrigin-Name: 452fb6de3984c3cb10d30b51dcdb2574578ca128a0c519b2bd43df0bdd343083
FossilOrigin-Name: 30e1b92d5663e24d2f325f2bab35f81b55848ef39d15688e40b9005269626303
…ons. FossilOrigin-Name: fc05a5b7f77cdbfcc659d49eb09569a64a172362cb90199e2861028085178f10
If page 1 changes, always send it last. FossilOrigin-Name: 2d8cd76691554578e987ce682cf0c42c083711dd1511a178148978182ef43ba2
FossilOrigin-Name: 9961334c8007e7cb6ae55885075b74acddc4fa701b359cf67e0f3c237d7eba4a
FossilOrigin-Name: 129aca54f6b791c222b51f3eb01569e1e569269860e153b005140eb65af378b9
…that can extend or truncate the database. Add the sqlite3-rsync utility program that make a copy of a live database over SSH. FossilOrigin-Name: b7a8ce4c8c5fc6a3b4744d412d96f99d2452eb4086ad84472511da3b4d6afec6
…d by the prior check-in. FossilOrigin-Name: 50762ba0783a04e0dcd9456a1ae17d875b0a9272f2f09854a23d9d5253761e9f
The SQLITE_USER_AUTHENTICATION deprecation #warning (added upstream in SQLite 3.46.0) is a hard error under MSVC's default traditional C preprocessor (error C1021: invalid preprocessor command 'warning'), breaking Windows builds. SQLite3MultipleCiphers force-enables SQLITE_USER_AUTHENTICATION by default, so the directive is active even though build.rs passes -DSQLITE_USER_AUTHENTICATION=OFF. Guard the directive with !defined(_MSC_VER) so the deprecation notice is kept for GCC/Clang and skipped on MSVC. This is purely a compile-time diagnostic; the extension's runtime behavior is unchanged on all platforms. Upstream removed this extension entirely in SQLite 3.48.0 (commit bc4df60), so this local patch becomes moot once the bundled SQLite is updated past 3.47.0. The patch is applied in libsql-sqlite3/src and mirrored into both regenerated bundled amalgamations.
The Windows job's `cargo build -p libsql --all-features` was being built with the `x86_64-pc-windows-gnu` toolchain (MinGW/GCC), set as the default host triple by `hecrj/setup-rust-action@v2`. MSVC was never exercised. So breaks that only fail under MSVC -- like the SQLite 3.47.0 `#warning` that errors with C1021 -- compiled green here. We discovered the gap when libsql-js (which builds with `--target x86_64-pc-windows-msvc`) failed downstream on the same code. Add `--target x86_64-pc-windows-msvc --release` so the Windows job actually compiles the bundled SQLite encryption amalgamation (sqlite3mc) with MSVC, and breaks surface in CI. `cargo clean -p libsql-ffi` runs first so sqlite3mc is rebuilt from source rather than restored from a cached target/.
The Windows job's `cargo build -p libsql --all-features` was being built with the `x86_64-pc-windows-gnu` toolchain (MinGW/GCC), set as the default host triple by `hecrj/setup-rust-action@v2`. MSVC was never exercised. So breaks that only fail under MSVC -- like the SQLite 3.47.0 `#warning` that errors with C1021 -- compiled green here. We discovered the gap when libsql-js (which builds with `--target x86_64-pc-windows-msvc`) failed downstream on the same code. Add `--target x86_64-pc-windows-msvc --release` so the Windows job actually compiles the bundled SQLite encryption amalgamation (sqlite3mc) with MSVC, and breaks surface in CI. `cargo clean -p libsql-ffi` runs first so sqlite3mc is rebuilt from source rather than restored from a cached target/.
This updates the vendored SQLite3 Multiple Ciphers encryption extension from 1.8.1 to 1.9.0, the release that targets upstream SQLite 3.47.0 (the base version we now ship). The previous 1.8.1 vendoring lagged the SQLite base and was missing several upstream fixes, including a crash in sqlite3mcSetCodec(). The import replaces all upstream files with their v1.9.0 contents and re-applies libSQL's local patches via 3-way merge (base v1.8.1, ours libSQL, theirs v1.9.0). All merges were clean (no conflicts), and `cargo build -p libsql-ffi --features multiple-ciphers` builds. == Preserved as-is (not upstream sqlite3mc) == src/sqlite3.c, src/sqlite3.h — libSQL's own SQLite amalgamation, not sqlite3mc's vanilla copy. build.rs overwrites src/sqlite3.c with libsql-sqlite3's amalgamation at build time regardless, so these are left untouched. == Re-applied libSQL patches (3-way merged onto 1.9.0) == CMakeLists.txt — libSQL build options (LIBSQL_ENCRYPTION, LIBSQL_CUSTOM_PAGER_CODEC, LIBSQL_EXTRA_PRAGMAS, LIBSQL_ENABLE_WASM_RUNTIME), AES256-only cipher selection, and the arm/aarch64 guards around -msse4.2/-maes. src/sqlite3mc.c — `#include "sqlite3.c"` instead of "sqlite3patched.c" so the amalgamation pulls in libSQL's SQLite. src/codecext.c, src/sqlite3mc_vfs.c — libSQL codec hooks libsql_db_has_codec(), libsql_pager_codec_impl() (replacing sqlite3mcPagerHasCodec()/sqlite3mcPagerCodec()), and the cached hasCodec update in the rekey success path. src/cipher_config.c — libsql_extra_pragma(). src/cipher_wxaes256.c — libsql_generate_aes256_key(). src/codec_algos.c — libsql_generate_initial_vector(). == Notable upstream fixes now picked up == src/sqlite3mc_vfs.c — sqlite3mcSetCodec() now passes the VFS located by mcFindVfs() (pVfsMC) to mcFindDbMainFileName() instead of blindly casting db->pVfs. When the Multiple Ciphers VFS is not the top-level VFS (another VFS stacked on top), db->pVfs is not an sqlite3mc_vfs and the cast made mcFindDbMainFileName() dereference a bogus ->mutex and crash. This matches the fix already present in sqlite3mcGetCodec(). src/sqlite3mc_vfs.c — the mcRead* helpers now propagate codec errors via sqlite3mcGetCodecLastError() instead of silently resetting rc to SQLITE_OK after decryption.
libsql_stmt_interrupt() set a per-statement isInterrupted flag, but the VDBE execution loop only ever checked the connection-wide db->u1.isInterrupted. A statement already executing inside sqlite3_step() therefore ran to completion regardless of the request; the flag was only observed at the next step() entry. Check p->isInterrupted alongside db->u1.isInterrupted at both VDBE interrupt-check sites so an interrupt requested mid-execution aborts the running statement promptly with SQLITE_INTERRUPT, without touching the connection-wide interrupt state (other statements keep running). Set and clear the flag atomically, mirroring sqlite3_interrupt(), since the request may come from another thread. Tests: - test/interruptstmt.test: deterministic regression via a new sqlite_stmt_interrupt_count test hook, covering the in-loop check and the connection-flag-stays-clear property. - test/interrupttest.c: standalone multi-threaded test of the real cross-thread case (interrupt a step() in flight) and statement-level granularity. Build and run with `make interrupttest`.
The 0.10.0-pre.3 version bump (61d629a) updated the workspace Cargo.toml versions but not Cargo.lock, which stayed at 0.10.0-pre.2. Any cargo invocation (including the c-bundle-validate CI job's 'cargo xtask build-bundled') rewrites the lockfile to match, leaving an uncommitted Cargo.lock change that fails the job's 'git diff --quiet' check. Regenerate the lockfile so it matches the committed Cargo.toml.
Two one-line typo fixes for duplicated "the" in `libsql-server` doc-comments: - `libsql-server/src/main.rs` — "By default, the the period is 30 seconds." → "By default, the period is 30 seconds." - `libsql-server/src/rpc/streaming_exec.rs` — "/// Apply the response to the the builder, and return whether..." → "/// Apply the response to the builder, ..." No code/behavior change.
…2248) libsql_stmt_interrupt() set a per-statement isInterrupted flag, but the VDBE execution loop only ever checked the connection-wide db->u1.isInterrupted. A statement already executing inside sqlite3_step() therefore ran to completion regardless of the request; the flag was only observed at the next step() entry. Check p->isInterrupted alongside db->u1.isInterrupted at both VDBE interrupt-check sites so an interrupt requested mid-execution aborts the running statement promptly with SQLITE_INTERRUPT, without touching the connection-wide interrupt state (other statements keep running). Set and clear the flag atomically, mirroring sqlite3_interrupt(), since the request may come from another thread. Tests: - test/interruptstmt.test: deterministic regression via a new sqlite_stmt_interrupt_count test hook, covering the in-loop check and the connection-flag-stays-clear property. - test/interrupttest.c: standalone multi-threaded test of the real cross-thread case (interrupt a step() in flight) and statement-level granularity. Build and run with `make interrupttest`.
The windows-latest runner image moved to windows-2025-vs2026, which ships Visual Studio 2026. The cmake crate 0.1.54 pinned in Cargo.lock does not recognize VS 2026 and panics in the libsql-ffi build script with "couldn't determine visual studio generator" while configuring the SQLite3MultipleCiphers build. Support for the VS 2026 generator was added in cmake 0.1.55, so bump the lockfile to 0.1.58 (pulling in the newer cc and libc it requires).
The windows-latest runner image moved to windows-2025-vs2026, which ships Visual Studio 2026. The cmake crate 0.1.54 pinned in Cargo.lock does not recognize VS 2026 and panics in the libsql-ffi build script with "couldn't determine visual studio generator" while configuring the SQLite3MultipleCiphers build. Support for the VS 2026 generator was added in cmake 0.1.55, so bump the lockfile to 0.1.58 (pulling in the newer cc and libc it requires).
The setup-protoc action downloads protoc via unauthenticated GitHub API requests, which share a rate limit across all runners on the same IP and intermittently fail with "API rate limit exceeded". Our proto files are all plain proto3, so the protobuf-compiler package from the Ubuntu archive is sufficient and involves no GitHub API at all.
The setup-protoc action downloads protoc via unauthenticated GitHub API requests, which share a rate limit across all runners on the same IP and intermittently fail with "API rate limit exceeded". Our proto files are all plain proto3, so the protobuf-compiler package from the Ubuntu archive is sufficient and involves no GitHub API at all.
AryanSuvarna
marked this pull request as ready for review
September 29, 2026 14:51
AryanSuvarna
marked this pull request as draft
September 29, 2026 14:53
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
4 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.
No description provided.