stephan
|
40f8f74946
|
OPFS SAHPool VFS: fix an improbable corner case in which the VFS's in-memory state and its on-disk state could differ with regards to availability of a given SAH if a write failure happens at one very specific point during de-registration of the SAH. Addresses [bugs:432f260aa7|bug report 2026-10-05T17:51:35Z].
FossilOrigin-Name: 837fee6e45464bd3fcb5064cfdb91ec15d7cf0c010106797e28e744c3cb39bbc
|
2026-10-05 20:26:24 +00:00 |
|
stephan
|
0caf6b532d
|
SAHPool VFS tweaks contributed off-list by Jure Rotar: hardening against malicious inputs and I/O errors. Truncate to the shorter length when overwriting a db file with a shorter one via importDb(). Correct xOpen() to honor the combination of CREATE/EXCLUSIVE flags.
FossilOrigin-Name: 7af745868b873daacdbb57621201887f80782cef4e10ef24243a15ae88897f0d
|
2026-10-01 08:13:58 +00:00 |
|
stephan
|
ca9a8adfa4
|
Add tests demonstrating kvvfs locking errors across multiple connections to the same storage. Add sqlite3.oo1.Stmt.resetNoThrow() for reasons described in its API docs.
FossilOrigin-Name: 0fed9bcbe2c33d8db5b3500b506bcdfa6a2c718a65bfef4f1d2d21951f6341a8
|
2026-09-30 19:26:14 +00:00 |
|
stephan
|
c15f5adadb
|
Rework the JS kvvfs xLock(), xUnlock(), xCheckReserved(), and xSleep() analog to the recent opfs-sahpool changes to make it more robust when a given kvvfs store is accessed via multiple handles (all necessarily from the same thread). This does not break anything but also does not prove anything - tests are pending.
FossilOrigin-Name: d0e27fa55b05832dc2ad52d5bdb56e74dae2d95984e0f8f8d1831015c5fcd781
|
2026-09-30 18:40:09 +00:00 |
|
stephan
|
519e4c23a5
|
[fa44643f65d6] inadvertently removed its circular-cleanup protection by fixing the finalizer ordering. This check-in fixes that. Spotted in [forum:ba4cb328a7|forum post 2026-09-30T14:59:13Z].
FossilOrigin-Name: 771faec269d786f01fc8d9155fd546d09ea6ca2eba0f0c97e34e795f64719454
|
2026-09-30 15:46:02 +00:00 |
|
stephan
|
cdbfe6a938
|
Do not inherit the OPFS SAHPool's xSleep method from the default VFS, as doing so can cause the thread to freeze. This VFS only deals with contention originating in the same thread, so xSleep() needs to be a no-op. Addresses [forum:3f0794c5d8|bug report 2026-09-30T12:52:54Z].
FossilOrigin-Name: c9dd4d88e44a9eae4ea11d0ce5491c7ac2ac2d60a3bc83ed485812234be374da
|
2026-09-30 13:23:56 +00:00 |
|
stephan
|
3b28aa1ce7
|
Fix the OPFS SAHPool VFS's xLock/xUnlock to properly deal with multiple handles to the same underlying file within the same JS thread. Addresses [forum:5e66a600ba6e57de|bug report 2026-09-30T10:49:15Z]. Also (unrelated but code-adjacent) make sure that the VFS releases all of its internal mappings when it's paused.
FossilOrigin-Name: 9e2caaa382cee5d5df26ad3af7d727ab3f68a912e39655ffcc52f592d5a08a8d
|
2026-09-30 11:48:29 +00:00 |
|
stephan
|
737c12a04c
|
Minor JS doc correction.
FossilOrigin-Name: 6b470917b2c9ab25d1e7fed56f19f37b5c2fc3f2a3d504a752a7bb2db14a3ee6
|
2026-09-30 10:36:27 +00:00 |
|
stephan
|
9fd5a0bc55
|
Correct the xCheckReserveLock() impl of the OPFS SAHPool VFS to fix a corruption case involving multiple handles in the same JS thread to the same filename when one leaves a journal behind. Addresses [forum:b2fbb61642|bug report 2026-09-29T18:25:28Z].
FossilOrigin-Name: 9168a6f1bedba87117169d096f9a6fcadaac65249d707193e4bcbc25e59ccb54
|
2026-09-30 09:53:09 +00:00 |
|
stephan
|
226a2c0d6a
|
Generic internal JS cleanups. No functional changes. Update the preprocessor for the ability to use C-style comments in directive arguments.
FossilOrigin-Name: b2699eb9676e6f5459b8279661c264ba2c803ae133f85c703984deb7afa4fe60
|
2026-09-30 09:17:12 +00:00 |
|
stephan
|
ad1bbfc2bd
|
OPFS SAHPool: more cleanly handle an init failure. Add an init option which tells it not to wipe the storage if initialization fails. Based on the discussion and patch in [forum:5664cd4baee50236|forum post 5664cd4b]. Minor doc updates in unrelated code.
FossilOrigin-Name: 0560c3cc20f60caee42492e4d2063aed8d8d44b0a544a8c3aae56ee67da2e4a7
|
2026-09-29 08:31:55 +00:00 |
|
stephan
|
998173efcd
|
Update some long-dormant JS test code for refactoring-/renaming breakage induced since it was last used. Remove a separate long-dormant JS demo which cannot work with the current code.
FossilOrigin-Name: fc4e246dc26f3e3872dde42ceefcea70ba4158a92ba94361656b2ca0f4e3ef92
|
2026-09-26 06:23:08 +00:00 |
|
stephan
|
a8d1f1ff12
|
Fix a small potential buffer overwrite in the JS impl of kvvfs to resolve [bugs:8d4fe4cc4c|bug report 2026-09-23T10:50:46Z]. Remove some adjacent dead code.
FossilOrigin-Name: c43287d8855b3f9a4621545793ecb2a89ef456d563d8377f2b243812e6acb2b3
|
2026-09-23 14:00:07 +00:00 |
|
stephan
|
ae9c436852
|
User new, more compare JS syntax in a couple of places. No functional changes.
FossilOrigin-Name: 0d747adadf56a0aa3bd5666954a9f7c75c394254dbb938f28ddcd987ffff01b2
|
2026-09-19 08:13:27 +00:00 |
|
stephan
|
46c7da50b3
|
Minor internal API doc update.
FossilOrigin-Name: 346c56e02c39858b4364e4c8b8e4c88237e64470066ebb2687ae06d2b0e978b8
|
2026-09-18 08:00:34 +00:00 |
|
stephan
|
d457639847
|
Performance improvements in some of the JS sqlite3.oo1 APIs, based on analysis and suggestions from Jure Rotar.
FossilOrigin-Name: 8c6c586c3e42dd1d0af4f374b01a660bfc6afb47fd3b34cd7ef16c198ec66d0a
|
2026-09-18 07:56:22 +00:00 |
|
stephan
|
3d0687bade
|
Disable Module.instantiateWasm() in node builds to match branch-3.53's behavior. Fix a sqlite3_backup test (duplicate finalize). The aforementioned world-breaking bug was caused by my browser having gotten into a funky state and needing a restart.
FossilOrigin-Name: 7d77bb088641881fca7a625d9828503d269d0ca1e245583a2dcf24a4f09390f5
|
2026-08-10 12:35:49 +00:00 |
|
stephan
|
961cc1ec09
|
Add tests for the sqlite3_backup JS bindings but they are disabled while tracking down an unrelated problem which prevents those tests from working (most of the sqlite3.capi methods are being mis-mapped to the sqlite3InitModule function for reasons unimaginable).
FossilOrigin-Name: cb610a8a87cfd29d70ab6b825b4a43d44fd801feecfda1bf4fb7ae181806d84b
|
2026-08-10 12:18:07 +00:00 |
|
stephan
|
46441db7e8
|
Merge trunk into the js-backup-bindings branch.
FossilOrigin-Name: 692c4e8ad81f142a454c2038fcdc83317dbaeb71dd6d0497edb4621297c2e35c
|
2026-08-10 11:23:29 +00:00 |
|
stephan
|
d0afe9c60b
|
Lower the range of inputs accepted by the pstack allocator to address [bugs:4dbd096fe335f974dbd096fe3|/bugs/4dbd096fe335].
FossilOrigin-Name: fafefb59481530255a881f22b1f75b587731867d00eabea39eaec7da2199fdff
|
2026-08-07 05:31:08 +00:00 |
|
stephan
|
77ef17960c
|
Add long-missing sqlite3_backup bindings to JS/WASM, in response to [forum:dc96ed27b3fdb202|forum thread dc96ed27b3]. This does not yet include tests.
FossilOrigin-Name: 90e599757416059044c643121631fc56eab2b6eea76d04b2d99eae49b8adef44
|
2026-07-29 12:51:52 +00:00 |
|
stephan
|
76a15f2b95
|
Factor an extranous local var out of the previous check-in.
FossilOrigin-Name: 422b4303dae214e23a53b181e5cdf7888d159da33e1675fd8317cbe66834b78e
|
2026-07-26 15:56:06 +00:00 |
|
stephan
|
bdde38eb0c
|
Remove some extraneous I/O in kvvfs. Reported off-list by Jure Rotar.
FossilOrigin-Name: ce09bfe406728312761a9f227b3f82cbee7cc176dddac1a499e10480a7ee2c1c
|
2026-07-26 15:47:56 +00:00 |
|
stephan
|
fdb802581c
|
Fix a JS container API misuse in sqlite3.oo1.DB.close() which caused its fallback cleanup of still-open statements to not actually do so. Reported off-list by Jure Rotar.
FossilOrigin-Name: d5d8ee76d337820f765cea7b279f2e5644a24d4f0e76d53b35b4d52277edb767
|
2026-07-26 15:39:18 +00:00 |
|
stephan
|
7d1796e488
|
Add (#base64 -f filename) support to the preprocessor and slightly simplify the wasm embedding with it. Fix a bug in the preprocessor's @token@ call expansion when the length of the configured @ is greater than 1 byte.
FossilOrigin-Name: 1e90bedec99054f4856bfb89efc4d676486504db67bd1304c45d14c638ddfc7e
|
2026-07-22 14:06:41 +00:00 |
|
stephan
|
093ca6b1af
|
Consolidate a duplicated code block and ensure that the new filtering of the final JS files makes them read/write so that rebuilds without a clean between them work.
FossilOrigin-Name: 8b013184e42dbd7cd47c07bbb1edbb1f088dfb4efe540bfbc52a2e3c472c7b2d
|
2026-07-21 16:12:06 +00:00 |
|
stephan
|
1f2bc0ea64
|
Add support for (make base64.wasm=1), which embeds sqlite3.wasm into its corresponding JS file as base64. A very expensive calculator which was fed the resulting file sizes assures me that the two-file approach is, in the aggregate, more efficient both in terms of overall network bandwidth (+~20% bigger for base64) and client-side memory usage (streaming vs non-streaming wasm compilation), so we'll shelve this as an interesting experiment or maybe for use by custom builds.
FossilOrigin-Name: a167aee75198ff08b20b448fecd3267ed4a70af2472c02605318a5613f446fc6
|
2026-07-21 16:04:10 +00:00 |
|
stephan
|
bee0db8c60
|
This embeds sqlite3.wasm into sqlite3.js/mjs as base64 but is applied globally - it still needs to be made configurable.
FossilOrigin-Name: 0467a0f826b5b7bb65d9c74df80c7443f01824c1cc8184417849a74791f491b7
|
2026-07-21 13:28:52 +00:00 |
|
stephan
|
379e111bff
|
An initial attempt at embedding sqlite3.wasm into sqlite3.js using base64. This produces a chicken-egg situation in the build because the input file where the blob belongs is injested by emscripten (which generates the wasm file from which that blob is to be generated), so it does not work as-is but is being set aside for later. We'll need to add a second preprocessing step which uses a different #directive delimiter to keep it hidden inside the first pass, then squeeze the wasm file in through there.
FossilOrigin-Name: 538199cf62f07a2fe7ee5775c9f01aff1644d638bc0916cf631d61376462b5cb
|
2026-07-21 12:57:48 +00:00 |
|
stephan
|
886d2c743e
|
Add worker1 to the list of omit-api=X values, per downstream request. Add omit-api=all shorthand.
FossilOrigin-Name: 97ab248b4dd30de76bfbc74ab5cf087df9cd14fd38eb4369a3eadc75953b8c73
|
2026-06-29 11:36:44 +00:00 |
|
stephan
|
bf98a1f6b6
|
Teach the JS build to filter out various APIs/VFSes when passed omit-api='api names...'. Fully trimmed this saves just shy of 300kb from the JS. Make loud=1 mode distinctly louder when preprocessing.
FossilOrigin-Name: 2603c855795fc9466522b1e10433efd5e30975e68ee891680f27f06f3189390b
|
2026-06-24 09:09:02 +00:00 |
|
stephan
|
3fa3e3942e
|
Add preprocessor guards to various JS files working towards the ability to build, e.g., with only one of various VFSes included. The makefile does not yet integrate these - this is an incremental step towards addressing the discussion in [https://github.com/sqlite/sqlite-wasm/pull/168|the downstream npm project's PR#168].
FossilOrigin-Name: f6618a693b7b71f4b846fac257100d3a516dcda687b039d24bcaf29dd1c7454b
|
2026-06-23 20:42:48 +00:00 |
|
stephan
|
190034d268
|
Correct misuse of a loop control variable in writing of a kvvfs journal. Bug report [bugs:20e208fe17|2026-06-20T08:49:41Z]. Remove a snippet of dead code in the JS kvvfs side.
FossilOrigin-Name: a396d7c54c5f251daaaac1e68321c2a54d3b6969599027c29b7ca7abe7fb8c4e
|
2026-06-20 11:07:04 +00:00 |
|
stephan
|
b1bbff20a0
|
Revert the native impl of kvvfs's xOpen() (as distinct from the JS impl which the wasm build uses) to the historical db name restrictions of 'local' or 'session', failing with SQLITE_CANTOPEN if passed another name. [ec866b04d088e53b] overhauled support for kvvfs db names in JS but it turns out that the native impl still relies on those names in order to match journals to databases. Correct a related too-lenient assert() in the WASM pieces.
FossilOrigin-Name: 4b3ec30c63e3824163a2e6cacceb9630301b61f3ffd2b80f8a7b234f09bc251b
|
2026-06-15 12:36:24 +00:00 |
|
stephan
|
6f842b8da4
|
For the various OPFS VFS db-import routines, in SEE builds (only) relax the is-this-a-db check to allow for SEE dbs not having the usual header string in cleartext. Addresses [https://sqlite.org/see/forumpost/f84bef3552|SEE forum post f84bef3552].
FossilOrigin-Name: a5a9ac7ad6622b5b07193aa279deefde7d59ca17f81aba9dd78ee90718446993
|
2026-05-05 08:51:10 +00:00 |
|
stephan
|
d3abfd77f5
|
Reformulate how the OPFS async proxy Worker is loaded for the bundler-friendly builds to work around a bug in Vite in which it incompatibly rewrites the Worker's URL arguments. Addresses the discussion at [https://github.com/sqlite/sqlite-wasm/pull/159|npm package PR 159] and patch validated by Jure Rotar.
FossilOrigin-Name: ec877a30f5c5db30b9154745140744c5cc80d89673d79e50901d7377599cf2b4
|
2026-04-21 10:19:38 +00:00 |
|
stephan
|
9d950aa13b
|
A protype of .prompt support in fiddle but its utility there is in doubt because .open does not work there, so there's very little dynamic info worth showing in a prompt. Also, ANSI colorization can't work there, nor can we colorize the prompts using HTML because of limitations in jquery.terminal.
FossilOrigin-Name: 3b17500dc6338df23d1ed0572cd71593da5a9b048731f73242e27b9a76553a27
|
2026-04-10 23:05:41 +00:00 |
|
stephan
|
35f8abcfd0
|
API doc correction for sqlite3_js_retry_busy(). No code changes.
FossilOrigin-Name: 749c537812b907164b7e7ad1fddc1aee8106c59875d2686fe83d602cd0e07611
|
2026-04-10 17:55:31 +00:00 |
|
stephan
|
d4b74b9fac
|
Add some comments to the OPFS async proxy about potential future experimentation. No code changes.
FossilOrigin-Name: 9c14ee74305a5db3be95dbd9ae8d1db0abf39b323001be4e2f02b7226e6ca0d3
|
2026-04-04 08:07:25 +00:00 |
|
stephan
|
eb8edc0bac
|
Doc cleanups in opfs-wl. No code changes.
FossilOrigin-Name: 009b576dcd5fccaaa52459dbfaa1b2a6fcf6d372e4b8d63553dfa88aab9a759c
|
2026-04-01 11:39:13 +00:00 |
|
stephan
|
46a692c08b
|
Upgrade ext/wasm/c-pp-lite.c to its newer sibling because we've reached the older one's limits. This renames c-pp-lite.c to libcmpp.c to maintain both the SCM- and code lineage but that may end up looking a bit weird because the diff between the two is vast.
FossilOrigin-Name: 2e2339bd9e4293bad04ece7673a3048b99c2143cf9573ade2ec082d95744b981
|
2026-03-08 16:03:08 +00:00 |
|
stephan
|
02cbc96d09
|
JS doc updates.
FossilOrigin-Name: 699295f84ef2c4da363f13126abde6af07697d0959896cd868b14513d4278964
|
2026-03-08 12:42:34 +00:00 |
|
stephan
|
976b50620a
|
Merge opfs-wl into trunk, not so much for the "opfs-wl" VFS, which is stil tentative (working fine but with little perceived benefit), but for the "opfs" VFS cleanups which necessarily happened around it.
FossilOrigin-Name: bcaa876c6dafee0995d239f59b6ce7810b6f5faa219b9c392966211ae3dec17a
|
2026-03-08 08:32:22 +00:00 |
|
stephan
|
9358070cb3
|
Expose sqlite3_bind_zeroblob() to JS/WASM.
FossilOrigin-Name: 0c5d4fe0bd322b4c8dbd0800001317a21cd511b00244de74b4db3fd051153879
|
2026-03-08 05:53:47 +00:00 |
|
stephan
|
237fce8205
|
Extend the JS pre-bootstrapping configuration to include an option to disable inclusion of any given extension VFS (not the default VFSes, like memdb). The primary motivation for this is to give people who don't use OPFS to a way to keep the OPFS VFSes from loading their proxy workers, which are expensive.
FossilOrigin-Name: 9a07eb7a941479510891d1444aacbeb440efaad3e9a13f186833618d8b60a8da
|
2026-03-07 06:10:29 +00:00 |
|
stephan
|
78649bb421
|
Fix the mangling of the opfs async proxy script's URI caused by both OPFS VFSes modifying it to accound for sqlite3.dir. Add a link to speedtest1 with opfw-wl to index.html.
FossilOrigin-Name: c1cb61f473a11a325ef421ba8edfb20257203688565f7db62309044be183af5f
|
2026-03-07 04:19:08 +00:00 |
|
stephan
|
7d10030d75
|
Cleanups and docs in the opfs vfs and its concurrency tester. Experimentally add sqlite3.capi.sqlite3_js_retry_busy(), which runs a callback repeatedly until it _stops_ returning (or throwing) an SQLITE_BUSY error.
FossilOrigin-Name: 45c02ed21635f7ef45214ab5ec6230b6b8bd89e35ef5889db318523e1c679fea
|
2026-03-07 03:32:17 +00:00 |
|
stephan
|
9be458e611
|
Cleanups and docs in the new opfs code structure.
FossilOrigin-Name: 3b470c4c7a1fcc710e6b9eae32134c7f6c3f6008b24c7351257f66f5e8f70311
|
2026-03-07 02:19:23 +00:00 |
|
stephan
|
57d0fe254a
|
For backwards compatibility, ensure that the "opfs" VFS does not specifically require Atomics.waitAsync() (a new requirement of "opfs-wl"), but make use of it if available. Only apply jitter to the concurrency test runs at random intervals.
FossilOrigin-Name: f2175f526c00cfe562e8f332eb197b5ef2c3d6be1fff2aab1566c2c533a293ac
|
2026-03-07 01:01:13 +00:00 |
|
stephan
|
ac63b524ac
|
Add some jitter and duration info to the opfs concurrency tester. Ensure that Atomics.waitAsync() is available for opfs-wl. Further testing has shown that Web opfs-wl is consistently fairer about doling out contested locks but it's dog slow compared to the original VFS under moderate concurrency. In single-connection use they're effectively on par.
FossilOrigin-Name: a9aecc987512d60f2663973f43c769cf086fc14149edfbcb18c0aec9f5aa3dbf
|
2026-03-06 23:50:44 +00:00 |
|