Age | Commit message (Collapse) | Author | Files | Lines |
|
* Create `AbortSignal` without JS
* Don't leak
---------
Co-authored-by: Jarred Sumner <709451+Jarred-Sumner@users.noreply.github.com>
|
|
* fix(runtime): followup for `server.requestIP`
* oops
* yeah
* sure
* Update src/deps/libuwsockets.cpp
* Update Dockerfile
* lol
---------
Co-authored-by: Jarred Sumner <jarred@jarredsumner.com>
Co-authored-by: Jarred Sumner <709451+Jarred-Sumner@users.noreply.github.com>
|
|
(#6165)
* [server] requestIp and AnyRequestContext
Changed Request.uws_request to the new AnyRequestContext. This
allows grabbing the IP from a Request. Unfinished.
* [server] basic `requestIp` implementation
Currently using uws's requestIpAsText, which always returns a ipv6
string. We should return a `SocketAddress` object to the user instead,
which will contain the formatted address string and what type it is.
We'll have to use requestIpAsBinary and parse that ourselves.
* TypeScript docs, use `bun.String`, return `undefined` instead of `null`
if we can't get the ip.
* binary address formatting
* uws getRemoteAddress binding
* remove dead code
* working
* final touches:sparkles:
* I will abide by the results of this poll.
---------
Co-authored-by: Parzival-3141 <29632054+Parzival-3141@users.noreply.github.com>
|
|
Co-authored-by: Jarred Sumner <709451+Jarred-Sumner@users.noreply.github.com>
|
|
chunked encoding, fix case when stream do not return a promise on pull (#6086)
|
|
Co-authored-by: Jarred Sumner <709451+Jarred-Sumner@users.noreply.github.com>
|
|
* Fix HTTP listen behavior being non-compliant with node
* Add error code for address in use
* use SystemError
---------
Co-authored-by: SuperAuguste <19855629+SuperAuguste@users.noreply.github.com>
|
|
* fix buffering clean
* fix resolveMaybeNeedsTrailingSlash and try to fix ci/cd error
* fix resolveMaybeNeedsTrailingSlash and try to fix ci/cd error
* oops
---------
Co-authored-by: Jarred Sumner <jarred@jarredsumner.com>
|
|
Co-authored-by: Jarred Sumner <709451+Jarred-Sumner@users.noreply.github.com>
|
|
* html rewriter response buffering
* pipe the data when marked as used
* fix empty response
* add some fetch tests
* deinit parent stream
* fix decompression
* keep byte_reader alive
* update builds
* remove nonsense
* was not nonsense after all
* protect tmp ret value from GC, fix readable strong ref deinit/init
* fmt
* if we detach the stream we cannot update the fetch stream
* detach checking source
* more tests, progress with javascript and Direct sink
* drop support for pure readable stream for now
* more fixes
---------
Co-authored-by: Jarred Sumner <jarred@jarredsumner.com>
|
|
* check sendfile ctx
* add test
* undo blob check
* undo undo and add assert
|
|
|
|
* just kernel32 things
* more
* Update linux_c.zig
* Update windows_c.zig
* Add workaround
Workaround https://github.com/ziglang/zig/issues/16980
* Rename http.zig to bun_dev_http_server.zig
* Rename usages
* more
* more
* more
* thanks tigerbeetle
* Rename `JSC.Node.Syscall` -> `bun.sys`
* more
* woops
* more!
* hmm
* it says there are only 37 errors, but that's not true
* populate argv
* it says 32 errors!
* 24 errors
* fix regular build
* 12 left!
* Still 12 left!
* more
* 2 errors left...
* 1 more error
* Add link to Tigerbeetle
* Fix the remainign error
* Fix test timeout
* Update syscall.zig
---------
Co-authored-by: Jarred Sumner <709451+Jarred-Sumner@users.noreply.github.com>
|
|
* Automatically hot reload Bun.serve()
* Update doc
* Update example
---------
Co-authored-by: Jarred Sumner <709451+Jarred-Sumner@users.noreply.github.com>
|
|
* Update response.zig
* Comment this out for now
* Support unix domain socket in Bun.serve()
* Add test
* add types
* Update JSFetchHeaders.cpp
* comment this test out
* tls unix web socket serve options
---------
Co-authored-by: Jarred Sumner <709451+Jarred-Sumner@users.noreply.github.com>
Co-authored-by: dave caruso <me@paperdave.net>
|
|
Close: #4298
|
|
* WIP
* backup
* more change
* json related error sovled
* number related issue solved
* revert WriterType changed before
* destroy -> free
* jsonStringify related issues solved
* fix mem.free expected []T or *[_]T, passed [*]const u8
* fix expected []T or *[_]T, passed [*:0]const u8
* fix build script
* fix build script, for real
* replace 0.11.0-dev.4006+bf827d0b5 to 0.12.0-dev.161+6a5463951
* fix build on macOS, COPYFILE.DATA -> COPYFILE_DATA
* fix the last destroy on [*]ptr issue
---------
Co-authored-by: Dylan Conway <35280289+dylan-conway@users.noreply.github.com>
|
|
Co-authored-by: Jarred Sumner <709451+Jarred-Sumner@users.noreply.github.com>
|
|
Co-authored-by: Jarred Sumner <709451+Jarred-Sumner@users.noreply.github.com>
|
|
* Update WebKit
* Don't do async hooks things when async hooks are not enabled
* Smarter scheduling of event loop tasks with the http server
* less exciting approach
* Bump WebKit
* Another approach
* Fix body-stream tests
* Fixes #1886
* Fix UAF in fetch body streaming
* Missing from commit
* Fix leak
* Fix the other leak
* Fix test
* Fix crash
* missing duperef
* Make this code clearer
* Ignore empty chunks
* Fixes #3969
* Delete flaky test
* Update bun-linux-build.yml
* Fix memory issue
* fix result body, and .done status before the last callback, dont touch headers after sent once
* refactor HTTPClientResult
* less flasky corrupted test
* oops
* fix mutex invalid state
* fix onProgressUpdate deinit/unlock
* fix onProgressUpdate deinit/unlock
* oops
* remove verbose
* fix posible null use
* avoid http null
* metadata can still be used onReject after toResponse
* dont leak task.http
* fix flask tests
* less flask close tests
---------
Co-authored-by: Jarred Sumner <709451+Jarred-Sumner@users.noreply.github.com>
Co-authored-by: cirospaciari <ciro.spaciari@gmail.com>
|
|
`URLSearchParams`, `FormData`, `Worker`, `EventTarget` (#4187)
* Add pretty printers for `Headers`, `URLSearchParams`, and `FormData`
* [untested] Add way to code generate getInternalProperties
* bump
* Bump Webkit
* Ref the event loop while loaded
* wip
* checkpoint
* another checkpoint
* The code has been written
* Fixup exports
* Fix all the errors
* Fix bug
* [console.log] Fix bug when printing non-reified types missing values
* Fix loading hash table
* fix plugin
* Fix ref & unref
* auto-unref
* various fixes
* Update bun.zig
* Set toStringTag
* Delete code for macro JSX
* Delete code for `bun dev` HTTP JS
* Move Bun.serve to C++ API
* Delete JSC C API code
* :scissors: :skull: code
* Use JSC C++ for `confirm`, `Crypto`, `prompt`, `alert`
* more dead code
* Update exports.zig
* Use JSC C++ API for FFI
* Remove remaining usages
* Remove remaining usages
* Update ffi.ts
* Update InternalModuleRegistryConstants.h
* draw the rest of the owl
* Update webcore.zig
* bind it
* Fix performance regression in crypto.randomUIUD()
* Update js_parser.zig
---------
Co-authored-by: Jarred Sumner <709451+Jarred-Sumner@users.noreply.github.com>
|
|
* Let the debugger to pause/resume the event loop
* Add initial support for Debug Adapter Protocol
* Add progress
* Update Worker.cpp
* Fix require("console") #3820 (#4073)
* Fix #3820
* Add Module (#4074)
* Set exports to {} in user-constructed CommonJSModuleRecords (#4076)
* feat(bun/test): Implement "toSatisfy" & "toIncludeRepeated" (fwup) (#3651)
* Fix merge issues
* oop
* make codegen
* Fix build issues
---------
Co-authored-by: dave caruso <me@paperdave.net>
* worker tests (#4058)
Co-authored-by: Jarred Sumner <jarred@jarredsumner.com>
* feat(bun:test) add support for test.each() and describe.each() (#4047)
* rename callback to func
* update testscope to handle function arguments
* works
* big cleanup
* works in debug, not release
* fix memory issue & update tests
* catch & str test
* write types for each() & switch tests to ts
* rm & typo
* move some code around & support describe
* review changes
* Fix one of the astro segfaults, also fix bun init version (#4079)
* 4->16
* add assertions
* fix version stuff
* Remove unintentional logs from #4043
* Run prettier
* Update main-worker-file.js
* Update SIMDUTF (#4078)
* Fix constructing buffer from a UTF16 string with the Latin1 encoding. (#4086)
Close: #3914
* Add support for `bun --revision` (#4027)
Co-authored-by: Yash Sharma <yashsharma@Yashs-MacBook-Air.local>
* Updates
* Update __global.zig
* remove non-node node-fallbacks (#4082)
* remove non-node node-fallbacks.
* organize the imports
* Fix test
* Sync bun-polyfills branch (#4081)
* bun-polyfills: initial impl. & baseline refactor
* move @types/ws dep from root to /test/
* bun-types: remove ReadableStream.forEach method
(this does not exist, probably added by mistake)
* bun-polyfills: remove extraneous stream utils
* bun-polyfills: add types syncing file
* bun-polyfills: re-arrange global polyfills
* bun-polyfills: fix FileBlob streams types again
* bun-polyfills: sync all of @types/node
* bun-polyfills: typeguard all current polyfills
* bun-polyfills: fix import paths
* bun-polyfills: switch to wasm impl. of farmhash
* bun-polyfills: support default import of bun obj
* bun-polyfills: transpiler placeholder file
* bun-polyfills: loaderless import.meta polyfill
* bun-polyfills: refactor import.meta polyfill
* bun-polyfills: repl entrypoint & todo list index
* bun-types: Add null to return type of Bun.which
* bun-types: match Bun.sha with Bun.hash.SHA512_256
* bun-polyfills: new "repl" package.json script
* bun-polyfills: full refactor of toplevel hashes
* bun-polyfills: these are fixed
* bun-types: NODE_ENV is optional
* bun-polyfills: fix Bun.env types
* bun-types+polyfills: fix HeapSnapshot.version type
* bun-polyfills: fix some web streams type conflicts
* bun-polyfills: update internal FileBlob.slice
* bun-polyfills: fix subproc stdin conversions
* bun-polyfills: better internal fileblob types
* bun-polyfills: try to sync global performance type
* bun-polyfills: working zig wasm polyfills setup
* bun-polyfills: update scripts
* bun-polyfills: fix wasm file location resolution
* bun-polyfills: goodbye farmhash (replaced by zig)
* bun-polyfills: move all Bun.hash polyfills to zig
* bun-polyfills: reimpl. seeding of seeded hashes
* bun-polyfills: impl. undocumented murmur32v2
* bun-polyfills: switch zighash from jsdoc to .d.ts
* bun-types: partial fix of Hash types
* bun-polyfills: documented Hash.murmur32v2
* bun-polyfills: misc updates
* bun-polyfills: enable sourcemaps
* bun-polyfills: handle empty inputs to hash funcs
* bun-types: narrow down hash func types
* bun-polyfills: remove unnecessary bigint casts
* bun-polyfills: impl. Bun.isMainThread
* bun-polyfills: impl. Bun.sleep and fix sleepSync
* bun-polyfills: impl. indexOfLine
* bun-polyfills: impl. Bun.peek.status
* bun-types: fix hashing test
---------
Co-authored-by: Jarred Sumner <jarred@jarredsumner.com>
* Add remix guide
* Fix title
* add util.formatWithOptions (#4090)
* Add formatWithOptions
* tests and tweaks
* adjust
---------
Co-authored-by: Jarred Sumner <jarred@jarredsumner.com>
* bun test: format description of test.each (#4092)
* bun test: format description
* add tests for tests
* only
---------
Co-authored-by: Jarred Sumner <jarred@jarredsumner.com>
* Fixes #4062 (#4106)
* Fixes #4062
* Update encoding.zig
* Use faster C++ impl
* Update wtf-bindings.cpp
* undo
* Fixup
---------
Co-authored-by: Jarred Sumner <709451+Jarred-Sumner@users.noreply.github.com>
* zig fmt
* Update remix guide
* fs.zig: create temp files with 0o700, not 0o007 (#4107)
* Handle thundering herd of setInterval (#4109)
Co-authored-by: Jarred Sumner <709451+Jarred-Sumner@users.noreply.github.com>
* Fix memory leak in base64url (#4111)
Co-authored-by: Jarred Sumner <709451+Jarred-Sumner@users.noreply.github.com>
* run files without extensions (#4113)
* run script without extension
* process stdio write fix
* don't check for trailing slash, var stream
* More lazily initialize these static strings
* Remove assertion
* wip
* Add --inspect flag
* Deprecate loading `node_modules.bun`
* realpath
* regenerate schema
* More
* more
* Update cli.zig
* Debugger JS loads
* have fun
* Most of the code
* Its starting to work.
* more progress
* win some, lose some
* Update dap.ts
* Async Context Tracking
* untested websocket
* Emit Console messages
* Error handling
* errors
* Make profiling work better
* [clap] CLI arguments with optional values ignore positional params
In `bun --inspect foo.js`, `foo.js` should not be the value of `--inspect`. It is a positional parameter. `--inspect=foo`
* Support multiple simultaneous clients, automatically unpause on disconnect
* regenerate
* Update Makefile
* Update WebKit
* Update cli.zig
* Update InternalModuleRegistry+createInternalModuleById.h
* BaseURL
* Update WebKit
* Add web-inspector-bun
* Update build.ts
* formatting, mostly
* Update debugger.ts
* Update InternalModuleRegistryConstants.h
* wrap
* Update test
* Update test
---------
Co-authored-by: Ashcon Partovi <ashcon@partovi.net>
Co-authored-by: Dylan Conway <dylan.conway567@gmail.com>
Co-authored-by: dave caruso <me@paperdave.net>
Co-authored-by: Tiramify (A.K. Daniel) <94789999+TiranexDev@users.noreply.github.com>
Co-authored-by: Jacques <25390037+jecquas@users.noreply.github.com>
Co-authored-by: Jarred Sumner <709451+Jarred-Sumner@users.noreply.github.com>
Co-authored-by: Ai Hoshino <ambiguous404@gmail.com>
Co-authored-by: Yash Sharma <yashosharma@gmail.com>
Co-authored-by: Yash Sharma <yashsharma@Yashs-MacBook-Air.local>
Co-authored-by: Colin McDonnell <colinmcd94@gmail.com>
Co-authored-by: jhmaster <32803471+jhmaster2000@users.noreply.github.com>
Co-authored-by: Adhityaa Chandrasekar <github@adtac.in>
Co-authored-by: Dylan Conway <35280289+dylan-conway@users.noreply.github.com>
|
|
* Deprecate loading `node_modules.bun`
* realpath
* regenerate schema
* More
* more
* Update cli.zig
---------
Co-authored-by: Jarred Sumner <709451+Jarred-Sumner@users.noreply.github.com>
|
|
|
|
& Response (#3902)
* Atomize respsone.url & response.statusText
* Fix warning
* Atomize Request & Response URLs when possible
* Fix memory leak in response.clone()
bun/bench/snippets on jarred/atomize
❯ mem bun --smol request-response-clone.mjs
cpu: Apple M1 Max
runtime: bun 0.7.2 (arm64-darwin)
benchmark time (avg) (min … max) p75 p99 p995
-------------------------------------------------------- -----------------------------
req.clone().url 77.3 ns/iter (40.35 ns … 222.64 ns) 91.53 ns 128.11 ns 172.78 ns
resp.clone().url 162.43 ns/iter (116 ns … 337.77 ns) 177.4 ns 232.38 ns 262.65 ns
Peak memory usage: 60 MB
bun/bench/snippets on jarred/atomize
❯ mem bun-0.7.1 --smol request-response-clone.mjs
cpu: Apple M1 Max
runtime: bun 0.7.1 (arm64-darwin)
benchmark time (avg) (min … max) p75 p99 p995
-------------------------------------------------------- -----------------------------
req.clone().url 115.85 ns/iter (80.35 ns … 247.39 ns) 128.19 ns 181.93 ns 207.23 ns
resp.clone().url 252.32 ns/iter (202.6 ns … 351.07 ns) 266.56 ns 325.88 ns 334.73 ns
Peak memory usage: 1179 MB
* Update tests
* Update js_ast.zig
* Update test
---------
Co-authored-by: Jarred Sumner <709451+Jarred-Sumner@users.noreply.github.com>
|
|
* Fix bugs with connecting to localhost
* Update uws
* More logs
* Allow not setting a hostname
* Make server.hostname & server.protocol faster
* Fixup
* normalize listening host
* Fix test
---------
Co-authored-by: Jarred Sumner <709451+Jarred-Sumner@users.noreply.github.com>
|
|
* work to get async local storage working.
* a
* a
* everything but queueMicrotask
* sdfghj
* .
* finish
* tests
* test
* ok
* done
* im so stupid
* Upgrade WebKit
* refactor
* refactor
* changes requested
* oops
* cool
* fix runInAsyncScope
|
|
* upgrade
* more fixes
* Bump Zig
---------
Co-authored-by: Jarred Sumner <709451+Jarred-Sumner@users.noreply.github.com>
|
|
(#3257)
|
|
* Allow zero length WebSocket client & server messages
* Add test
* Clean this up a little
* Clean up these tests a little
* Hopefully fix the test failure in release build
* Don't copy into the receive buffer
* Less flaky
---------
Co-authored-by: Jarred Sumner <709451+Jarred-Sumner@users.noreply.github.com>
|
|
* add socket parameter support
* refactor #socket
* add test and more fixs
* some fixes
* bump uws
* handlers fix
* more fixes
* fix node net and node tls tests
* fix duplicate port
* fix deinit on CallbackJobs
* cleanup
* add setImmediate repro
* add test to setImmediate
* this is necessary?
* fix prependOnce on native listener
* try to findout the error on nodemailer CI
* show error message
* Update bun.lockb
* prettier
* Use exact versions of packages
* add alpnProtocol support
* update
* emit error when connect fails on net.Socket
* format
* fix _write and cleanup
* fixup
* fix connect, add alpn test
* fix socket.io
* add socket parameter to TLSSocket
* add TLSSocket socket first parameter
* fixup and _start
* remove flask tests
* fmt
---------
Co-authored-by: Jarred Sumner <709451+Jarred-Sumner@users.noreply.github.com>
|
|
* Use `BunString` in SystemError
* Use Bun::toStringRef when we will de-ref strings
* Move `napi_create_error` to C++ to support `code` being a Symbol potentially
* Update blob.zig
* Make this test less flaky
---------
Co-authored-by: Jarred Sumner <709451+Jarred-Sumner@users.noreply.github.com>
|
|
* Make HTTP Request struct use 8 bytes less memory
* Update server.zig
---------
Co-authored-by: Jarred Sumner <709451+Jarred-Sumner@users.noreply.github.com>
|
|
* fix crash in readablestate
* make node:https request+get actually use https
* use a native readablestream in IncomingMessage
* tweaks
* fix abort crash
* emit close by default
* remove abort. this isnt a real function
* add validate functions, fixup some other requested changes. not done yet
* Update WebCoreJSBuiltins.cpp
* Update JSReadableState.cpp
* Add some missing exports
---------
Co-authored-by: Jarred Sumner <709451+Jarred-Sumner@users.noreply.github.com>
|
|
* Support reading embedded files in compiled executables
* :nail_care:
---------
Co-authored-by: Jarred Sumner <709451+Jarred-Sumner@users.noreply.github.com>
|
|
* progress
* finish `@memset/@memcpy` update
* Update build.zig
* change `@enumToInt` to `@intFromEnum` and friends
* update zig versions
* it was 1
* add link to issue
* add `compileError` reminder
* fix merge
* format
* upgrade to llvm 16
* Revert "upgrade to llvm 16"
This reverts commit cc930ceb1c5b4db9614a7638596948f704544ab8.
---------
Co-authored-by: Jarred Sumner <jarred@jarredsumner.com>
Co-authored-by: Jarred Sumner <709451+Jarred-Sumner@users.noreply.github.com>
|
|
* fixup checks
* throw when tls is not a object also fix socket
* fix error message
* null or undefined on tls option in Bun.serve or sockets should not throw
* add tests
* fix tests and socket validation
* remove unnecessary check
* add listen tests
|
|
* Fix crash in CJS
* Add std.heap.ArenaAllocator
* Use our arena allocator
* Reduce JS parser memory usage and make HMR faster
* Write some comments
* fix test failure & clean up this code
* Update javascript.zig
* make arena usage safer
---------
Co-authored-by: Jarred Sumner <709451+Jarred-Sumner@users.noreply.github.com>
|
|
* Fix `make headers`
* [JS parser] Fix bug with printing non-ascii import paths in ascii mode
* Introduce `bun.String`
* Add test for non-ascii imports & entry points
* Add comment
* Fix build issue
* Support HTTP server
* Make it print the same
---------
Co-authored-by: Jarred Sumner <709451+Jarred-Sumner@users.noreply.github.com>
|
|
|
|
This exposes the WebKit inspector debugger protocol over WebSockets at the endpoint `/bun:inspect` when `Bun.serve()`.
To enable, pass:
```js
Bun.serve({inspector: true, development: true, fetch(req){ /* rest of params *... });
```
Both `development` and `inspector` must be true, as this is very security sensitive to expose publicly.
|
|
cc @cirospaciari, if you use `mimalloc-debug` it catches things like this
|
|
* Fix crash in test.todo
* remove usages of JSC C API in bun:test
* Remove additional JSC-C API usages
* fix `make headers`
* URLSearchParams.length
* FormData length
* URLSearchParams length
* Fix `make headers`
* very fancy length
* Fix bug with exceptions being ignored sometimes
* Add tests for extension toHaveLength
---------
Co-authored-by: Jarred Sumner <709451+Jarred-Sumner@users.noreply.github.com>
|
|
* patch server.fetch
* add tests and fix types
|
|
This changes `publishToSelf` to be `false` by default instead of `true`.
This is a breaking change because it means that `ws.publish("foo", "bar")` will exclude `ws` from the list of websockets to broadcast.
We are making this change because many people asked for this and were confused by the status quo - that `w.publish` publishes to self.
|
|
Previously, this defaulted to "uint8array", so this shouldn't be a breaking change unless you make use of `.slice()` in which case it will now be a reference to the same ArrayBuffer rather than a clone.
The rationale for this change is most usages of Uint8Array on the server need a little more than just the bytes. Many npm packages expect Buffer rather than Uint8Array. Directly returning it for binary websocket messages is faster than creating another one.
|
|
|
|
No performance or memory usage change
Just removing some duplicate code
|
|
|
|
|