Repository navigation
Releases
dev-2026-11a hotfix
Contains dev-2026-11 release + hotfix addressing an enum clash between BNG + LZ4.
dev-2026-11
BREAKING Changes
- [#7724]
core:fmt: Makefmtpad widths with spaces unless the 0 flag is given, keeping to C-styleprintfrules - [#8153]
core:compress: Deflate compression added, rewritten Inflate, core:compress overhaul.
- Speed (Silesia, 5950X / M4 Pro):
- Inflate: 2.1-2.2x the old core inflate; 1.26-1.33x / 1.39-1.46x
vendor:zlib1.3.2. - Deflate L1-9: 1.54-2.43x / 1.18-1.86x zlib 1.3.2, sizes 3% smaller to 1.1% larger than
vendor:zlib. - Level 10 (new
Best): 0.64% smaller than zlib -9 over the corpus, at about 1.4x the time of level 9. Adds 3-byte matches and cost-based block splitting. Image.pixelsandContext.outputare[dynamic]byte(wasbytes.Buffer);Imagesavers take^[dynamic]byte.core:compress/zlibandgzipare@(deprecated)wrappers aroundcore:compress/deflate.- Removed: The old LSB bit-reader helpers in
core:compress,COMPRESS_OUTPUT_ALLOCATE_MIN, and the unused error valuesResize_Failed,Payload_Size_Exceeds_Max_Payload,Unsupported_Compression_LevelandInflate_Error.
- Inflate: 2.1-2.2x the old core inflate; 1.26-1.33x / 1.39-1.46x
New Language Features
- [#7809] General Feature Improvements
@(align=N)for variables: globals, locals,@(static),@(thread_local)- New intrinsics:
return_address,frame_address,stack_pointer,address_of_return_address
- [#7816] Rejects atomics that LLVM cannot lower, and checks atomic memory orderings across the whole program (which many C/C++ compilers do not do).
#+analysis, e.g.#+analysis escape !atomic@(futex=.Wait)/@(futex=.Wake), markingcore:sync's futex procedures, so the analysis checks futex words instead of skipping them.
New Compiler Improvements
Speedups
- [#7823] Debug Info Improvements. Both faster to generate and easier to read in debuggers.
- Each record type is defined once, in one of eight debug types modules generated in parallel; every other module only forward declares it.
- Debug info is finalized per module in parallel, and work for unused debug info is skipped.
- [#8070] Add a fast path to
for inoverstringandstring16 - [#8136] Inline small run-time-length
mem_copyat-o:speed - [#8146] Drop redundant zeroing of literals and declarations
High Impact
- [#8006] Add Android support:
core,NDKandJavabindings, andAPK/AABbundling- Adds Android as a target:
-subtarget:androidfor arm64, arm32, amd64, and i386, at API level 29 or higher.- An app starts at
main, just like on every other platform. odin bundle androidmakes a signed APK or AAB.
- An app starts at
- Android target triples now name a vendor. Without one, LLVM got the ABI wrong (x86 stack alignment, arm64's x18).
- Links use
--no-undefined: a symbol newer than the target API level is a link error. odin bundle androidruns aapt2, adds a prebuiltclasses.dex, aligns, and signs.examples/android: a minimal app, with how tobuild,bundle, andrunit.- A CI job that
checks,builds , andbundles it for all four ABIs. - (more)
- Adds Android as a target:
- [#7878] Update Windows' version of LLVM to 22.1.8
- [#7809] General Feature Improvements
- Matrices now align to the largest power of two dividing their column (row if
#row_major) stride, capped at 16, still with no padding (e.g.matrix[4, 4]f32is 16-byte aligned) - Fixed-capacity
[dynamic; N]Tis now padded to its full size whenTis over-aligned - Naming a
@(deprecated)procedure in a procedure group no longer warns; only a call that selects it does #constparameters can now use their default values
- Matrices now align to the largest power of two dividing their column (row if
- [#7802] Flow-based Escape Analysis of Stack Memory + Show where parapoly procedures and records were instantiated in errors within them
- Tracks where stack memory can end up
- Replaces the bodgy syntactic approach that
check_unsafe_returndid previously - An explicit
rawptr/uintptrconversion ortransmuteopts out - Interprocedural
- The entry point is skipped for executables
- Store warnings suggest clearing the store
- [#7816] Rejects atomics that LLVM cannot lower, and checks atomic memory orderings across the whole program (which many C/C++ compilers do not do).
- Atomic Ordering Analysis. On by default,
-no-atomic-analysisto disable - Warns on
.Releasewrites only ever loaded.Relaxed, and.Acquirereads only ever stored.Relaxed, per global, static, or struct field. -vet-atomic-access(opt-in) errs on plain reads of what is accessed atomically, outside ofcore:synclocks.
- Atomic Ordering Analysis. On by default,
- [#7819] Improved Atomic Analysis: Fences, Weak Compare-Exchange, Futex Wakes, and Volatile Memory Access
- Fences: Only order what they are reached after
(acquire)or before(release), including across calls.
Warnings: - Weak compare-exchange with its second result unused.
- Futex woken before it is written.
- Atomics on a local nothing else can access; suggests
volatile_*. volatile_*on what is accessed atomically.atomic_signal_fencewhereatomic_thread_fencewould pair.
Errors:- Atomics on what may be misaligned, e.g. a
u32converted to^u64.
- Fences: Only order what they are reached after
- [#7823] Debug Info Improvements. Both faster to generate and easier to read in debuggers.
- Procedures and globals are named by package and enclosing procedure (
pkg::proc,pkg::[file.odin]::namefor file-private,proc@linefor literals). - Polymorphic instances share one name; exported and custom link names are kept.
typeidis an enum of the type table, so debuggers show which type it isbit_sets are flag enums on DWARF.- RAD Debugger views for unions,
string16,#soaand other generated types; big-endian types keep their names in CodeView, and the integer ones are byte-swapped by RAD views. - gdb and lldb pretty printers (
base/runtime/odin_debugger.py) formaps,unions, strings, slices, dynamic and#soaarrays, allocators andany. T- They load automatically: embedded in
.debug_gdb_scriptson ELF, copied into the.dSYMon macOS.
- They load automatically: embedded in
- Optimized procedures are marked, and assignment tracking is on unless
-o:none, so more variables keep a location. - DWARF 5 on Darwin and Linux, unique type identifiers, and MD5 file checksums on LLVM 22+.
- Procedures and globals are named by package and enclosing procedure (
- [#7828] Atomic analysis: add `@(synchronizes=...)``, wrapper inference, and more atomic memory model checks
- Fences in callees and callers count; fences in other branch arms don't
- Error on atomic writes to
@(rodata) - Warn on atomics on an unshared
@(thread_local) - Warn on Dekker-style store/load pairs not ordered by
.Seq_Cst - Warn on mixed-size atomics on one location
-vet-atomic-access- Checks plain writes too
@(synchronizes=<Atomic_Memory_Order>)and@(synchronizes_shared=...)replace the lock-name heuristic- Unmarked procedures are inferred from their calls
- Please read the PR's description for how Odin's atomic memory ordering analysis works
- [#7848] Error Reporting/Recovery Improvements for Editors and External tooling (e.g. OLS)
-overlay:<file>: Go-style JSON overlays for unsaved buffersodin check <dir> <dir2>: Check several packages at onceodin check -export-semantics:<json|cbor> -export-semantics-file:<file>- (more)
- [#7871] Record the enum field an implicit selector picks in
-export-semantics - [#7876] General fixes in the odin parser to improve
-export-semantics - [#7877] More Type Inference Improvements
- [#7874] Publish the backend's entity names atomically so parallel modules never read a half-written name
- [#7886] Reduce Type and Polymorphic instantiation Memory Usage in the Checker
- [#7887] Reduce Checker Memory Usage
Scope: 568 -> 104 bytes- AST nodes: 40 bytes smaller
- Fewer constant-value allocations
DeclInfo: 560 -> 312 bytesodin checknow totals around 16.5% less memory on large projects, and is also a smidge faster
- [#7905] Bounds Checking Analysis: Leave out bounds checks that are proven to pass
- Elides bounds checks that are proven will pass by an earlier check, see PR for details
- [#7828] Atomic analysis: add
@(synchronizes=...)``, wrapper inference, and more atomic memory model checks, adds-vet-atomic-access` - [#8068] Only add type info for union type assertions when the assertion can panic
- [#8101] Add
enumsupport to#configand properly check types - [#8162] Fix Windows i386 executables, DLLs and threading. Add to CI to catch 32-bit cleanliness issues.
General fixes and improvements
- [#7795] Fixes #7737 and #7727: Regressions caused by temp copy of underaligned values
- [#7796] Parapoly: Fix regression with proc group containing parapoly enumerated array specialization
- [#7800] Make
@(static)locals visible in the debug info - [#7761]
LLVMBropcode split toLLVMCondBrandLLVMUncondBrin LLVM >= 23 - [#7068] Fixes
(@fast_math)plumbing to LLVM - [#7805] Fix #7798: Sort
link_nameswithProceduresbeforeForeignProcedures - [#7804] Make
vendorcollection optional - [#7807] Upgrade
vendor:zlibto 1.3.2 - [#7808] Fix #7806:
-lto:thin/-lto:thin-filesno longer defaulted to-linker:lldon Windows - [#7814] Update
-vetflag help message - [#7821] Fix #7820: Keep only the origins of what a call's flow loads from memory of an unknown type in the escape analysis
- [#7826] Fix debug info printers
- [#7825] Fix LLVM backend bugs:
- core:simd/x86: replace 35 intrinsics that LLVM has removed (link errors like
undefined reference to llvm.x86.sse41.pmaxsb) with generic ops, as clang's headers do. - core:simd/x86: fix
_mm_cvtsd_ss,_mm_cvtss_sd,_mm_prefetchand__rdtscp("Intrinsic has incorrect argument type"), and AVX wrappers that did not compile. - A multi-value call in a matrix literal hit an assertion. In a #simd literal it put later values in the wrong lanes.
- Slicing a
[dynamic; N]Twith consta...
- core:simd/x86: replace 35 intrinsics that LLVM has removed (link errors like
dev-2026-10
New Language Features
- [#7664]: Allow
bit_setto be backed by arrays of integers
New Compiler Improvements
Speedups
- [#7720]: LLVM Backend Performance Improvements to Reduce Compile Times Part 1
- [#7729]: LLVM Backend Performance Improvements to Reduce Compile Times Part 2
- [#7747]: Compiler Performance Improvements to Reduce Compile Times Part 3
- [#7685]: Complete Parapoly Rewrite + Race condition fixes
- [#7694]: Make global checking codegen order-independent
- [#7686]: Parse integer literals without allocating per digit and reject non digits in their exponent
- [#7739]: Make
radlinkdefault linker on Windows - [#7701]: Multithread The Entire Semantic Checker
- [#7728]:
mapperformance improvements
General fixes and improvements
- [#7654]: Experimental support for cross-linking Windows executables
- [#7788]: Add the new
-windows-sdk-rootto theodin help buildcommand - [#7497]: Fix #7488:
-veterror incore_builtin_soa.odinwhen appending multiple empty structs to#soa[dynamic] - [#7500]: Fix #5851: Detect and error on chained
offset_ofexpressions - [#7505]: Apply mojibake magic to test runner, too
- [#7509]: Fix #7499: Add ctrl-c handler to
run_subprocess - [#7070]: Fix
vendor:ENet.Bufferdata/length ordering issue - [#7524]: Fix alias
nullptrICE - [#7528]:
intrinsics.type_elem_type()added case forType_FixedCapacityDynamicArray - [#7531]: Improve error message for constant indexing
- [#7530]:
simd_abscodegen improvement + fix for unsigned integers - [#7545]: Fix #6917: LLVM IR
va_arginstruction uses wrong stride for C varargs on Windows - [#7548]: Fixes
simd_approx_recipandsimd_approx_recip_sqrt - [#7549]: Fixes
simd_selectandsimd_runtime_swizzleoperand construction - [#7574]: Fix #7573: Check Enumerations iota fits explicit base type
- [#7570]: Fix #7544: Constant
bit_fieldfield access returning zero value - [#7568]: Fix
side_effectsvsvolatiletypo in error message - [#7561]: Fixes #6888: SIMD gather/scatter and masked load/store randomly crashing the compiler
- [#7550]: Fix syntax error type with
-json-errors - [#7577]: Fix
#locationin global variable initializers - [#7582]: Fix #7581: Compiler crash when checking code with swizzled SOA fields
- [#7605]: Fix #7604: Allow building at
/ - [#7607]: Fix #7598: Keep entity iteration valid across a scope map rehash
- [#7623]: Add
-filesuggestion whenmainis duplicated or package names differ - [#7628]: Fix quadratic type checking case: don't cache polymorphic record specializations
- [#7631]: Fix record instantiated poly proc on selector when used as a proc value
- [#7629]: Fix #7596 and #7511
- [#7621]: Fixes the attributes of the context parameter in dummy procedures
- [#7641]: Fix field sorting for
bit_fieldliterals - [#7656]: Fix #7609: Emit union constants for non-constantable unions
- [#7658]: Fixes
#must_tailon WASM targets - [#7652]: Fix #6951 and #5214:
$procedure arguments matching the wrong instantiation - [#7639]: Fix #6302: Broadcasting arrays of matrices and matrices
- [#7661]: Fix #7587: Full length swizzles through a pointer keeping the pointer type
- [#7594]: Fix #7583:
-strict-stylerequires comma before newline after proc parameters which have a default value - [#7588]: Fix #7427 and #7193: Error on OOB tuple element in compound literals
- [#7668]: Fix #5318: Procedures passed to
$parameters share an instantiation's name - [#7666]: Add
ExactValue_Rationalto improve precision on compiler float computation - [#7602]: Adds lifetime markers for named locals
- [#7674]: Fix #5105 and #5569: Deferred call in short circuits causes LLVM codegen failure
- [#7682]: Fixes the
nocaptureattribute on LLVM >= 21 - [#7679]: Fix #7477 and #7506: Casts between nested arrays of the same depth
- [#7697]: Fix compound literals of array backed
bit_fields - [#7692]: Disallows the "odin" calling convention with
#must_tail - [#7706]: Fix #7430: Fixed capacity dynamic array literals lose their length
- [#7709]: Fix #7708: Compare a polymorphic record's constant parameters by value in
subst_unify - [#7711]: Fix #7700: Constant
#soafield access yields the per-element field array; compare and canonicalizetypeidconstants - [#7716]: Fix
intrinsics.procedure_ofwith parapoly specializations - [#7714]: Fixes #3995, #4337, #5950, #7197, and #7640: Handle globals initialized with the address of a non constant literal
- [#7718]: Fix #7490: Preserve zero-initialized arrays in
intrinsics.concatenate - [#7722]: Use a response file for
MSVC-style linkers (link.exeandradlink.exe) when the command line exceeds theCreateProcesslimit of 32KiB - [#7723]: Fix mark a generated polymorphic procedure before publishing it to other threads
- [#7520]: Fix
#soapointer field access throughusing - [#7743]: Allow
usingon#soapointers, and make the fields ausing#soapointer parameter promotes assignable - [#7746]: Fix
-linker:radlinkbeing rejected on Windows - [#7740]: Fix #7689: Incorrect debug source location at the end of if/else chains
New Packages
- [#7523]:
asm: Generalize memory operand syntax to a flat term list
Package Improvements
Speedups
- [#7630]:
core:strconv: Ported Eisel-Lemire's fast float algorithm to Odin - [#7672]:
core:strconv: Second part of the Fast Float port (SWAR based scanning) - [#7725]:
core:strconv: Adds a#configto opt out of fast float (and its table) - [#7726]:
core:hash/xxhash: Add a few more#no_bounds_checktoXXH3 - [#7728]:
base:runtime:mapperformance improvements
General fixes and improvements
- [#7496]:
core:sys/windows: Add bindings for advanced mouse/pointer/pen manipulation - [#7508]:
vendor:windows/wasapi: Fixes - [#7516]:
core:mem: Revert #6476, which misunderstood themem.Scratchallocator's semantics - [#7517]:
core:mem: Implement tests for themem.Scratchallocator - [#7489]:
core:simd/arm: Implement SIMD Arm Neon shift left - [#6787]:
core:fmt: Improvedfmt_osdocumentation - [#7527]:
core:simd/x86: Fixed typo inavx.odin - [#7539]:
vendor:curl(Linux): Remove import declarations for crypto libraries that differ by distro - [#7532]:
core:sys/windows: Add line-drawing functions ingdi32.odin - [#7563]:
core:simd/arm: Implement SIMD Arm Neon shift right - [#7576]:
vendor:sdl3/ttf: Use[^]byterather thancstringand addExpandStringwrapper - [#7572]:
vendor:libc-shim: Replacemem_zerowith null termination instrncpy - [#7565]:
core:sys/windows: Add performance counterPSAPIbindings - [#7553]:
vendor:box2d: Add missingGetVersionbinding - [#7546]:
vendor/box3d: SupportBOX3D_SHAREDon Windows - [#7543]:
vendor:egl: Bind more procedures and constants - [#7599]:
core:os: Fix the 0th env var being duplicated on Linux - [#7600]:
vendor:OpenEXRCoreFixes - [#7536]:
core:odin: Fix #7526: Handle triple quote/tick string - [#7603]:
core:simd/arm: Implement SIMD Arm Neon zip unzip extract - [#7615]:
base:runtime: Fix #7562: Windows AMD64 no-CRT stack probe support - [#7620]:
core:dynlib: Improvements and explanations in example - [#7622]:
core:strconv: Fixes three bugs in hex float parsing - [#7624]:
core:math/big: Fix squaring path writing the denominator into the numerator - [#7633]:
core:encoding/json: Size the unquote buffer for replaced invalidUTF-8 - [#6722]:
core:sys/linux: FixTime_Valprecision comment - [#7634]:
core:crypto/_fiat: Fix P-256fe_sqrtaddition chain, unblocking compressed SEC1 point decoding - [#7627]:
core:rexcode: Add 512bitvmovntdq(a) - [#7659]:
core:encoding/base64: Add strict decoding options - [#7663]:
core:simd/x86: Implement simd x86 avx2 avx512 shift logical - [#7632]:
core:simd/arm: Implement simd arm neon transpose reverse - [#7579]:
core:c: Fixes #7540: fastX type widths - [#7584]:
core:sys/windows: Extend Win32 common dialog bindings - [#7610]:
core:sys/windows: Fix #7547: Double-null-terminate path inSHFileOperationW - [#7665]:
core:container/bit_array: Addcreate_from_enumhelper. Improve documentation, fix examples. - [#7671]:
core:container/xar: Fix chunk slot count cappingArraycapacity - [#7484]:
core:sort: Document all public procedures and Interface types - [#7676]:
core:crypto/aegis: Minor documentation cleanups - [#7673]:
core:encoding/json: Removes@(private)fromunmarshal_valuefor use in user unmarshalers - [#7590]:
core:os: Allocate the long-path buffer in the caller's allocator - [#7320]:
vendor:sdl3: WASM foreign import library configurable - [#7684]:
core:rexcode: Changedappendtoappend_elemfor a sizable improvement - [#7683]:
core:simd/x86: Implement simd x86 avx512 evex512 - [#7690]:
core:encoding/base64: Add unpadded base64 encoding - [#7695]:
core:simd/x86: Implement simd x86 avx512 typo fix - [#7703]:
core:encoding/csv: Preserve final CSV records in multiline mode - [#7704]:
core:encoding/cbor: Fixbit_setbyte order - [#7643]:
vendor:sdl2: AddRenderSetVSync - [#7721]:
core:sync/chan: Accept.Recvchannels inchan.recvandchan.try_recv - [#7733]:
core:odin: Fix parser assert with invalid for-in stmt - [#7731]:
core:time/timezone: Preserve fractional timestamps in timezone conversions - [#7738]:
core:math/linalg/glsl: Fixmix_vec4andmix_dvec4implementations - [#7753]:
core:fmt: Fix incorrect number of bytes written returned when formatting floats withh/H - [#7751]:
core:simd/x86: Implement simd x86 sse2_mm_insert_epi16fix - [#7755]:
core:sys/windows: AddGetCursorandGetDpiForSystem - [#7756]:
vendor:box3d: Fixing typo fromuserMaterialdtouserMaterialId
Miscellaneous
- [#7495]: macOS Intel removed from CI tests and releases, due to general flakiness ...
dev-2026-09
New Language Features
- Experimental: Inline
asmtemplates for x86/x64, ARM and RISC-V in #7271, #7406 and #7465 - Documentation: https://odin-lang.org/docs/inline-asm/
- ``` and """ strings to declare multi-line string literals that normalize indentation up to the closing quotation marks.
New Compiler Improvements
- Use
posix_spawnpinstead ofsystem()when invoking the linker in #7485 - Fix default linker timing label on non-Windows platforms in #7240
- Fix
odin doc -doc-formataborting on any package that imports another package in #7242 - Fix bigint negative printing in #7243
- Fix constant
transmutefrom unsigned to signed wrapping one value too many in #7244 - Handle
#soafixed compound literals for vars in #7239 - Validate floating-point array counts as integers in #7251
- Fix typo in enumerated array in #7252
- Make redeclaration diagnostics stable under threading in #7253
- Improve procedure group overload scoring in #7227
- Fix #7012 - a type isn't its own specialization in #7257
- Fixes swizzles on array elements in soa containers in #7259
- Fix Name Canonicalization Rotate Left typo in #7267
- Fix constant
string16carrying UTF-8, and inverted slice indices reachingsubstringin #7268 - Fixes chain indexing for array elements in
#soacontainers in #7270 - Fixes range loop over array elements in
#soacontainers in #7277 - Fix riscv64 ABI classification for padded, over-aligned, and pointer-bearing structs in #7282
- i386: Pass aggregate arguments by value on the stack in #7283
- i386: Return aggregates through the hidden pointer on System V targets in #7284
- Use enum value init expression instead of numeric value in doc format in #7289
- Remove dead
simd_remin #7269 - Fix duplicated name for temporary
.llfiles in #7291 - Don't clear the operand value on the simd success path in #7315
- Fix #7281, apply
#max_field_alignin #7314 - Add diagnostic for missing field name to
type_field_bit_offset+type_field_bit_sizein #7293 - Fix byte swap on untyped int-like untyped types in #7294
- Bound constant arguments before range-checking them in #7295
type_is_comparable: false for an unspecialized polymorphic record in #7296- Fix guards combining two required conditions with
&&in #7297 - Correct the endianness guard messages in #7301
- Make the three endianness predicates partition in #7307
simd_extract_lsbs/simd_extract_msbsReject a lane count that exceeds thebit_setlimit in #7308- Clarify
type_has_shared_fieldsbehavior in code doc in #7309 - Matrix: enforce the element limit on multiplication and on overflowing dimensions in #7310
- Hoist
ifstmt in enumerated_array in #7311 - Check enumerated array length in #7312
check_matrix_type_hint: Require the element type and layout to match in #7313- Fixes
f32to 64-bitint/uintconversion in #7319 - Fixes
floatto 128-bitint/uintconversion in #7321 - Classify amd64/arm64 aggregates from the source type in #7292
- Fix
long->int64_tfor win64 in #7324 - Fix
bit_fieldintrinsic by @kalsprite in #7325 - Fix out of bounds for zero arg
#load_directoryin #7329 - Fix compilation error when using the fixed-point write procedure and
-vet-unusedin #7330 - Fix backend support for
#unrollon[dynamic]Tin #7337 - Fix
bit_setupper endpoint representability check in #7340 - fix
bit_setsubset/superset constant folding in #7346 - Fix three faults in
&~constant folding in #7345 - Enforce the
#simdconstruction rules and stop two crashes in #7344 intrinsics.type_integer_to_unsigned(rune)returnsf16in #7343quaternionandcomplexnan cmp & accessors in #7342- Improved stable sort, from
insertion_sorttomerge_rotatein #7349 - Set console codepage to utf-8 (on Windows), and restore old one on exit in #7351
- Reenables
LLVMSetUnnamedAddressfor global constants merging in #7333 - Range-check a constant passed to an
#any_intparameter in #7360 - Add the missing
breakinadd_min_dep_type_infoin #7359 - riscv64: Read a flattened aggregate's members from their real offsets in #7334
- Fix eol diagnostic in #7358
- Fixes the
%%operator in #7353 - Fix SIGBUS/segfault when compiling recursive
#soaslice/dynamic array types in #7361 - ABI Conformance harness + fixes in #7327
- Fix
c_va_startguard andc_va_argformat mismatch in #7367 - riscv64: use a slice for the coerce offsets + missed arm in riscv ABI in #7368
- Reject a built-in procedure assigned to in #7369
- Do not register map runtime dependencies under bedrock in #7371
- Range-check the constant
absfolds in #7372 - Fix #7377 in #7378
- Let FreeBSD use default
gb_copy_filepath in #7381 - Add AVX-512 VNNI VPDPWSSD support to inline assembly in #7380
- A base prefix with only digit separators has no digits in #7370
- A
bit_fìeldconstant is range-checked only while it is untyped in #7385 - A self-referential
bit_fieldhangs the compiler in #7390 - Codegen worker must not exit process with running siblings in #7391
- Multi-instruction asm template computed with one operand twice in #7393
extendfamily emits mnemonic that doesn't exist in #7392- Let compiler check error() fmt strings + fixes in #7395
- Peel union constant to its variant in #7396
complexandquaterniontake their type from their arguments in #7399- A union constant in a ternary or builds an ill-typed phi #7397
- Implement fixes for constant unions in #7087
- Only reuse a compound literal's storage for a variable of its type in #7400
- Fix
or_elsediverging with single return value in #7394 - Remove the
os.Error == 0compatibility hack in #7403 - Restore the
Did you want 'nil'?diagnostic everywhere in #7405 - Address
printfformat warning in #7404 - Better hints when assigning or comparing to
0when it is not allowed in #7407 - Fix reordered compound literal comparison in #7409
- Fix compiler panic on a selector into
#branch_locationin #7415 - Fix compiler segfault on
distinctof a type alias that cycles back in #7416 - Fix segfault comparing two foreign declarations of the same symbol in #7417
- Fix compiler segfault on a bare
..in a type position call in #7418 - Asm cast guard in #7437
- Reject asm operands in slots only a named hardware register can fill in #7431
- Diagnose in an asm template instead of aborting with SIGILL in #7434
- Fix
-no-crton Windows in #7426 - Fixes
-o:sizein #7440 string_split_iteratorfix in #7445- Big endian bool cmp in #7448
- Fix
min(T) % -1kills the process with SIGFPE, at every hardware width in #7450 - Fix canonical names for types in procedure literals in #7439
- Inline
asmtemplates - CFG based analysis in #7444 - Normalize bool cmp in #7449
- Reject a non-constant argument to an asm template's immediate in #7432
- Add missing
i128le,i128beto endian switch in #7459 - Convert constants before narrowing to
i64in the remaining checker sites in #7460 - Delete dead code that crashed the checker on an
asmtemplate group with no parameters in #7435 - Fix a swizzle target in a multi-assignment reading an already-written source in #7458
- Broadcast a constant into every array level in #7457
- Give asm templates a documentation category instead of a blank header and a null
%sin #7438 - Do not deduplicate cases in tagless switches in #7468
New Packages
Package Improvements
- [core:unicode/utf8] Fix truncated text for grapheme clusters in #7003
- [core:thread] Clarify completed task handling in thread pools in #7247
- [core:simd] Fix grammar in #7203
- [core:bytes] Optimize
index_anywith a byte bitset in #7255 - [core:simd/arm] Implement simd arm neon table lookup in #7256
- [core:odin/tokenizer] Fix broken double-checked lock in
initin #7258 - [core:sys/windows] Add
UnregisterHotKeyin #7264 - [core:odin/parser] Fix unparenthesized attributes missing end position in its AST node in #7265
- [vendor:box3d] Add Box3D support for WebAssembly in #7266
- [core:simd/arm] Fix simd arm neon table lookup on big endian in #7273
- [core:unicode/utf8] + [core:text/edit] Add
textfield toutf8.Grapheme, fix caret translation by grapheme in #7274 - [base:runtime] Fixes
non_zero_append_elem_fixed_capacity_stringin #7317 - [base:runtime] Adds the
popfamily of procedures for#soaarrays in #7290 - [core:simd/arm] Implement simd arm neon extended table lookup in #7275
- [core:simd/arm] Implement simd arm neon logical in #7332
- [core:encoding/hex] Add non-allocating
decode_into_bufferin #7347 - [base:runtime] Fix #7262. Do not free the original block on a failed heap resize in #7328
- [core:os] Fix #7350 in #7352
- [core:mem] Keep
Dynamic_Arenaallocations aligned after reset in #7355 - [core:slice] Fix
slice.getdoing unnecessary bounds check in #7363 - [core:slice] Fix
slice.simple_equalmisbehaviour on empty slice with non-nil data in #7364 - [core:image/tga] Fix
.alpha_add_if_missingfor TGA in #7366 - [core:sys/windows] Fix
FILE_NOTIFY_INFORMATIONstruct member casing in #7374 - [core:odin] Remove
inlineandno_inlinekeywords in #7375 - [core:sys/wasm/js] Support shared WebAssembly memory in browser APIs in #7272
- [core:container/handle_map] Nicer organization, fix broken test, add test in #7383
- [core:rexcode/x86] Merge tables in #7384
- [core:thread] An exiting worker must not leave published in #7388
- [core:rexcode] Improvements in #7362
- [core:image/jpeg] Fixed negative read bug in #7410
- [core:hash] Add murmur3 implementation in #7419
- [core:nbio] Fix use-after-free, spurious poll readiness, and double completion on Windows in #7401
- [vendor:raylib] Fixed typo in
raylib.odinin #7451 - [core:os] Fix simple typo in
process.odinin #7454 - [core:fmt] Fill left-justified field with spaces ...
dev-2026-08
New Language Features
- Allow custom
-collections to be used forforeign imports in #6999
New Compiler Improvements
- Disallow proc group assignment to blank identifier in #6996
- Increase stack size for threads in #6997
- Reject chained deferred procedures in #7013
- Fix arm64 ABI: '#simd' vectors and vector aggregates not passed in registers (#7010) by in #7015
- Fix darwin arm64 argument passing abi in #7030
- Fix playdate
-subtargetissue where floats were incorrectly compiled in #7026 - Make object and assembly file extensions user-changeable again in #7041
- Fix subtype comparisons (#6151, #3689, #7020 and #4734) in #7039
- Fix constant aggregate rebuild in #7046
- Keep polymorphic probes read-only in #7052
- Fix
switchfallthrough inlb_emit_struct_ev(#7050) in #7051 - Fix arm64 argument alignment abi for composite type of size <= 16 bytes in #7045
- Preserve recursive struct aliases in #7042
- Compare
mmapresult toMAP_FAILEDrather thannullptr(#7053) in #7065 - Stricter type coercions in
convert_to_typedin #7056 - Fix backend deadlock where
func_raw_types_mutexandtypes_mutexwere waiting on each other by in #7060 - Add
typeidconstant case tocheck_comparison(#7027) in #7064 - Fixes compile time equality comparison for
string16in #7071 - Allow fixed-capacity arrays in constant unions (#7089) in #7090
- Fixes spurious
#no_aliascompile errors during polymorphic overload resolution in #7116 - Reject the use of labels as expressions (#5830 and #7108) in #7130
- Fix hang for poly procs with default proc literal params in #7146
- Fix
expand_valuespanic on type expressions in #7157 - Handle spaces in args for 'odin run . -- ' on Windows in #7162
- Don't assert on
-define:values with an exponent in #7164 - Fix error hint for vet tag in #7132
- RISC-V fixes/improvements in #7144
- Fix LLVM codegen for matching on
nilforunion{}in #7139 - Allow
#caller_expressionon constant parameters which are procedures in #7119 - Fixes compile error due to
#no_aliasshifting position in procs with indirect return in #7214 - Fixes a compiler segfault with zero sized parameters on targets that ignore these in the LLVM signature in #7215
- Fix every named argument in "Given argument types" being labelled with the first argument's name in #7217
- Fix
levenstein_distancein #7218 - Fix signed Int Constants being permitted to exceed size in #7220
- Fix
string16oob in #7221 - Fix new line suppression by @kalsprite in #7222
- Improve grammar on import name error in #7223
- Fix matrix col error message that said rows instead in #7224
- Minor typos in #7225
- Fix crash on poly in #7226
- Fix
-vet-packagesnot working in certain cases in #7210 - Pack tuple results into variadic arguments (#7167) in #7185
- Polymorphic procedures resolving to
nil(#6753 and #) in #7208
New Packages
[core:crypto/rsa]Add RSA support in #6857[core:encoding/asn1]Add strict DER +[core:crypto/x509](X.509 v3) in #6838[core:crypto/turboshake]Add TurboSHAKE in #7043[core:rexcode]Add rexcode, a high-performance multi-architecture instruction encoder/decoder/printer package in #6839[core:debug/trace]Replaced with a much improved implementation in #7141
Package Improvements
-
[core:container/rbtree]Matchfind_or_insertsemantics with AVL in #6988 -
[core:container/rbtree]Addupsertin #7002 -
[vendor:box3d]Fix missing return value of*_GetUserDatafunctions in #7006 -
[core:sys/windows]Add additional bindings in #7022 -
[vendor:raylib]Updaterlglbindings to raylib 6 in #7023 and #7028 -
[core:strings]Add@(require_results)where useful in #7038 -
[core:log]Make multi logger use all loggers instead of breaking on first non match in #7057 -
[core:encoding/csv]Update example to use the newcore:osin #7076 -
[base:runtime]Fix markdown table in docs in #7077 -
[base:runtime]Fix map reporting incorrect capacity after shrinking in #7097 -
[core:os]Fixes file open emulation of posix open behavior on mode/perm mismatch on Windows in #7098 -
[base:runtime]Fixes SoA dynamic arrays not zeroing all elements after shrink + grow in #7103 -
[core:odin/tokenizer]Set token kind to float for hex floats in #7117 -
[vendor:sdl3]Fix(S/G)etDefaultTextureScaleModein #7133 -
[core:unicode]Fix duplicated entries withinunicode_spacesandspace_rangesin #7142 -
[base:runtime]Fixes growth check for SoA dynamic arrays in #7145 -
[core:debug/trace]Sync tracking allocator withcore:memin #7147 -
[core:debug/trace]Sync allocator withcore:memin #7200 -
[core:sys/windows]FixLoadLibraryEx_Flagsoff-by-one in #7151 -
[core:mem/virtual]Close dangling handle inmapimplementation on Windows in #7159 -
[core:testing]RespectTERM=dumbin test runner to prevent ANSI sequences, color in #7160 -
[core:thread]Addget_nameandset_namein #5464 -
[base:runtime]Shrinkmapto minimum capacity withshrinkin #7152 -
[core:rexcode/x86]Label addressing for RIP-relative disp and movabs imm in #7168 -
[core:simd/arm]Implement simd arm crc32 in #7174 -
[core:compress/gzip]Fix example code in #7178 -
[core:simd/arm]Implement simd arm pmull in #7184 -
[base:runtime]Follow up fix ofshrink_map#7182 -
[core:odin/parser]Correct end position for fields with a specialization in #7180 -
[core:text/scanner]Fix octal escape parsing in #7176 -
[core:os]Update staleos.flagsfile doc in #7187 -
[core:mem]Zerotracking_allocator.allocation_mapandtracking_allocator.bad_free_arrayindestroyin #7195 -
[base:runtime]Make soa builtins go vroom in #7196 -
[base:runtime]Cleans up obsolete comments in_append_soa_elemsin #7204 -
[base:runtime]Annotates the array parameter of dynamic array builtins with#no_aliasin #7149 -
[core:simd/arm]Impl fix simd arm SHA in #7205 -
[core:os]Fixglobnot cloning pattern when no meta characters present in #7206 -
[core:crypto/_fiat]Fix conditional inensurein #7211 -
[vendor:box3d]Update box3d (#7177) to erincatto/box3d@3fc20f5 in #7212 + #7231 -
[core:crypto]Minor fixes in #7228 -
[core:os]Resolve relative path beforeSHFileOperationWin_remove_allin #7230 -
[core:net]Fix percent encoding for bytes < 0x10 (#7109) in #7234 -
[core:simd/arm]Implement simd arm neon bit manipulation in #7216 -
Various typo fixes across the packages
Miscellaneous
[vendor:*]Replace makefiles in with shell scripts in #7034- CI: Update NetBSD to
10_2026Q1in #7085 - Improve build scripts portabability and make them
pwdindependent in #7084 - CI: Make NetBSD CI less prone to breaking when scheduled on old CPUs
What's Changed
Full Changelog: dev-2026-07a...dev-2026-08
dev-2026-07a
What's Changed
- Add pull request policy to
vendor/README.md - Fix
vendor:raylib6.0 bindings - Update
vendor:box3dto latest commit in #6986 - Fix for
@(objc_implement)being broken indev-2026-07release in #6990 - Ensure identical types have same hash by in #6974
- Prevent git output with signature information in #6973
- Update wgpu.js in #6975
- Fix parser failing on comments inside ternary expressions in #6968
- fix
assign_at_elems_fixed_capacity_dynamic_arrayin #6993 - Fix #6979 in #6987
- Remove
defaulthelp option for stack protector in #6984 - Update
kb_text_shapeto v2.21 in #6944
Full Changelog: dev-2026-07...dev-2026-07a
dev-2026-07
New Language Features
- Add restricted Odin subset mode
-bedrockin #6654 - Add Playdate support as a
-subtargetof-target:freestanding_arm32in #6900 - Support
q.xyzon quaternion*
New Compiler Improvements
- New Feature: Stack canaries #6830
- Change default settings to enable pie and full relro on Linux in #6790
- Make the type hash 63-bit to not crash OMG JIT in WebKit under WASM in #6829
- Add suggestion to use 'typeid_of(type)' when trying to do
typeid(type)in #6846 @(test)procs in#+privatefiles are now processed in #6844- Reject transitive
#subtype-onlyimplicit conversion in checker in #6845 - Fix initialization of global
anys #6849 - Fix: tuples are not handled properly in struct literals in #6859
- Move away from
systemforodin runin #6848 - Raise child process error signal if
odin runchild gets terminated #6870 - Flush accumulated errors before printing assert in #5508
- Fix #6874 in #6880
- Fix #6885 in #6886
- Resolve panic on 64-bit constant overflow in #6904
- Fix constant union stuff against tests in #6905
- Report range type correctly on out-of-bounds in #6906
- Add more constant type cases in #6913
- Use
lb_add_values_to_arrayforcvarargsin #6908 - Only allow direct
CVarArgusage inc_va_startin #6916 - Add ways to diff anon procs with same signature in #6920
- Fix compiler segmentation fault when field names start with '.' in #6928
- Fix UTF-8 decode error check in #6930
- Improve custom output location (
-out) support for LLVM-IR, OBJ + ASM in #6932 - Fix typo in union error in #6929
- Make checking of untyped to union conversions more strict in #6925
- Allow casts to change mode for array programming in #6921
- Avoid code regeneration for fns in unroll in #6939
- Make diverging
or_elseinherit#loadlhs type in #6934 - Add diverging (
!),#optional_okto proc canon name in #6935 - Fix misuse of
lb_emit_(min|max)insimd_(min|max)in #6924 - Remove support for LLVM 14, suggest apt.llvm.org if
llvm-config-14can't be found in #6918 - Use specific format to make
f64strings canonical in #6949 - Disallow types in
#any_intparams in #6950 - Convert poly const int to
f64when needed in #6948
New Packages
Package Improvements
- [core:terminal/ansi] Add Alternate Screen Buffer in #6813
- [core:nbio] Fix UAF op
op.lin cross-thread exec in #6819 - [core:slice] Fix wrong variable name in
slice.rotatein * fixed wrong variable name in #6828 - [vendor:vulkan] Update Vulkan in #6824
- [core:container/pool] Fix ABA race in the free list in #6822
- [core:bufio] Improve
reader_read_slice's documentation using the power of punctuation in #6463 - [base:runtime] Added optional allocator error to
map_insertin #6836 - [core:odin] Correct divergence from C++ parser with expr levels, causing failures in #6861
- [vendor:stb/truetype] Amend
PackFontRangestypes in #5662 - [core:mem] Rename
DEFAULT_PAGE_SIZEtoPAGE_SIZEin #6471 - [core:os] Improve consistency in
os.read_entire_file_from_filein #6671 - [core:flags] Improve
bit_setsupport in #6435 - [vendor:x11/xlib] Update
Xlibdocumentation link to current PDF in #6875 - [core:text/regex] Improve regex compiler performance in #6873
- [core:mem] Fix
mem.Dynamic_Arenaalignment in #6878 - [core:crypto/wycheproof] Update the tests to match recent changes in #6887
- [vendor:sdl3] Remove non-exported
joystick_locksymbol in #6893 - [vendor:windows/GameInput] Remove
GameInputprefix from interface type names in #6903 - [core:encoding/entity] Fix whitespace normalization in #6914
- [core:container/handle_map] Remove unused
base:builtinimport in #6931 - [vendor:wgpu] Update to v29.0.1.1 in #6947
- [core:encoding/cbor] Fix segfault decoding map/big.int nested in slice elements in #6945
- [core:encoding/cbor] Fix data corruption in #6954
- [vendor:wgpu] Update to v29.0.1.1
- [core:container/xar] Forward caller location to make in
array_push_back_*in #6958
Miscellaneous
- Fix nightly static Linux build in #6808
- Added missing
LDFLAGSforOpenBSDin #6728 - Deduplicate macOS and Linux nightly builds in #6812
- Remove dependency on
mbedx509,mbedcryptofor Darwin in #6842
What's Changed
Full Changelog: dev-2026-06...dev-2027-07
dev-2026-06
New Language Features
- Add
@(fast_math)attribute (PR #6676) - Support
**asexpand_valuesoperator:**x==expand_values(x)(PR #6731)
New Compiler Improvements
- Better error reporting for unary
++/--(PR #6649) - Removed Haiku from supported targets (PR #6631)
- Better SIMD generation for square matrix multiplication and
intrinsics.transpose(PR #6676) - Fix
deferred_returnand segfault in docs writer (PR #6668) - Fix issue #6429,
odin dochandling of Fixed Capacity Dynamic Arrays inwrite_expr_to_string(PR #6670) - Add duplicates checking in type switches by @Creativty (PR #6702)
- Fix
niltyping for procedure call arguments (PR #6701) - Fix issue #6096, honor
-no-thread-localfor variables declared at procedure scope (PR #6696) - Improve
simd.runtime_swizzleand its documentation (PR #6712) - Improve thoroughness of generic procedure type detection (PR #6711)
- Fix race condition relating to type specialization (PR #6716)
- Fix issue 6692,
usingon bit fields (PR #6742) - Fix error pos style environment variable overwriting commandline argument (PR #6741)
- Properly check
#any_intparam operands (PR #6740) - Fix assertion failure when comparing array-backed
bit_fields (PR #6773) - Disallow
*and/forbit_sets (PR #6777) AArch64assembly support forforeign import(PR #6784)- Fix error message "
donot on the same line asfor" (PR #6794)
New Packages
- Add ML-KEM (FIPS 203) post-quantum key encapsulation support (PR #6635)
- Add ML-DSA (FIPS 204) post-quantum digital signature support (PR #6695)
- Add intial PEM support (PR #6755)
Package Improvements
- Improve consistency between platforms for
os.copy_directory_all(PR #6651) - Use
context.assertion_failure_procin tracking allocator bad free callback (PR #6659) - Add
DefSubclassProcbinding (PR #6664) - Make
make_alignedandnew_alignedbuiltins (PR #6677) - Handle invalid utf8 when parsing json (PR #6669)
- Fix xlib binding signatures (PR #6680)
- Add unicode mapping between 'ẞ' <-> 'ß' (PR #6672)
- Fix missing
only_if_existsparameter inx11.XInternAtoms(PR #6684) - Convert
(f)statcalls tostatxin Linuxcore:osimplementation (PR #6683) - Additional xlib binding changes (PR #6687)
- Fix
read_slicetypo incore:io/utils(PR #6697) - Document
strconv.parse_intandstrconv.parse_uint(PR #6706) core:math/linalg: Add Gram-Schmidt orthonormalize (PR #6707)vendor:wgpu: Fix incorrect proc signatures (PR #6661, PR #6724)- Add support to unmarshal the new fixed capacity dynamic arrays (PR #6713)
core:encoding/base64: Misc fixes and improvements (PR #6733)core:strconv: Fix upper bound of Decimal inround_shortest(PR #6734)vendor:directx/d3d12: AddIDevice10interface (PR #6602)vendor/README.mdfix inverted SDL2 and SDL3 (PR #6749)- Swap
panic_contextlessandunimplemented_contexlessdocumentation (PR #6758) core:sys/linux: Remove parapoly fromrt_sigaction(PR #6756)core:slice: Fixrotate_leftcrash on empty array (PR #6771)core:sys.info: Prefer mem info from/proc/meminfooversysinfo()(PR #6765)core:mem/core:testing: Fix buffer overflow when running a test which shrinks a dynamic array (PR #6791)
Miscellaneous
- Add
/srcas safe directory for nightly linux arm workflow (PR #6658) - Link against
libstdc++expfor backtrace support (PR #6732)
Full Change Log
Full Changelog: dev-2026-05...dev-2026-06
dev-2026-05
New Language Features
- Native Array Casting Semantics, e.g.
cast([4]f32)4x_u32(PR #6605) - Increase
matrixlimit from 16 to 64 elements, allowing for up tomatrix[8, 8]T
New Compiler Improvements
- Add -debug default behavior usage docs (PR #6511)
- Fix output of filename when outputting docs with the
-in-source-orderoption (PR #6536) - Fix
#no_bounds_checkregression (PR #6541) - Fix target feature lookup, canonicalize
target_features_set(PR #6544) - Fix (Issue #6344) - Field-first index writes on
#soa[dynamic]Tand#soa[]T(PR #6539) - Fix
aligned_allocnot defined on Android/Bionic (PR #6519) - Fix
break/continuebeing allowed in a nested unrolled range loop (PR #6521) SROAand simplification passes on-o:minimal(PR #6553)- Fix array counts for floats that are exactly representable as integers e.g.
[1.1e4]int(PR #6583) - Prevent coredump when child process gets killed by signal (PR #6586)
- Support multiple return values within compound literal elements (PR #6596)
- Fix (Issue #5422) - Make
@(rodata)on constants yield "not supported" message (PR #6601) - Fix (Issue #6594) - Report cycle when assigning variable to itself (PR #6599)
- Fix (Issue #6621) - Report cycle for
type_of(PR #6624) - Removed moribund
EssenceOS from supported targets (PR #6623) - Add
intrinsics.c_va_*(PR #6629) - Fix type info not being generated for types only used in
[]typeidliterals (PR #6630) - Fix compiler warning on Linux (PR #6638)
New Packages
- [core:crypto/noise] Add Noise implementation (PR #6573)
Package Improvements
- [core:os] Add
dirprocedure (PR #6517, PR #6528) - [base:runtime] Native simd width (PR #6545)
- [core:os] Fix
_volume_name_lento handle paths of purely path separators (PR #6547) - [core:math/linalg] Faster
dot,cross3,floor,ceil, and addtrunc(PR #6549) - [core:math/linalg] More matrix related calls:
pseudo_inverse,inverse_lu_decomposition,matrix_inverse_gauss_jordan - [core:mem/virtual] Fix arena overcommit bug (PR #6552)
- [core:sys/linux] Fix missing argument in
adjtimexsyscall (PR #6502) - [vendor:wgpu] Update to v29.0.0.0 (PR #6551)
- [core:sys/freebsd] Add
ioctlandstdioFILENOconstants (PR #6537) - [core:nbio] Handle
EOFin poll on Posix (PR #6556) - [core:container/xar] Fix a typo in example (PR #6559)
- [core:testing] Ensure
make_*style bounds traps are handled in test runnerexpect_signalon Windows (PR #6563) - [core:sys/windows] Add I/O Ring API bindings (PR #6562)
- [vendor:directx/d3d12] Fix
D3D12_BARRIER_SUBRESOURCE_RANGEstruct (PR #6557) - [core:testing] Fix stop reason being wrong for OOB on Windows (PR #6570)
- [core:fmt] Remove redundant 'defer bufio.writer_flush' from
fprintprocs (PR #6567) - [vendor:zlib] Fix signature of
inflateInit_(PR #6566) - [vendor:sdl3] Mixer: Removed
#by_ptrfrom args that are valid asnil(PR #6576) - [core:mem/virtual] Fix
map_fileusage example (PR #6577) - [core:mem/virtual] Add
address_hinttoreserve(PR #6558) - [core:os] Fix
stemon empty path (PR #6579) - [core:path/filepath] Create wrappers for allocating procedures (PR #6575)
- [core:fmt] Omit value when formatting
map[T]struct{}sets (PR #6581) - [vendor:vulkan] Add version macros (PR #6565)
- [core:sys/windows] Add Registry Association API (PR #6525)
- [core:sys/windows] Fix
LockFileExandUnlockFileEx(PR #6625) - [core:math/linalg] Fix negated quaternions in
angle_from_quaternion(PR #6607) - [core:sys/linux] Emulate
dup2better on arm64 and riscv (PR #6632)
Full Change Log
Full Changelog: dev-2026-04...dev-2026-05
dev-2026-04
New Language Features
- Fixed Capacity Dynamic Arrays, adds
[dynamic; N]T
New Compiler Improvements
- Limit maximum exponent of integer literals to avoid unnecessary work
- Add
-in-source-orderoption toodin docto show docs in source order within each file - Support clang-style
+/-prefixes in target features - Fix #6396: handle full-width
bit_fieldliteral masks - Fix #6401: generic cycle deadlock in struct layout
- Add support for other Android architectures
- Fix #5602: "Internal Compiler Error: Type_Info for 'x' could not be found" for poly procs with named parameters
- Continuous Compiler Improvements
- Add initial LLVM 22 support on non-Windows platforms
- More accurate
-march:nativefeature selection - Allow pointers to types which have subtype fields at offset 0 to be assignable in proc parameters.
offset_of(some_bit_field)now raises a proper error- Add
intrinsics.type_field_bit_offset(typeid, string)to get a compile-timebit_fieldfield offset in bits - Add
intrinsics.type_field_bit_size(typeid, string)to get a compile-timebit_fieldfield size in bits (the value after|) - Fix #6407: Global unions do not get initialized correctly
- Fix
>=comparisonbase:runtimedependencies forstringandcstringtypes - Fix
GB_PANICformat string inlb_emit_convinvalid subtype cast - Fix
[^]u16tocstring16transmute condition inlb_emit_conv - Remove
Tildebackend - Fix #6484: Two
whenblocks which evaluate totruein aforeignblock misses following decls - Fix declaration order bug #6506: False-positive
where-clause failures in recursive polymorphic subtype cases - Fix declaration order bugs #5572 and #5961: Incorrectly rejecting valid in-progress/forward union references
New Packages
core:crypto/argon2idinitial importvendor:windows/wasapivendor:sdl3/mixer
Package Improvements
- Add more JS to
examples/all - Update Unicode database to newer version
- Fix modular exponentiation in
core:math/big - Fix
assign_atdocumentation core:fmtnow formatstime.Durationwith a space between duration and unit when using% v- Correct SIMD
shrexample documention - Fix
sync.Auto_Reset_Eventsignal deadlock core:sys/darwin/Foundation: fixNSTimerbinding- Partial 1.619.x d3d12 bindings
core:mem/virtual: Fix for virtual arena overcommit bug #5821core:crypto: Add ARM AES hardware accelerationcore:crypto: Add ARM SHA256 hardware accelerationcore:fmt: Improvecore:container/small_arrayformattingvendor:stb/vorbis: Add WASM supportcore:encoding/cbor: Fix order-dependent partial unmarshalscore:encoding/json: Fix user unmarshaler example- Fix possible OOM in
core:os.get_working_directory()on Linux - Fix #6455:
core:mem/virtual.map_file_from_pathignores flags - Fix for corner case in the
core:memscratch allocator. - Minor cleanup in
base:runtimeandcore:mem - Fix #6495: Handle starting separator during name comparison in
core:os - Improved
delete_keydocumentation to mention it's safe to delete keys while iterating a map as long as no resizes take place - Fix source size check in
core:crypto/aesECB encrypt/decrypt
Full Change Log
Full Changelog: dev-2026-03...dev-2026-04