v1.0.0

Announcing the first stable release of Pyjion!

What's Changed

• Improve performance of CALL_METHOD by @tonybaloney in #258
• Update to .NET 6 (preview 3) by @tonybaloney in #260
• Optimize COMPARE_OP for floats by @tonybaloney in #261
• Add simple test for function declaration by @tonybaloney in #259
• Remove old escaping/boxing API by @tonybaloney in #265
• Update internal opcode and opargs to fixed width types by @tonybaloney in #263
• Fix native disassembly offset reporting by @tonybaloney in #269
• Patch to 6.0.100-preview.4.21255.9 by @tonybaloney in #273
• Do escape analysis and unbox floats for binary operations (where it makes sense) by @tonybaloney in #271
• Use a different compile flag for instruction graphs by @tonybaloney in #275
• Optimizations table now reflect that OPT-4 is enabled by default by @thedrow in #278
• Unbox integers by @tonybaloney in #276
• Refactor absint to use consistent types for opcode index, opcode and oparg by @tonybaloney in #281
• Unbox boolean values by @tonybaloney in #283
• fix typo by @jugmac00 in #284
• Implement LOAD_FAST, STORE_FAST and LOAD_CONST for unboxed values. by @tonybaloney in #285
• Fix CALL_METHOD assertions by @tonybaloney in #292
• Update to .NET 6 preview 5 by @tonybaloney in #294
• Add a runpy entrypoint by @tonybaloney in #295
• Add a python API for instruction graphs. by @tonybaloney in #297
• Change the import to not throw a system exit signal. by @tonybaloney in #298
• Add some extra tests for the test_ast crash by @tonybaloney in #293
• Implement generators and the YIELD_VALUE opcode by @tonybaloney in #277
• Remove overuse of size_t with fixed-width types to avoid runtime errors by @tonybaloney in #305
• Add a pickle test and a spectral norm benchmark by @tonybaloney in #307
• Fix NumPy comparison operators returning arrays being yielded back to bools by @tonybaloney in #313
• Fix pyyaml decoding on PGC compile by @tonybaloney in #311
• Correct the ord() asserted return type by @tonybaloney in #316
• Print offsets as comments to help the lexers by @tonybaloney in #317
• Integer power and inplace power will always yield an abstract big integer (native python long) by @tonybaloney in #318
• Update to preview 6 by @tonybaloney in #320
• Make the instruction dumps easier to read by @tonybaloney in #321
• Add a symbol table and improve the disassembler output by @tonybaloney in #322
• Fix a bug where the contains operation wasn't being checked for null values by @tonybaloney in #324
• Initialize the JIT after import. by @tonybaloney in #319
• Clean up some forward declaration warnings by @tonybaloney in #325
• Fix a bug in floating point arithmetic with negative values by @tonybaloney in #327
• Refactor tests into base class. Fix crash on import from cache by @tonybaloney in #314
• Rewrite documentation by @tonybaloney in #332
• Fix the native disassembler double line break by @tonybaloney in #333
• Implement a better Pyjion CLI by @tonybaloney in #334
• Fix the crashing macOS test by @tonybaloney in #335
• Update init.py by @AlexWaygood in #337
• LOAD_BUILD_CLASS will infer a function type instead of Any by @tonybaloney in #338
• Remove stelem by @tonybaloney in #340
• Add abstract types for all builtins by @tonybaloney in #339
• Test all builtins by @tonybaloney in #342
• More unit tests by @tonybaloney in #343
• Capture optimisations made in compilation by @tonybaloney in #344
• Conditional branches in instruction graph by @tonybaloney in #346
• Fix branch yield by @tonybaloney in #347
• Add the nqueens and xml etree benchmarks. by @tonybaloney in #341
• Fixed a deoptimization bug. by @tonybaloney in #348
• Fix a regression on unboxed integers, caused by PGC values being set as Any by @tonybaloney in #350
• Math float miscalculation by @tonybaloney in #351
• Pyjion will infer that range(n) generates integers in iterator to improve unboxing by @tonybaloney in #353
• .NET 6 Preview7 by @tonybaloney in #358
• try..except statements. by @tonybaloney in #355
• Update jump instructions by @tonybaloney in #362
• update to .net6 rc1 by @tonybaloney in #364
• Windows ssize patch by @tonybaloney in #365
• Adopt the new 3.10 frame state fields by @tonybaloney in #363
• Add exception check for LOAD_METHOD by @tonybaloney in #370
• Implement PEP626 for 3.10 by @tonybaloney in #371
• Make 3.10 the main develop branch by @tonybaloney in #374
• Add a test for the strides crash by @tonybaloney in #376
• Change the API to have a single config method by @tonybaloney in #380
• Update JitInfo to include the tracing and profiling by @tonybaloney in #383
• Apple Arm64 support by @tonybaloney in #386
• Update to 3.10 proper by @tonybaloney in #387
• Relocation implementations for ARM by @tonybaloney in #388
• Add a truthy float test for ARM64 by @tonybaloney in #389
• Dont try native dis on anything other than x64 by @tonybaloney in #390
• Implement generators for 3.10 by @tonybaloney in #385
• Fix the way optimization levels are set and add a test for it by @tonybaloney in #395
• Reformat C++ code using CLang-format by @tonybaloney in #394
• Add a specialised container type pattern by @tonybaloney in #396
• Linux ARM64 builds by @tonybaloney in #398
• add PGC to STORE_FAST by @tonybaloney in #397
• Fixes #401 by @tonybaloney in #402
• Update to .NET 6 RC2 by @tonybaloney in #403
• Create an unboxed big integer type by @tonybaloney in #405
• Improve function call performance by @tonybaloney in #406
• Rip out bigints by @tonybaloney in #409
• Add an option to optimistically unbox integers when the values were observed to be "small" by @tonybaloney in #411
• Add an unboxed range iterator by @tonybaloney in #412
• Fix the optimization flag numbers by @tonybaloney in #418
• 1.0 release by @tonybaloney in #416
• Observe class attributes to establish a better graph by @tonybaloney in #419

New Contributors

• @thedrow made their first contribution in #278
• @jugmac00 made their first contribution in #284
• @AlexWaygood made their first contribution in #337

**Full Changelog...

1.0.0rc4

• Added ARM64 for Apple macOS (M1 Silicon). The JIT will emit native ARM64 assembly. Disassembler not supported.
• Added Linux/ARM64 support and a manylinux wheel
• Added Windows/ARM64 support
• All JIT configuration is now set using the pyjion.config() function instead of enable_feature()/disable_feature()
• Fixed a bug where pyjion.graph() would cause an exception if graphing was not enabled when the function was compiled
• Added support for the SETUP_ANNOTATIONS opcode
• Added support for generators in 3.10
• Fixed a bug in the DUP_TOP_TWO opcode with escaped variables
• UNPACK_SEQUENCE will emit types for constant tuples of string, int and float
• STORE_FAST is now profiled during the probe stage of PGC, leading to drastically improved performance in many cases where unboxed ints and floats can be used
• Added a simple benchmark suite script (Tests/benchmarks/suite.py)
• Upgraded scikit-build to 0.12 for native VS2019 support

1.0.0rc3

• Updated to .NET 6 RC1
• Added Python 3.10 support
• Experimental try..except statements, stacked exception handlers and exception filtering. Disabled by default
• Fixed a bug on LOAD_METHOD not raising an exception immediately if the method could not be resolved
• Remove enable_tracing() and enable_profiling functions in 3.10 builds. Profiling and tracing is detected on compilation and available via the .info() method
• Added support for PEP626 (fast line tracing)
• Extended profiling support for CFunction calls
• PyTrace_RETURN now correctly sends the return value as an argument as per CPython spec
• Fixed a bug causing SEG when range iterators are escaped

1.0.0rc2

• BINARY_MULTIPLY and BINARY_POWER will assume the resulting integer is a big integer (not unboxed)
• Introduced two optimizations IntegerUnboxingMultiply and IntegerUnboxingPower which are applied at optimization level 2. Try level two if you work with integers, but at smaller values to see better performance.
• Pyjion will infer that range(n) generates integers in iterator to improve unboxing
• LOAD_BUILD_CLASS will infer a function type instead of Any (#42)
• Instruction graphs will include the name of fast locals
• Instruction graph const values are capped to 40 characters
• Added abstract types for all builtins (#339)
• pyjion.info() will now return a JitInfo object
• Optimization flags that were applied to a function during compilation are available in JitInfo.optimizations
• All optimizations are now runtime flags instead of compile-time features
• Unboxing PGC errors will raise pyjion.PyjionUnboxingError (ValueError) instead of ValueError
• Instruction graphs will show conditional branches (in orange)
• Fixed a bug in generators where mixed unboxed/boxed fast locals would yield the wrong values with PGC between the first and second compilation stages
• Fixed a de-optimization that happened in rc1 due to PGC asserting the abstract kind on the profiling phase and then always asserting that integers were big-integers
• Fixed a bug where unboxed locals were decrefing frame locals on yield
• Generators will not unbox fast locals for stability reasons
• Fixed a regression on unboxed integers, caused by PGC values being set as Any
• Assert return types for float object methods (as_integer_ratio, conjugate, is_integer, hex)

1.0.0rc1

• Added pyjion command-line script to complement the python -m pyjion command
• The pyjion CLI has flags for enabling profiling, tracing, optimization level, graphs and debugging
• Unboxing integers that don't fit into long long will raise a ValueError.
• Pyjion will mark any values above 1 billion as "big integers" and not escape them to reduce the chance of overflows.
• Floating point __pow__ with negative values matches all behaviour of CPython
• Raising 0 to a negative power will raise a ZeroDivisionError
• PGC no longer uses a reference to probed values, dramatically reducing memory consumption between the first and second compile cycles
• Fixed a bug where statistics.variance([0, 0, 1]) would raise an assertion error because of an overflow raised in Fraction arithmetic (#326)
• Fixed a bug on calling sys.settrace(None) would cause a segmentation fault (#330)
• Fixed a bug on optimized calls to custom types would crash on the 3rd execution because of the way PGC held and released references.
• Refactored Pyjion's test suite to Pytest
• Rewrote the documentation site
• Fixed a bug in the native disassembler printing empty comment lines
• Corrected the type signature of pyjion.get_offsets()
• Fixed a bug on changed methods for object (like a global) causing crashes because of the way load_method was optimized (#335)

1.0.0b7

• Added pyjion.symbols(callable) API to fetch the dictionary of external call tokens
• Extended the dis() and dis_native() methods with a flag to not print the program counter (print_pc=False)
• Improved the dis_native() method to print the name of the method call after the call instructions as a line comment
• Fixed a bug in dis_native() which showed unresolved sequence points at the top of the printout
• Fixed a bug where in (CONTAINS_OP) result wasn't being checked for exceptions and the next operation would segfault if the in operation returned an error result.
• The IL in dis() is closer in syntax to ILDasm and easier to read
• Added a pyjion.status() method to get runtime data on the JIT
• Windows will now observe the DOTNET_ROOT and DOTNET_LIB_PATH environment variables

1.0.0b6

• Updated to .NET 6 preview 6
• Fixed a bug where ord() builtin would return the wrong type (#315)
• pyjion.dis.dis() and pyjion.dis.dis_native() will show sequence points as comments
• The BINARY_POWER and INPLACE_POWER opcodes will always return a native python long instead of an escaped integer, to avoid overflows

0.13.0

• The compiler will now fail (and default back to CPython) if .NET emits a FAST_FAIL helper
• UNPACK_SEQUENCE is rewritten to be more efficient and use optimized paths for LIST and TUPLE types
• f-string (BUILD_STRING) is rewritten to be more efficient
• UNPACK_EX is rewritten to remove the requirement for dynamic heap allocation (and the stack canary) and leverage .NET compiler's dynamic eval stack
• PGC implemented for UNPACK_SEQUENCE
• PGC implemented for BINARY_SUBSCR
• PGC implemented for CALL_FUNCTION/OPT-14

0.11.0

• Added OPT-13 (OPTIMIZE_TYPESLOT_LOOKUPS) to optimize the type slots for all binary operators and resolve the precedence at compile-time (only for known types)
• Added OPT-14 (OPTIMIZE_FUNCTION_CALLS) to optimize calls to builtin functions
• Optimize all frame locals by determining abstract types on compilation
• Bugfix: Fixed a crash on f-strings with lots (>255) arguments
• Bugfix: Will now skip all functions containing the use of exec() as it contains frame globals which are not supported
• Updated to .NET 5.0.3
• Updated the containers to Ubuntu 20
• Added fileobject abstract type
• Added enumerator abstract type
• Added code object abstract type
• Added integration tests for reference leaks for all binary operations (thanks @amaeckelberghe)
• Added module type (thanks @vacowboy75)

0.10.0

• Added OPT-12 (OPTIMIZE_BUILTIN_METHOD) to pre-lookup methods for builtin types and bypass LOAD_METHOD (PyObject_GetMethod)
• Optimized LOAD_METHOD to recycle lookups for the same object
• Expanded OPT-8, OPT-9, OPT-11, OPT-12 for nested stacks (e.g. inside expressions)
• Added a frozen set abstract type