This is a patch release for issue #2690

This is a break patch fixing On-Demand for wildcard matches. I usually do not release patches that are code breaking but we have a design flaw with respect to wildcard matches and we need to remove it from circulation before people start using.

Full Changelog: v4.6.1...v4.6.2

What's Changed

• parser.load(string_view) does not respect view length when opening files by @jmestwa-coder in #2659

Full Changelog: v4.6.0...v4.6.1

What's Changed

• Fix fopen usage with string_view by @uwezkhan in #2652
• correct inverted clamp in set_max_capacity by @jmestwa-coder in #2651
• add SIMDJSON_INSTALL option by @safesintesi in #2649
• Avoid overflow in padded_string_builder::reserve by @uwezkhan in #2654
• Add overflow checks in padded_string_builder::reserve by @uwezkhan in #2653
• add get_int32() and get_uint32() to the On Demand API by @eyupcanakman in #2638

New Contributors

• @uwezkhan made their first contribution in #2652
• @jmestwa-coder made their first contribution in #2651
• @safesintesi made their first contribution in #2649
• @eyupcanakman made their first contribution in #2638

Full Changelog: v4.5.0...v4.6.0

What's Changed

• Fix PCH build failure with on non-C++20 compilers by @Raju1173 in #2646
• feat: Add DOM tape support for big integers (opt-in) by @harshanagd in #2640

New Contributors

• @gnusi made their first contribution in #2641
• @Raju1173 made their first contribution in #2646
• @harshanagd made their first contribution in #2640

Full Changelog: v4.4.2...v4.5.0

What's Changed

• Fix RISC-V compilation without RVV by @kxxt in #2637

New Contributors

• @kxxt made their first contribution in #2637

Full Changelog: v4.4.1...v4.4.2

What's Changed

• fix failing CI and rename misleading workflow name by @jaja360 in #2630
• fix build issue under POWER when altivec not enabled by @lemire in #2631

Full Changelog: v4.4.0...v4.4.1

What's Changed

• Add LoongArch LSX support for find_next_json_quotable_character by @Sudhanshu-S3 in #2608
• simplifying the find_next_json_quotable_character fnc for neon processors by @lemire in #2613
• new test: iterating over a simdjson_result<padded_string> by @lemire in #2614
• Add RISC-V RVV support for find_next_json_quotable_character by @Sudhanshu-S3 in #2617
• altivec version of find_next_json_quotable_character by @lemire in #2622
• adding memory-file mapping by @lemire in #2625
• disabling loongson LASX runtime dispatching everywhere but GCC15+ by @lemire in #2626

New Contributors

• @Sudhanshu-S3 made their first contribution in #2608
• @Asmod4n made their first contribution in #2616

Full Changelog: v4.3.1...v4.4.0

What's Changed

• fix bad hint (minor perf issue) by @lemire in #2610

Gotchas

If you run your code with development checks (in debug mode, or by setting #define SIMDJSON_DEVELOPMENT_CHECKS 1 prior to including the simdjson header), then the 4.3 series has additional checks for unsafe coding practices with respect to iterators. It is therefore possible that code that appeared to work with 4.2 or earlier might not work anymore. Refer to our documentation for the safe use of iterators.

Full Changelog: v4.3.0...v4.3.1

What's Changed

• Add runtime dispatching for loongarch by @lemire in #2575
• moving the builder files in their own directory by @lemire in #2578
• Build perf: speed up simdjson build by adding PCH, parallel compilation and optimizing linker settings by @es110 in #2588
• work on the ondemand iterators by @lemire in #2590
• adding padded string builder by @lemire in #2592
• Add FracturedJson formatting support for DOM serialization by @FranciscoThiesen in #2580
• Add Vector-Length-Specific RISC-V Vector extension support using generic backend by @camel-cdr in #2593
• when calling get_string with a mutable string parameter, we want to only use the string buffer as scratch space by @lemire in #2595
• faster amalgamation script by @lemire in #2597
• updating the reflection benchmarks by @lemire in #2598
• inlining get_single_implementation by @lemire in #2600
• perf: SIMD string escaping and batch integer formatting optimizations by @FranciscoThiesen in #2605
• When using C++11, we could violate the one definition rule by @lemire in #2606

Gotchas

If you run your code with development checks (in debug mode, or by setting #define SIMDJSON_DEVELOPMENT_CHECKS 1 prior to including the simdjson header), then the 4.3 series has additional checks for unsafe coding practices with respect to iterators. It is therefore possible that code that appeared to work with 4.2 or earlier might not work anymore. Refer to our documentation for the safe use of iterators.

New Contributors

• @arthurscchan made their first contribution in #2574
• @es110 made their first contribution in #2588
• @camel-cdr made their first contribution in #2593

Full Changelog: v4.2.4...v4.3.0