Resolve the conflict in docs/mkdocs/docs/features/files.md (keep the
measured ifstream ratio of this PR, use develop's json_view/index.md link)
and regenerate the single include headers.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Split the json_view feature page into sub-pages
The single features/json_view.md page grew to ~600 lines covering eight
topics. Split it into features/json_view/ with an overview page and
pages for ownership and lifetime, differences from basic_json, reading
values, generic code, editing, and images.
The new differences page collects what was spread over several sections
("What is the same", "What is different", "Duplicate keys", and notes in
the conversion and serialization sections) under one summary table, and
adds the parts of the basic_json interface a view does not have. The
other pages now link to it instead of restating the rules.
Internal links across the documentation point to the new pages.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Docs: fix stale comments and trim repeated text in json_view pages
- fig-jsonview.js/.css: point the header comments at the new pages
- reading.md: views' strings are valid as long as the document (and a
borrowed source text), as lifetime.md says
- reading.md, generic_code.md: link to differences.md instead of
restating the member-order and repeated-key rules
- differences.md: drop the third mention of ordering comparisons, name
rbegin()/rend() in the iterator row
- exceptions.md: link editable documents to editing.md
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
---------
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
#5835 replaced the switch on the value type in operator*, operator->,
operator++, operator-- and operator== of iter_impl with an if chain that
tests for an array first. Every step over an object then pays for a
failed test and a taken branch: with the iterator inlined, walking
twitter.json took 14% more cycles than with 3.12.0, and citm_catalog.json
3.5% more.
Test for objects first. Walking every value of the three files now
takes fewer cycles than with 3.12.0 (Apple M1 Max, Apple clang 21):
citm_catalog.json 1.21x, twitter.json 1.04x, canada.json 1.17x as fast.
canada.json, which holds only arrays, gives back part of the gain of
#5835 (develop: 1.62x); the other orders tried (arrays first without
hints, objects first without hints, a switch) each left one of the
files slower than 3.12.0.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* docs: add the blog post "How 3.13.0 got faster"
A post for the documentation (Home > Blog) on how 3.13.0 became faster:
eight themes with the pull requests behind them, the ideas taken from other
libraries, what did not work, how the fast paths are tested, and the
compile-time cost. Every number comes from a run on 2026-10-27 (3.12.0 against
develop plus the open non-draft PRs, Apple M1 Max) or from the linked PR.
- docs/mkdocs/docs/home/blog/how_3_13_0_got_faster.md: the post.
- docs/mkdocs/scripts/figures/perf_blog.py: draws its 12 figures (overview,
eight timeline panels, the SWAR test, the cliffs, json_view throughput) as
inline SVGs in the nj-charts style; data/perf_timeline.csv holds the
timeline.
- tests/benchmarks/history: history.cpp and history.py time one fixed program
against the header of every commit that changed it, interleaved, and write
one CSV row per commit with its PR. The timeline of the post comes from it.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* docs: compare json_document with Glaze, sonic, RapidJSON and Rust parsers
Extends the comparison in "How 3.13.0 got faster" from yyjson, simdjson and
Boost.JSON to eleven engines: sonic-cpp, RapidJSON and Glaze (generic DOM,
lazy_json, reading into structs) in C++, sonic-rs, simd-json and serde_json in
Rust. Measured on an Apple M1 Max (Apple clang 21, Rust 1.99) and an x86-64
Haswell server (GCC 13, Clang 18). Every engine must return the same values,
string bytes and bit-identical number sums before it is timed.
The post now answers whether json_document is the fastest parser: it is the
fastest document parser at parse+dump everywhere and at parsing on Apple
silicon, ties with simdjson's DOM on x86-64, and loses to simdjson On-Demand
and Glaze structs when only a few fields are read.
tests/benchmarks/json_view/others holds the programs (bench_all.cpp, the Rust
crate with its Cargo.lock), summarize.py and the pinned versions and checksums.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* docs: captions, test-file table and figure fixes for the 3.13.0 post
- caption every table and figure
- describe what makes twitter.json, citm_catalog.json, canada.json and
jeopardy.json hard, and what each timeline panel exercises
- split the json_view layout figure into the tree and the index, with
the example text above; fix the arrows into the string boxes
- arrows and a "less is better" axis label in the cliffs figure
- name nlohmann/json in the comparison tables and chart
- licenses of the adapted code in the thanks section
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* docs: byline and licenses for the 3.13.0 post
- author and date below the title, styled as .nj-byline; its link gets
no external-link icon
- the license of every library in the thanks section, checked against
each project's license files; credit sonic-rs for QUOTE_TAB, which
orjson adapted from it
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* docs: dump comparison and split acknowledgments in the 3.13.0 post
- a dump workload in bench_all and bench_rs: each engine writes a
document parsed beforehand, and every output must read back with
bit-identical numbers; an optional argument selects one workload
- the post's comparison gains a write table, a throughput chart, a dump
row for all three configurations, and the x86-64 cases it loses
- the thanks section is split into adapted code and borrowed ideas
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* docs: x86-64 dump numbers with #5908's branch-free token digits
The dump of canada.json with GCC on x86-64 lost to yyjson by 0.78 because
GCC compiled #5908's token-digit selections as branches; 9fad526888 fixed
that. The post's x86-64 dump and parse+dump rows of canada.json are now
measured with it (GCC 0.95, Clang 0.91 against yyjson), theme 5 tells the
story, and "How we measured" says which rows include the commit.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Compare JSON libraries by use case in the json_view harness
bench_all and bench_rs now time use cases (validate, sum, select, edit,
minify, write) instead of library functions. Every library may solve a
use case in every way it offers; each variant's result is checked, and
three malformed inputs give its validation level (full, no UTF-8,
partial). summarize.py reports each library's fastest fully validating
variant and a faster one at a lower level, as text and as JSON.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* docs: restructure the 3.13.0 post around themes and use cases
- deeper technical detail and code in the eight themes; pull request
numbers move to the appendix, timeline steps are labelled by theme
- comparison with other libraries by use case, as heatmaps per
platform, with validation levels and the caveats next to the results
- appendix: test files, all 182 suite rows with a histogram, timeline
steps, algorithmic fixes, per-file tables, behaviour changes, setup,
pull requests by theme
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Measure each library's fastest modes in the json_view harness
Address the points of a review from the other libraries' side:
- simdjson 5.0.8 (On-Demand structs, key selectors) and Glaze 9.1.0
- number-skipping modes: yyjson and RapidJSON raw numbers, Boost.JSON
number_precision::none
- reused parsers and allocators for yyjson, RapidJSON and Boost.JSON
- Boost.JSON parse_into, Glaze generic with double numbers, and the
same struct types for nlohmann, Glaze, simdjson and serde
- minified copies of twitter.json and citm_catalog.json
- a fourth probe (trailing garbage) and Glaze "(strict)" variants
- a new random variant order in every round, as a variant's time could
depend on the one before it by up to 15 % on the Mac
- --files for other test files and --conformance for JSONTestSuite,
with conformance-rs.sh for the Rust program
summarize.py also reports each library's best reused variant.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* docs: rerun the library comparison of the 3.13.0 post with fairer rules
The figures and Appendix E use the reruns with each library's fastest
modes. Skipping number conversion while reading is now allowed for
every library; copying number text is still not chosen when writing.
The text now says who set the rules, where libraries that check less
win select, how minified files change the picture, and that the
reused jeopardy.json claim holds with every library reused. A new
section covers 19 held-out files and JSONTestSuite, including where
json_document loses, and lists what was not measured. Every
comparison number was recomputed; Glaze is credited in the intro.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* docs: measure the 3.13.0 post against develop with #5964 and #5908
The suite (Appendix B, the short-version table, the overview and distribution
figures) and the timeline endpoint now measure develop d20c4dcd89 with #5964
and #5908 merged. The geometric mean stays 2.9x (2.95); Dump/signed_ints and
Dump/unsigned_ints gain 1.5x and 2.1x from #5964; GetVectorDouble/floats drops
from 10.3x to 7.0x because the old candidate's suite binary happened to compile
that loop faster (a standalone build takes 0.97 ms with both headers).
The timeline endpoint is the 10-27 value scaled by the ratio of the new to the
old candidate, both measured on 10-30, so that it stays comparable with the
other points of the timeline.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* docs: add three figures to the 3.13.0 post
- short strings: two overlapping 16-byte loads and stores
- eight digits at a time: the three multiplications, lane by lane
- validate on the 19 held-out files, per file, Mac and x86-64
The prose the figures replace is trimmed.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* docs: measure the 3.13.0 post against develop with all open pull requests
The baseline is now develop 9c349b1d7c with every open non-draft pull
request merged, including #5968, #5970, #5971, #5975, #5976, #5978,
#5979, #5980, #5981 and #5982. The suite was rerun on 11-01 with every
step repeated when the load rose above 5 while it ran; the comparison
with other libraries and the held-out files were rerun on 10-31.
- suite: 3.0x geometric mean (median 2.9x); Appendix B, the overview
and the distribution regenerated
- timeline endpoint, Appendix D and the cliffs figure updated
- the held-out figure compares json_document with the fastest other
library
- new paragraphs for the pull requests listed above
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* docs: measure the 3.13.0 post against develop with the open pull requests of 11-02
Suite, timeline endpoint, Appendix D and the scalar-run numbers now come from
develop b2674a4a2f with #5990, #5991, #5992, #5996, #5997, #5998, #5999 and
#6000 merged. The scalar runs are measured against a build without them on
develop's code, which no longer checks for a scalar before entering the loop.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* docs: measure the 3.13.0 post against develop with all open pull requests of 11-02
Suite, timeline endpoint, Appendix D and the scalar-run numbers now come from
develop b2674a4a2f with #5990-#5992, #5996-#6006 and #6008 merged. With #6008,
arrays of one element skip the scalar run, so nested arrays no longer pay for
it. The suite ran three rounds per version; the round of 3.12.0 and of 3.13.0
that differed from the other two was dropped, as the setup table says.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* docs: fix the facts the review of the 3.13.0 post found wrong or imprecise
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* docs: re-measure the 3.13.0 post and apply the review
Measure everything again against develop with the open pull requests of
11-03 (suite, timeline, comparison on the Mac and on x86-64, compile
times, stream gap), add the suite on an Intel MacBook Pro as a check,
mark the ways that keep number text as "N" in the use-case charts, move
the loop-shape box to theme 5, and apply the style and wording fixes of
the review.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* docs: make the comparison of the 3.13.0 post fairer
Three reviews of "How it compares" (a maintainer audit of the harness, a
methodology review, a reader's red team) led to a full rerun on the Mac and
x86-64 with simdjson 5.1.1 and new ways in the harness:
- in-place parses with the copy made before the timer ("(copy untimed)",
never chosen as a library's best) and json_document::parse_copy();
- RapidJSON through StringStream, and with default precision in validate,
which reads no number;
- BENCH_ONLY takes several prefixes.
The post now gives the copy's real cost (11-14 us for twitter.json on the
Mac, not 30), what yyjson gains without it, the run-to-run noise, the LTO
check, the single-file losses that were missing, and qualifies the headline
claim. Every number in the section and Appendix E is recomputed; the
NUL-byte remark moved out of the body.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Format bench_all.cpp with astyle
The amalgamation check runs astyle over tests/ and failed on bench_all.cpp. The variant tables keep one variant per
line between *INDENT-OFF* and *INDENT-ON*; astyle formats the rest, and the list of use cases becomes a named array so
the loop header stays on one line.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Print the history benchmark's results with a stream
Flawfinder flagged the printf in history.cpp (CWE-134), although its format string is a constant. The stream prints the
same text, so history.py reads it as before.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* docs: address review comments on the performance blog post
- fix wording: "are new", "for each input source", "the strings", "the last zero to seven bytes", "by" Danila
Kutenin, "the result of that query"
- introduce `json_document` where it first gets a detail, say that the `-mbranches` flag is not used, explain
why the position and token bookkeeping overloads are empty, and say that a step's percentage is relative to
the time before the step and that a median of 4 is the mean of the two middle values
- write percentages without a space before `%` in the post and in the price figure (regenerated)
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* docs: wrap the prose of the performance blog post at 120 columns
Only line breaks change: the rendered page is identical after whitespace normalization.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* docs: blog post: address the second round of review comments
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* docs: blog post: note that Boost.JSON's tree lead comes from its arena
The comparison builds Boost.JSON's tree in a monotonic_resource. With its
default allocator and precise numbers, boost::json::parse() takes as long
as json::parse() on twitter.json, 5% longer on citm_catalog.json and 17%
less time on canada.json (M1, C++23 -O3, best of 3 x 30 runs; the arena
saves Boost.JSON 28-41%).
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* docs: blog post: comparing already passes the depth; copying measured slower without the counter
operator== walks the standard containers itself and passes the depth as an
argument (#5840). A copy that does the same, inserting element by element,
measured 4-16% slower on an M1 Max with Apple clang, so copying keeps the
thread_local counter.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* docs: blog post: name every user of the depth counter and every input the document copies
The "Bounded recursion" paragraph said only copying and equality for custom
containers use the thread_local counter. Converting from another basic_json
specialization and the ordering comparisons use it too; equality for the
standard containers passes the depth instead.
The "Strings stay in the input" bullet named only parse_copy() and an rvalue
std::string as owning inputs. Streams, FILE* and other input that is not one
contiguous block of bytes are read into an owned buffer as well.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
---------
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
Resolve the conflict in docs/mkdocs/docs/features/performance.md: keep
the description of the FILE* bulk scan from #6030 and the stream timing
from this branch.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
- "default parser" instead of "library parser" in comments and tests
- the three comments in json.hpp say the same thing
- current() replaces the repeated `p != end ? *p : 0`
- the backslashes of each macro are aligned
- the docs say that JSON_DIAGNOSTIC_POSITIONS turns the fast path off
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Scan the buffer of a FILE in bulk with glibc and the Apple C library
Parsing from a FILE* still read one character at a time through
getc_unlocked(), while a std::istream has its get area scanned in bulk.
file_input_adapter now offers the lexer the same kind of window into the
buffer of the FILE: it reads the fields that the C library's own
getc_unlocked() macro reads (glibc: _IO_read_ptr and _IO_read_end; Apple:
_p and _r) and consumes exactly the characters getc_unlocked() would
return, so the file is left where it was left before. The part of the
current token in a buffer that is about to be refilled is kept for error
messages, as for streams. The window needs the file locked for the whole
parse, so JSON_DISABLE_UNLOCKED_FILE_IO and MemorySanitizer builds turn it
off; uClibc, which defines __GLIBC__ but has other fields, does not get it.
parse() from a FILE* gets 2.0-2.6x as fast on an Apple M1 Max and 1.7-2.2x
on x86-64 (glibc, GCC 15 and Clang 21), accept() 2.9-4.4x; that is about as
fast as parsing from a std::ifstream. The new tests read with FILE buffers
of 1 to 70 bytes and unbuffered files, after ungetc(), with SAX positions
and with binary strings across the buffer's edges.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Tests: no setvbuf() buffer below 2 bytes on Windows
The Microsoft C runtime rejects a _IOFBF buffer size below 2 as an
invalid parameter and ends the process (0xc0000409), which made
test-file_input fail on all MSVC and clang-cl jobs.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Fix CI: astyle formatting in unit-file_input.cpp
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Tests: limit the multi-character ungetc() test to glibc and Apple
The Windows C runtime only supports one pushed-back character, which
broke the Windows CI jobs.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01N5NasEsrzHhSNouxSoEwqz
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Tests: sweep_allocator::construct() accepts pointers to const
Recent libc++ (emsdk latest) probes construct(const T*, const T&) in an
unevaluated context while checking for trivial copies; the static_cast to
void* in the noexcept specification then fails to compile.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01N5NasEsrzHhSNouxSoEwqz
(cherry picked from commit 794bf70823c12a2a1cb2a8a95f34dfb6357746a8)
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Tests: guard sweep_allocator::construct() against const types instead of const_cast
The const_cast added in the previous commit is rejected by
cppcoreguidelines-pro-type-const-cast (clang-tidy). Make construct() not
viable for const U instead, which is what libc++ probes for.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01N5NasEsrzHhSNouxSoEwqz
(cherry picked from commit de099fa607001b710862fd1bbecef3609c54cbc4)
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Fix CI: astyle and clang-tidy for sweep_allocator::construct()
astyle 3.4.13 reformats the enable_if<!...> default template argument
(Check amalgamation). Delete the const overload instead, which makes the
libc++ probe for const objects fail in the SFINAE context just the same
and needs no <type_traits>.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01N5NasEsrzHhSNouxSoEwqz
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
---------
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com>
* Binary readers: append strings straight from contiguous input
get_bytes() read every string into a zero-filled resize() and copied the
bytes over it, so each byte was written twice. A string that the input holds
in one block (bulk_remaining() at least its length) is now appended from it
directly. The claimed length is checked against the input first, so a huge
claimed length still allocates nothing it cannot fill; other inputs and
binary values keep the chunked loop.
from_cbor() and from_msgpack() of twitter.json and citm_catalog.json get
5.5-9.1% faster on an Apple M1 Max and 2.7-8.6% on x86-64 (GCC 15, Clang 21);
canada.json, which has few strings, is unchanged.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Binary readers: keep the tag argument of append_from_block() astyle-proof
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Fix CI: bugprone-signed-char-misuse in append_from_block
Convert the (positive) length to its unsigned type before widening it
to std::size_t, as clang-tidy flags the int8_t to std::size_t cast.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Tests: sweep_allocator::construct() accepts pointers to const
Recent libc++ (emsdk latest) probes construct(const T*, const T&) in an
unevaluated context while checking for trivial copies; the static_cast to
void* in the noexcept specification then fails to compile.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01N5NasEsrzHhSNouxSoEwqz
(cherry picked from commit 794bf70823c12a2a1cb2a8a95f34dfb6357746a8)
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Tests: guard sweep_allocator::construct() against const types instead of const_cast
The const_cast added in the previous commit is rejected by
cppcoreguidelines-pro-type-const-cast (clang-tidy). Make construct() not
viable for const U instead, which is what libc++ probes for.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01N5NasEsrzHhSNouxSoEwqz
(cherry picked from commit de099fa607001b710862fd1bbecef3609c54cbc4)
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Fix CI: astyle and clang-tidy for sweep_allocator::construct()
astyle 3.4.13 reformats the enable_if<!...> default template argument
(Check amalgamation). Delete the const overload instead, which makes the
libc++ probe for const objects fail in the SFINAE context just the same
and needs no <type_traits>.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01N5NasEsrzHhSNouxSoEwqz
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
---------
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com>
* Tests: make sweep_allocator::construct() not viable for const types
The newest libc++ (emsdk latest, used by ci_emscripten) probes
construct(const T*, const T&) in an unevaluated context while checking
whether copies are trivial. For the test allocator, instantiating the
noexcept specification then fails to compile, because static_cast<void*>
cannot drop the const qualifier. Constrain construct() to non-const types,
which makes the probe fail in the SFINAE context instead of with a hard
error.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01N5NasEsrzHhSNouxSoEwqz
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Fix CI: astyle and clang-tidy for sweep_allocator::construct()
astyle 3.4.13 reformats the enable_if<!...> default template argument
(Check amalgamation). Delete the const overload instead, which makes the
libc++ probe for const objects fail in the SFINAE context just the same
and needs no <type_traits>.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01N5NasEsrzHhSNouxSoEwqz
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
---------
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com>
astyle 3.4.13 reformats the enable_if<!...> default template argument
(Check amalgamation). Delete the const overload instead, which makes the
libc++ probe for const objects fail in the SFINAE context just the same
and needs no <type_traits>.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01N5NasEsrzHhSNouxSoEwqz
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
astyle 3.4.13 reformats the enable_if<!...> default template argument
(Check amalgamation). Delete the const overload instead, which makes the
libc++ probe for const objects fail in the SFINAE context just the same
and needs no <type_traits>.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01N5NasEsrzHhSNouxSoEwqz
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
The const_cast added in the previous commit is rejected by
cppcoreguidelines-pro-type-const-cast (clang-tidy). Make construct() not
viable for const U instead, which is what libc++ probes for.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01N5NasEsrzHhSNouxSoEwqz
(cherry picked from commit de099fa607001b710862fd1bbecef3609c54cbc4)
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
Recent libc++ (emsdk latest) probes construct(const T*, const T&) in an
unevaluated context while checking for trivial copies; the static_cast to
void* in the noexcept specification then fails to compile.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01N5NasEsrzHhSNouxSoEwqz
(cherry picked from commit 794bf70823c12a2a1cb2a8a95f34dfb6357746a8)
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
The const_cast added in the previous commit is rejected by
cppcoreguidelines-pro-type-const-cast (clang-tidy). Make construct() not
viable for const U instead, which is what libc++ probes for.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01N5NasEsrzHhSNouxSoEwqz
(cherry picked from commit de099fa607001b710862fd1bbecef3609c54cbc4)
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
Recent libc++ (emsdk latest) probes construct(const T*, const T&) in an
unevaluated context while checking for trivial copies; the static_cast to
void* in the noexcept specification then fails to compile.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01N5NasEsrzHhSNouxSoEwqz
(cherry picked from commit 794bf70823c12a2a1cb2a8a95f34dfb6357746a8)
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
The SSE2/NEON tests for the bytes a string run stops at compared with the
quote, with the backslash and with 0x20 separately. Flipping bit 1 maps
0x00-0x1F onto itself and the quote onto 0x20, and every other ASCII byte
to 33 or more, so (c ^ 2) < 33 finds the quote and the control characters
with one compare: signed where the bytes from 0x80 stop the run too,
unsigned (or with a saturating subtraction / minimum on SSE2) where they
do not. This saves one vector operation per 16 bytes in all twelve vector
tests: the lexer's vector_plain_run() and vector_utf8_run(), dump()'s
copy_plain_halves() and copy_plain_vectors(), and json_view's scanner and
serializer. The DEL tests of the ensure_ascii paths stay.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
The usual test for a zero byte, (x - ones) & ~x & high, needs `& ~x` only
to keep bytes from 0x80 out. The tests for quotes, backslashes and control
characters in string_scan.hpp and the serializer also flag every byte from
0x80 with `v & high`, so the `& ~x` terms can be dropped. Clang finds this
reduction itself; GCC does not and emits 24 instead of 16 instructions.
The masks are the same for 1.3 billion random words and for every pair of
adjacent byte values. With GCC 15 on x86-64: accept() -2 to -4 %,
sax_parse() -2 %, dump() -3 %. Clang is unchanged.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
accept() needs the value of a number only for a float whose exponent may
reach the largest value. skip_digits() finds the end of the digits without
the multiply-adds of accumulate_digits(), and once an 8-byte word fails the
digit test, the remaining loop needs no bounds check: a non-digit is in
that word. The value is computed afterwards for the rare floats that need
it.
accept() of canada.json: -16 % on Apple M1, -22 % with GCC 15 and -21 %
with Clang 21 on x86-64; files with many numbers -9 to -15 %.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
- json_sax_error_reporter::parse_error: mark the exception unused, as
JSON_THROW does not use it with JSON_NOEXCEPTION
- unit-contiguous_parser.cpp: compare the exceptions of both paths only
when exceptions are enabled (invalid input aborted the test otherwise)
- silence cpplint's readability/braces false positives on the if
statements inside the parser's macros
- add contiguous_parser.hpp to BUILD.bazel
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
The parser and the lexer exchange one token at a time, and the lexer keeps
its state (the current character, a pending unget, where the token starts)
in members, which every character pays for: json::accept() of twitter.json
took 2.8x as long as json_document::accept(). detail::contiguous_parser
reads text in one block of memory through a local pointer, with the
grammar, the whitespace and the tokens in one loop (after yyjson's reader),
and drives a SAX handler: parse()'s DOM handler, accept()'s handler, or the
user's handler of sax_parse(). It is used with the default options (no
callback, no comments, no trailing commas; sax_parse() with strict = true).
Errors are reported by the library parser, so the exception, its message and
its position stay the same. contiguous_parser stops at the token that the
library parser would reject, before reporting it, so the events up to the
error are the same; then:
- parse() runs the library parser with a handler that only reports the
error (json_sax_error_reporter), with the lexer of parse(), which is
instantiated anyway;
- sax_parse() runs it with a handler that forwards only parse_error() to
the user's handler (json_sax_error_forwarder), so no event arrives twice;
- accept() returns the verdict.
A handler that returns false stops the parse as before. An error at the end
of a large input costs parse() about 1.2x of what it cost before.
On an Apple M1 Max: accept() 1.6-2.0x as fast, parse() 10-16% faster,
ordered_json 11-16%, sax_parse() with a user's handler 1.36-1.46x, small
messages 12-22% (parse()) and 2.1-2.4x (accept()). On x86-64 (GCC 15,
Clang 21): accept() 1.6-1.9x, parse() 5-17%, ordered_json 12-19%,
sax_parse() 1.32-1.57x. A translation unit that parses a string compiles 3%
(-O0) and 9% (-O2) slower.
unit-contiguous_parser.cpp compares the contiguous path with the byte path
(std::list iterators) on corner cases, generated documents, their
truncations and mutations, deep nesting, and narrow number types: the same
values with the same number types and float bits, exception messages,
accept() verdicts and sax_parse() events.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
Add sections on switching off std::format and ranges support to the
compile-times page, with measurements (Apple clang, GCC 16, -std=c++20),
the functionality that is lost, and chart rows (issue #5294).
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Binary reader and writer: remove checks that invariants make dead
Branch coverage of the test suite and of 24k mutated inputs showed
these conditions as never taken; each is implied by a check made
earlier on the same path:
- check_bson_document_size: `document_size < 0` (the size was rejected
below 5 before it was stored)
- get_ubjson_ndarray_size: `second != 'N'` ('N' is rejected as a type
marker before `second` can be set)
- bjd_type_name() == nullptr for an ndarray dtype: every marker
is_valid_optimized_type() accepts for BJData has a name (now an assert)
- the early is_bjd_excluded_optimized_type() check in
get_ubjson_size_type(): is_valid_optimized_type() rejects the same
eight markers with the same error
- `input_format == bjdata` next to `is_ndarray`, bit 8 of the type, or a
'B'/'u'/'m'/'M' marker: these only arise for BJData
- `first != npos` next to bit 8: the bit is only set after a count
- MessagePack writer: the `<= max` halves of the negative int8/16/32
ladder (the ladder is in the `else` of `n >= 0`)
No change in behaviour: a build that aborts on each removed side passes
the test suite and the fuzz corpora, and a 1.6M-input differential run
of the reader against the original headers found no difference.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Remove is_bjd_excluded_optimized_type(), which has no caller left; fix two comments
is_valid_optimized_type() rejects the markers BJData excludes, so the reader
no longer needs its own list; the test now checks is_valid_optimized_type()
of a BJData reader against the writer's list. The BSON size comment now says
that open_bson_document() rejects a size below 5.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Make the BJData reader in the lookup test const (misc-const-correctness)
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
---------
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
json_sax_dom_parser::insert_key_with_index: own the new object_key_index
in a unique_ptr before key_indexes may reallocate; a throwing reallocation
leaked it.
Signed-off-by: Christoph Grüninger <foss@grueninger.de>
output_vector_sink grew the vector by resizing it to its whole new
capacity. resize() zeroes the characters it adds, so every doubling
zeroed megabytes of a fresh allocation at once: that faults in all its
pages and evicts them from the cache before the writer reaches them,
and the writer's stores then miss again. The sink now reserves the same
capacity as before but resizes the vector only to 64 KiB
(vector_sink_window) ahead of the position, which is still cached when
the writer's stores come.
to_cbor(), to_msgpack(), to_ubjson(), to_bjdata() and to_bon8() of an
array of one million numbers into a fresh vector: 4-26 % faster on x86
(Skylake laptop with Apple clang 13, Haswell VPS with GCC 16 and
Clang 23), 1-2.5 % on Apple M1 Max; other documents unchanged (within
0.4 % in reruns). The capacity and growth of the vector are as before.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* UBJSON/BJData reader: reject a container count equal to SIZE_MAX
The reader marks a container whose count is not known by npos, which
is SIZE_MAX. A count of exactly that value was stored unchanged and
then read as "no count", so the BJData input
[ # M ff ff ff ff ff ff ff ff i 01 ]
parsed as [1] instead of throwing out_of_range.408 as a count of
2^64-2 does. Where size_t is 32 bits, the same holds for 'm' and 'L'
counts of 2^32-1, in UBJSON too.
The counts of 'm', 'M' and the signed types now go through
store_ubjson_count(), which rejects npos with "integer value overflow"
(the message a count beyond size_t already gets), as the CBOR reader
does in get_cbor_container_size(). An ND-array dimension of 2^64-1 gets
the same message now, where it was "excessive ndarray size caused
overflow" (the product check) on 64-bit platforms.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* MessagePack reader: reject an array/map 32 length equal to SIZE_MAX
On a platform with a 32-bit std::size_t, an array 32 or map 32 length of
0xFFFFFFFF equals unknown_size(), so start_array()/start_object() were
told the size was unknown and skipped the max_size() check. Route both
through the CBOR length check, renamed get_container_size(), which
already rejects that value.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
---------
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Fix the build of develop: tests from #5997 and #5991
- unit-msgpack-cpp17: assign the result of json::parse in
CHECK_THROWS_WITH_AS (GCC -Werror=unused-result)
- unit-json_view: failing_allocator::allocate is [[noreturn]]
(clang -Wmissing-noreturn)
- unit-class_lexer: build the quoted string once
(clang-tidy performance-inefficient-string-concatenation)
- unit-allocator: sweep_allocator provides pointer, reference and size
typedefs, which GCC 4.8's std::map takes from the allocator
- unit-allocator: skip the "sax_parse into an existing value" sweep with
iterator debugging, like the parser callback sweep: MSVC's std::vector
default constructor is noexcept but allocates its debug proxy through
the allocator, so a failing allocation in create<array_t>() terminates
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Fix the remaining CI failures of the tests from #5997 and #5991
- unit-custom-object-key-type: drop the redundant member initializer and
return a braced init list (clang-tidy); direct-initialize the converted
value, as copy-initialization makes GCC 4.8 instantiate the copy
constructor, which needs a range constructor strict_object lacks
- unit-stream-bulk-scan: pick the claimed get area size by tag dispatch
instead of a constant condition (MSVC C4127)
- unit-allocator: skip the parser callback and sax_parse sweeps with
Clang 3.4 and 3.5, which leak memory when an element constructor throws
inside a libstdc++ std::map copy or std::vector reallocation
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Fix the CI failures that the first failing TU of each job hid
- unit-stream-bulk-scan: compute the claimed size without a condition;
the tag-dispatched template was unused with a 64-bit std::size_t
(-Wunused-template)
- unit-json_view: give failing_allocator the typedefs, rebind and
members that GCC 4.8's std::map takes from the allocator directly
- clang-tidy hits in the tests from #5997: raw string literals
(unit-json_view), redundant c_str() and a NOLINT for &text[0]
(unit-locale-cpp), reserve() before a loop (unit-ordered_json), and a
NOLINT for a return that a braced list would turn into an array
(unit-regression1)
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Skip the 3 GiB get-area test with a 32-bit std::size_t in the preprocessor
The claimed size was a constant 0 on Win32, so `claimed == 0 || ...` was a
constant condition and MSVC reported C4127. Guard the section with
`#if SIZE_MAX > UINT32_MAX` instead, as unit-msgpack-custom-types.cpp does.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Silence modernize-use-transparent-functors for failing_map in unit-json_view
std::less<> needs C++14, and the test is also built as C++11.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
---------
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
json_view number(): #6006 folded the check of the digit after a minus
sign into `else if (!negative || ...)`, which MSVC 19.44 reports as
C4127 (conditional expression is constant) and /WX turns into an
error in every MSVC and clang-cl job. The check now sits in the
`if (negative)` block, right after the sign is skipped; the error and
its position are unchanged.
unit-locale-cpp.cpp: the locale_switching_token added in #5997 held a
`const char*`, so GCC's -Weffc++ asked for a copy constructor and copy
assignment. The locale name is now a std::string.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
Since #6004 made the "while parsing" part unconditional, the three appends
at the start of exception_message() are one fixed string around the
context. Review follow-up to #6004.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
enter_cbor_array and enter_msgpack_array called the out-of-line scalar
run for every non-empty array. When the array's element is an object or
an array, as on every level of [[...]] or the single ring of a GeoJSON
polygon, the call reads nothing; it made 1,000 arrays nested in each
other cost 4 % more instructions (GCC 13) and up to 8 % more time (Apple
clang) than without the run. #5975 had fixed that with an inline
pre-check, which was later dropped after measuring canada only.
The run is now entered only for arrays of two or more elements: the same
comparison as before, so no other row changes. Arrays of one scalar,
which the run read faster, now take the general path; they are rare in
documents (one in citm_catalog, none in twitter or canada). The new
benchmark FromBinaryShape/singles measures them, next to
FromBinaryShape/nested.
Instructions per repetition, new/old, GCC 13 callgrind: nested arrays
0.973 (CBOR) and 0.971 (MessagePack), arrays of one scalar 1.037 and
1.021, every other row within 0.15 %.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Create the DOM parser's member value before placing it (#5985)
json_sax_dom_parser::set_member() constructed the new value in place over
the null member. basic_json(value_t) sets the type before the json_value
allocates, so a std::bad_alloc left a member typed object/array with a
null pointer, and the destructor crashed instead of propagating the
exception. Construct the value first and move it over the member (the
move constructor is noexcept).
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Fix CI for the parse() bad_alloc sweep test (#5985)
- sweep_allocator: provide max_size(), construct() and destroy(), which
GCC 4.8's std::map calls on the allocator directly.
- Declare the non-throwing test lambdas noexcept (-Werror=noexcept on
recent GCC).
- Give the sax_parse lambda an explicit return type; Clang 3.4 crashes on
the deduced one in C++11 mode.
- Skip the parser callback sweep with iterator debugging (MSVC debug
builds, clang-cl), where containers allocate debug proxies inside
noexcept move constructors and terminate instead of throwing, as in the
other allocator tests.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01N5NasAgbDGG28MqXjQvLqp
Signed-off-by: Claude <noreply@anthropic.com>
* Construct a scalar member in place; only a throwing value goes through a temporary
Creating the value first and moving it over the member is needed only when
creating it can throw (a string, a binary value or a container). A scalar
or null is again constructed in place, without the extra move and destroy
per member (and, with JSON_DIAGNOSTICS, without the extra set_parents()).
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Spell the nothrow trait without && and fix the construct_member comments
The dispatch in set_member tests std::is_nothrow_constructible, not
whether the value is a scalar. Say so in the comments, and drop the &&
from the trait argument (equivalent for a forwarding reference) so that
astyle no longer inserts spaces around the template brackets.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Say "for example" in the construct_member comments
The overloads are selected by std::is_nothrow_constructible, not by the
kind of value: a custom type that cannot throw takes the first one.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
---------
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
Signed-off-by: Claude <noreply@anthropic.com>
Co-authored-by: Claude <noreply@anthropic.com>
* basic_json, iterators, serializer: remove checks that invariants make dead
Branch coverage of the test suite and of 24k mutated inputs showed
these conditions as never taken; each is implied by the code around it:
- copy_object_level(): `src_it != src_object.cend()`; dst was built from
exactly src_object.size() unique keys and src_it advances at most once
per member
- the general comparison operators (equal_general, less_general,
three_way_general): the five same-type scalar cases, which the inline
fast path of every operator handles before calling them (now an
assert); and the NaN half of compares_unordered() in the mixed-type
branch, reached only after all six number pairs returned, where only
a discarded operand is unordered
- serializer, ensure_ascii: the reset of `bytes` under error_handler_t::keep
and the surrogate-pair escape in the UTF8_ACCEPT case; every
well-formed multi-byte character is escaped by write_u_escapes() before
decode() sees it, so decode() only ever accepts DEL here, a single byte
of which nothing was buffered (now asserts)
- iter_impl(pointer): the if/else chain assigned value-initialized
iterators to the value-initialized m_it
No change in behaviour: a build that aborts on each removed side passes
the test suite, the fuzz corpora and 23.2M dumps of all 1-, 2- and
boundary 4-byte strings in both ensure_ascii modes.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* iter_impl: keep the type-dependent initialization in the constructor
The check is not needed for correctness (m_it is value-initialized), but
without it GCC 13 generates slower iteration loops: a recursive walk over
a parsed document takes 0.5-2.7 % longer (callgrind: +1.9 % instructions
on twitter.json, +0.95 % on canada.json). Clang generates identical code
either way. See the measurements in #6005.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
---------
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
Branch coverage of the test suite and of 24k mutated inputs showed
these conditions as never taken; each is implied by the code around it:
- view_serializer::write_codepoint<EnsureAscii>(): the only caller
passes the byte a run of plain output stopped at (a quote, a
backslash or a control character), so the function is now
write_escaped(), which writes that byte's escape sequence
- load_image(): `owned_image.empty() || image != owned_image.data()`;
both callers hand in a fresh document_data, and for load(vector&&)
the image is owned_image itself
- set_moved(): reserve_moved(); all three callers reserved first or
have the moved flag set (the documented precondition)
- view_side::string() and json_side::string() in equal(): only used
after both sides were found to be strings, so they read the string
directly (and are noexcept)
- number<false>(): the '1'..'9' test after a sign-less dispatch, which
only happens for a digit
- value_range(): the pre-loop `is_new || off >= size` test, which the
loop's first pass repeats on the same node
No change in behaviour: a build that aborts on each removed side passes
the test suite and the fuzz corpora.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
Without the same-type scalar cases, the macro body has no use for it, and
JSON_IMPLEMENT_SCALAR_CASES is now used only by JSON_IMPLEMENT_SCALAR_FAST.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
Branch coverage of the test suite and of 24k mutated inputs showed
these conditions as never taken; each is implied by the code around it:
- lexer: `next_unget` in get_codepoint_bulk() and scan_literal_bulk().
unget() is only called at the end of scan_number(), in scan_comment()
and in skip_bom(); every token scan starts with a plain get() in
skip_whitespace() that consumes a pending unget, and nothing between
that get() and the bulk helpers ungets (now asserts, as in
get_after_string_run())
- lexer: `number_type == value_integer` in convert_integer(): both
callers exclude value_float and the type has three values
- parser: `!context.empty()` in exception_message(): all callers pass a
literal
- input_stream_adapter: `refill` in get_character_slow(): the function
is only called after get_character() found the get area exhausted,
and the variable recomputed that comparison
- json_sax_dom_callback_parser: `ref_stack.back()` in key() (the null
case returned a few lines earlier, and the callback in between cannot
reach ref_stack), `is_structured()` in end_object() (non-null entries
are only pushed for objects and arrays), `!key_stack.empty()` in
current_key() (a kept object pushed its key before any value)
No change in behaviour: a build that aborts on each removed side passes
the test suite, the fuzz corpora and 40k random documents through every
input adapter in four macro configurations.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* CI: measure line coverage without lcov's unexecuted-block adjustment
Drop geninfo_unexecuted_blocks=1 from the lcov capture of ci_test_coverage:
it reset every line of an inlined or templated function to 0 as soon as one
copy had an unexecuted block, which understated the line coverage by about
three points.
Add -fvar-tracking-assignments to both coverage builds. At -O0, GCC inlines
always_inline functions before it places the coverage counters, so the call
sites of such helpers (put_literal() and others) got no line record and
showed as 0 in an -O3 translation unit that never reaches them. With the
flag, the call sites carry counts in every translation unit. The generated
code is identical.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Remove unreachable code found by the coverage analysis
Remove code that no input can reach, and mark the lines that cannot be
covered with LCOV_EXCL_LINE:
- BJData writer: the dry_run parameter of the ND-array element writers was
always true; replace them by range checks.
- BON8 writer: call throw_integer_out_of_range() directly instead of
write_bon8_value() for values that always throw.
- binary_writer: integer_bits() is only called for integers of up to 64 bits.
- serializer, view serializer, lexer: replace guards that the callers make
unreachable by assertions or by the simpler code (surrogate pairs under
ensure_ascii, next_unget, '-' at the end of the buffer).
- json_view: drop unreachable branches in scalar(), set(), check_utf8_scan(),
append_input_impl() and unescape().
- Mark exception-cleanup braces, 2 GiB limits, big-endian and 32-bit code, and
fall-through case labels with LCOV_EXCL_LINE.
No behavior changes for any input that reaches these functions.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Tests: cover the remaining lines of the library headers
Add tests for the lines of include/nlohmann that were not covered: the
binary readers and writers, the lexer and locale paths, the json_view
modules, the custom number, key and array types, json_pointer, json_patch,
diff and the iterators.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Fix test failures in the coverage tests
- unit-custom-object-key-type: remove the unused make_strict_key(strict_key)
overload, the unused range constructor and default constructor, and guard
data() with JSON_DIAGNOSTICS (unused-function/-member-function errors with
GCC and clang -Weverything).
- unit-json_view_compare: move the two new sections to the new
unit-json_view_compare_values.cpp, as the object file got too large for the
MinGW assembler again, and use a std::map with std::greater instead of
std::unordered_map, which does not accept the incomplete basic_json
(clang 3.9, icpc).
- unit-json_view_dump: rename the variables that hid outer ones (MSVC C4456).
- unit-stream-bulk-scan: restore the real size of the oversized get area
before the adapter is destroyed (JSON_PRECISE_STREAM_POSITION reads the
character at the read position), and skip the test where the streambuf
cannot hold a get area larger than an int (MSVC).
- unit-regression1: with JSON_NO_THREAD_LOCAL, copies are built member by
member and keep correct timestamps, so check the corresponding outcome.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01N5NasAgbDGG28MqXjQvLqp
Signed-off-by: Claude <noreply@anthropic.com>
* Return std::size_t from the hash helper in the ordered_json test
MSVC (Win32) rejected the implicit conversion of the unsigned long long
product to the 32-bit std::size_t (C4244 as an error). Also note when
JSON_NO_THREAD_LOCAL is set in the regression test.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01N5NasAgbDGG28MqXjQvLqp
Signed-off-by: Claude <noreply@anthropic.com>
---------
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
Signed-off-by: Claude <noreply@anthropic.com>
Co-authored-by: Claude <noreply@anthropic.com>
* Keep the byte range in exceptions about the root value (#5987)
With JSON_DIAGNOSTICS and JSON_DIAGNOSTIC_POSITIONS both set, exception::diagnostics()
and the json_view diagnostics() returned early for a value without a path (the
root), dropping "(bytes a-b)". Build the "(path) " prefix only when there are
tokens and always append the byte range.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
* Guard the null value of a value-initialized iterator; drop the unneeded found flag
With the root's byte range, diagnostics() reaches get_byte_positions() also
with JSON_DIAGNOSTICS when there is no value, as for a value-initialized
iterator (invalid_iterator.212). get_byte_positions() now returns no range
for a null value, the same change as #6003, so either can be merged first.
In the view, a value that value_path() does not find leaves no tokens, so the
found flag added nothing to the tokens.empty() test.
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
---------
Signed-off-by: Niels Lohmann <mail@nlohmann.me>
Niels Lohmann