* Make it possible to check that an implementation is supported at runtime.
* add CI fuzzing on arm 64 bit
This adds fuzzing on drone.io arm64
For some reason, leak detection had to be disabled. If it is enabled, the fuzzer falsely reports a crash at the end of fuzzing.
Closes: #1188
* Guarding the implementation accesses.
* Better doc.
* Updating cxxopts.
* Make it possible to check that an implementation is supported at runtime.
* Guarding the implementation accesses.
* Better doc.
* Updating cxxopts.
* We need to accomodate cxxopts
Co-authored-by: Paul Dreik <github@pauldreik.se>
* Adding new files.
* Better.
* Fixing minifier and adding tests.
* Adding benchmarks.
* Including the array header.
* Replacing old stream-based code by the new code.
* Doubling up the itoa.
* Hidden away to_chars in internal namespace.
* Removing the repetitions.
* Documented the atoi functions.
* Tuning the escape sequences.
* Moving the operators off the main namespace.
* Added more tests.
* Tweaking the implementation so that it works with and without exp.
* The string_builder template and mini_formatter class
are not part of our public API and are subject to change
at any time!
* Adding a benchmark and some optimization.
* Cleaning.
* Strictly speaking, this header is needed.
* This would disable bash scripts under FreeBSD.
* Let us also disable GIT.
* Let us try to just disable GIT
* Nope. We must have both bash and git disabled.
In the parse_many function, we have one thread doing the stage 1, while the main thread does stage 2. So if stage 1 and stage 2 take half the time, the parse_many could run at twice the speed. It is unlikely to do so. Still, we see benefits of about 40% due to threading.
To achieve this interleaving, we load the data in batches (blocks) of some size. In the current code (master), we create a new thread for each batch. Thread creation is expensive so our approach only works over sizeable batches. This PR improves things and makes parse_many faster when using small batches.
This fixes our parse_stream benchmark which is just busted.
This replaces the one-thread per batch routine by a worker object that reuses the same thread. In benchmarks, this allows us to get the same maximal speed, but with smaller processing blocks. It does not help much with larger blocks because the cost of the thread create gets amortized efficiently.
This PR makes parse_many beneficial over small datasets. It also makes us less dependent on the thread creation time.
Unfortunately, it is going to be difficult to say anything definitive in general. The cost of creating a thread varies widely depending on the OS. On some systems, it might be cheap, in others very expensive. It should be expected that the new code will depend less drastically on the performances of the underlying system, since we create juste one thread.
Co-authored-by: John Keiser <john@johnkeiser.com>
Co-authored-by: Daniel Lemire <lemire@gmai.com>
Although it passes user-defined options, if the project is build in Debug mode or with Clang (since
CXX defaults to gcc on Linux) results can flactuate
* This will change the default of the parse benchmark so that it work over hot buffers
by default, thus omitting memory allocation as part of the benchmark.
* Everyone should be using '-H' from now on.
* This is an implementation of "size()" for arrays and objects.
* Adding benchmark
* Adding a size() remark in the documentation.
* Extending size() to result types.
To avoid using data belonging to a temporary, the parse functions are ref qualified to get a compile error if used on an rvalue. See https://github.com/simdjson/simdjson/issues/696
Compilation tests are also added, to make sure bad usage fails to compile.
Reviewed by jkeiser.
* Some users are interested, as a metric, in the number of documents parsed per second.
Obviously, this means reusing the same parser again and again.
* Adding a sentence
* This update the parsingcompetition benchmark so that it displays the number of documents parsed per second.
* Fallback should use our scalar code.
* parse should have a nicer error message.
* Making it so that "minify" can use different architectures.
* Let us change the minifier competition so that it tests all implementations.
* Documenting the untaken optimization opportunity.
Co-authored-by: John Keiser <john@johnkeiser.com>
* If we are going to have a google benchmark flag, we better make sure that we test it out minimal (it should build).
* Fix bench_dom_api
Co-authored-by: John Keiser <john@johnkeiser.com>
* Make architecture implementations virtual functions
- Easier to add new architectures (add implementation to implementation.cpp)
- Easier to add new algorithms / functions to architecture selection
(add to implementation.h, implement)
- Automatically select best implementation in static initialization
- Allow user to explicitly select implementation with a string (i.e.
parameter)
- Allow user to inspect current implementation name/description
- Allow user to list available implementations
- Eliminate architecture enum and architecture-based templating
- Add noexcept in non-inline functions
* Move implementation static methods to their own classes
* Detect best supported implementation on first use
* available_implementationsI() -> available_implementations
This creates a "document" class with only user-facing document state (no parser internals).
- document: user-facing document state
- document::iterator: iterator (equivalent of ParsedJsonIterator)
- document::parser: parser state plus a "docked" document we parse into (equivalent of ParsedJson)
Usage:
```c++
auto doc = simdjson::document::parse(buf, len); // less efficient but simplest
```
```c++
simdjson::document::parser parser; // reusable parser
parser.allocate_capacity(len);
simdjson::document* doc = parser.parse(buf, len); // pointer to doc inside parser
doc = parser.parse(buf2, len); // reuses all buffers and overwrites doc; more efficient
```
* Fix for issue467
* Updating single-header
* Let us make it so that JsonStream is constructed from a padded_string which will avoid dangerous overruns.
* Fixing parse_stream
* Updating documentation.
* This revert the code back to how it was prior to the silly "run two stages" routine and instead
adds an option to benchmark the code over hot buffers. It turns out that it can be expensive,
when the files are large, to allocate the pages.
* Instead of emulating the whole parsing as stage 1 + stage 2, let us
benchmark the real thing.
* Adding explicit constructor.
* Adding warning to the benchmark user.
* Making re-running optional.
* string literal + integer means unintended and incorrect pointer arithmetic
fixes a clang warning. it could not be triggered, because it can only be
triggered if the string given to getopt is not covered among the
cases in the switch.
* handle review comment
* I think we can align the numbers better (so it is prettier).
* Remove space before %, align third line better
Co-authored-by: John Keiser <john@johnkeiser.com>
John Keiser
Paul Dreik
Daniel Lemire
Daniel Lemire
Furkan Usta