Skip to content


Switch branches/tags

Name already in use

A tag already exists with the provided branch name. Many Git commands accept both tag and branch names, so creating this branch may cause unexpected behavior. Are you sure you want to create this branch?


Failed to load latest commit information.
Latest commit message
Commit time
March 6, 2021 01:08
September 5, 2023 21:17
March 5, 2019 18:17
January 16, 2020 00:42
August 20, 2022 17:31
February 12, 2022 13:25
February 10, 2018 14:41
February 23, 2022 09:45
September 18, 2021 14:14
August 20, 2022 17:31

Rapid YAML

MIT Licensed release PyPI Gitter



Or ryml, for short. ryml is a C++ library to parse and emit YAML, and do it fast, on everything from x64 to bare-metal chips without operating system. (If you are looking to use your programs with a YAML tree as a configuration tree with override facilities, take a look at c4conf).

ryml parses both read-only and in-situ source buffers; the resulting data nodes hold only views to sub-ranges of the source buffer. No string copies or duplications are done, and no virtual functions are used. The data tree is a flat index-based structure stored in a single array. Serialization happens only at your direct request, after parsing / before emitting. Internally, the data tree representation stores only string views and has no knowledge of types, but of course, every node can have a YAML type tag. ryml makes it easy and fast to read and modify the data tree.

ryml is available as a single header file, or it can be used as a simple library with cmake -- both separately (ie build->install->find_package()) or together with your project (ie with add_subdirectory()). (See below for examples).

ryml can use custom global and per-tree memory allocators and error handler callbacks, and is exception-agnostic. ryml provides a default implementation for the allocator (using std::malloc()) and error handlers (using using std::abort() is provided, but you can opt out and provide your own memory allocation and eg, exception-throwing callbacks.

ryml does not depend on the STL, ie, it does not use any std container as part of its data structures), but it can serialize and deserialize these containers into the data tree, with the use of optional headers. ryml ships with c4core, a small C++ utilities multiplatform library.

ryml is written in C++11, and compiles cleanly with:

  • Visual Studio 2015 and later
  • clang++ 3.9 and later
  • g++ 4.8 and later
  • Intel Compiler

ryml is extensively unit-tested in Linux, Windows and MacOS. The tests cover x64, x86, wasm (emscripten), arm, aarch64, ppc64le and s390x architectures, and include analysing ryml with:

  • valgrind
  • clang-tidy
  • clang sanitizers:
    • memory
    • address
    • undefined behavior
    • thread

ryml also runs in bare-metal, and RISC-V architectures. Both of these are pending implementation of CI actions for continuous validation, but ryml has been proven to work there.

ryml is available in Python, and can very easily be compiled to JavaScript through emscripten (see below).

See also the changelog and the roadmap.

Table of contents

Is it rapid?

You bet! On a i7-6800K CPU @3.40GHz:

  • ryml parses YAML at about ~150MB/s on Linux and ~100MB/s on Windows (vs2017).
  • ryml parses JSON at about ~450MB/s on Linux, faster than sajson (didn't try yet on Windows).
  • compared against the other existing YAML libraries for C/C++:
    • ryml is in general between 2 and 3 times faster than libyaml
    • ryml is in general between 10 and 70 times faster than yaml-cpp, and in some cases as much as 100x and even 200x faster.

Here's the benchmark. Using different approaches within ryml (in-situ/read-only vs. with/without reuse), a YAML / JSON buffer is repeatedly parsed, and compared against other libraries.

Comparison with yaml-cpp

The first result set is for Windows, and is using a appveyor.yml config file. A comparison of these results is summarized on the table below:

Read rates (MB/s) ryml yamlcpp compared
appveyor / vs2017 / Release 101.5 5.3 20x / 5.2%
appveyor / vs2017 / Debug 6.4 0.0844 76x / 1.3%

The next set of results is taken in Linux, comparing g++ 8.2 and clang++ 7.0.1 in parsing a YAML buffer from a travis.yml config file or a JSON buffer from a compile_commands.json file. You can see the full results here. Summarizing:

Read rates (MB/s) ryml yamlcpp compared
json / clang++ / Release 453.5 15.1 30x / 3%
json / g++ / Release 430.5 16.3 26x / 4%
json / clang++ / Debug 61.9 1.63 38x / 3%
json / g++ / Debug 72.6 1.53 47x / 2%
travis / clang++ / Release 131.6 8.08 16x / 6%
travis / g++ / Release 176.4 8.23 21x / 5%
travis / clang++ / Debug 10.2 1.08 9x / 1%
travis / g++ / Debug 12.5 1.01 12x / 8%

The 450MB/s read rate for JSON puts ryml squarely in the same ballpark as RapidJSON and other fast json readers (data from here). Even parsing full YAML is at ~150MB/s, which is still in that performance ballpark, albeit at its lower end. This is something to be proud of, as the YAML specification is much more complex than JSON: 23449 vs 1969 words.

Performance reading JSON

So how does ryml compare against other JSON readers? Well, it's one of the fastest!

The benchmark is the same as above, and it is reading the compile_commands.json, The _arena suffix notes parsing a read-only buffer (so buffer copies are performed), while the _inplace suffix means that the source buffer can be parsed in place. The _reuse means the data tree and/or parser are reused on each benchmark repeat.

Here's what we get with g++ 8.2:

Benchmark Release,MB/s Debug,MB/s
rapidjson_arena 509.9 43.4
rapidjson_inplace 1329.4 68.2
sajson_inplace 434.2 176.5
sajson_arena 430.7 175.6
jsoncpp_arena 183.6 ? 187.9
nlohmann_json_arena 115.8 21.5
yamlcpp_arena 16.6 1.6
libyaml_arena 113.9 35.7
libyaml_arena_reuse 114.6 35.9
ryml_arena 388.6 36.9
ryml_inplace 393.7 36.9
ryml_arena_reuse 446.2 74.6
ryml_inplace_reuse 457.1 74.9

You can verify that (at least for this test) ryml beats most json parsers at their own game, with the only exception of rapidjson. And actually, in Debug, rapidjson is slower than ryml, and sajson manages to be faster (but not sure about jsoncpp; need to scrutinize there the suspicious fact that the Debug result is faster than the Release result).

Performance emitting

Emitting benchmarks also show similar speedups from the existing libraries, also anecdotally reported by some users (eg, here's a user reporting 25x speedup from yaml-cpp). Also, in some cases (eg, block folded multiline scalars), the speedup is as high as 200x (eg, 7.3MB/s -> 1.416MG/s).

CI results and request for files

While a more effective way of showing the benchmark results is not available yet, you can browse through the runs of the benchmark workflow in the CI to scroll through the results for yourself.

Also, if you have a case where ryml behaves very nicely or not as nicely as claimed above, we would definitely like to see it! Please submit a pull request adding the file to bm/cases, or just send us the files.

Quick start

If you're wondering whether ryml's speed comes at a usage cost, you need not: with ryml, you can have your cake and eat it too. Being rapid is definitely NOT the same as being unpractical, so ryml was written with easy AND efficient usage in mind, and comes with a two level API for accessing and traversing the data tree.

The following snippet is a quick overview taken from the quickstart sample. After cloning ryml (don't forget the --recursive flag for git), you can very easily build and run this executable using any of the build samples, eg the add_subdirectory() sample.

// Parse YAML code in place, potentially mutating the buffer.
// It is also possible to:
//   - parse a read-only buffer using parse_in_arena()
//   - reuse an existing tree (advised)
//   - reuse an existing parser (advised)
char yml_buf[] = "{foo: 1, bar: [2, 3], john: doe}";
ryml::Tree tree = ryml::parse_in_place(yml_buf);

// Note: it will always be significantly faster to use mutable
// buffers and reuse tree+parser.
// Below you will find samples that show how to achieve reuse; but
// please note that for brevity and clarity, many of the examples
// here are parsing immutable buffers, and not reusing tree or
// parser.

// API overview

// ryml has a two-level API:
// The lower level index API is based on the indices of nodes,
// where the node's id is the node's position in the tree's data
// array. This API is very efficient, but somewhat difficult to use:
size_t root_id = tree.root_id();
size_t bar_id = tree.find_child(root_id, "bar"); // need to get the index right
CHECK(tree.is_map(root_id)); // all of the index methods are in the tree
CHECK(tree.is_seq(bar_id));  // ... and receive the subject index

// The node API is a lightweight abstraction sitting on top of the
// index API, but offering a much more convenient interaction:
ryml::ConstNodeRef root = tree.rootref();
ryml::ConstNodeRef bar = tree["bar"];
// A node ref is a lightweight handle to the tree and associated id:
CHECK(root.tree() == &tree); // a node ref points at its tree, WITHOUT refcount
CHECK( == root_id); // a node ref's id is the index of the node
CHECK( == bar_id);   // a node ref's id is the index of the node

// The node API translates very cleanly to the index API, so most
// of the code examples below are using the node API.

// One significant point of the node API is that it holds a raw
// pointer to the tree. Care must be taken to ensure the lifetimes
// match, so that a node will never access the tree after the tree
// went out of scope.

// To read the parsed tree

// ConstNodeRef::operator[] does a lookup, is O(num_children[node]).
CHECK(tree["foo"].key() == "foo");
CHECK(tree["foo"].val() == "1");
CHECK(tree["bar"].key() == "bar");
// maps use string keys, seqs use integral keys:
CHECK(tree["bar"][0].val() == "2");
CHECK(tree["bar"][1].val() == "3");
CHECK(tree["john"].val() == "doe");
// An integral key is the position of the child within its parent,
// so even maps can also use int keys, if the key position is
// known.
CHECK(tree[0].id() == tree["foo"].id());
CHECK(tree[1].id() == tree["bar"].id());
CHECK(tree[2].id() == tree["john"].id());
// Tree::operator[](int) searches a ***root*** child by its position.
CHECK(tree[0].id() == tree["foo"].id());  // 0: first child of root
CHECK(tree[1].id() == tree["bar"].id());  // 1: first child of root
CHECK(tree[2].id() == tree["john"].id()); // 2: first child of root
// NodeRef::operator[](int) searches a ***node*** child by its position:
CHECK(bar[0].val() == "2"); // 0 means first child of bar
CHECK(bar[1].val() == "3"); // 1 means second child of bar
// NodeRef::operator[](string):
// A string key is the key of the node: lookup is by name. So it
// is only available for maps, and it is NOT available for seqs,
// since seq members do not have keys.
CHECK(tree["foo"].key() == "foo");
CHECK(tree["bar"].key() == "bar");
CHECK(tree["john"].key() == "john");
// CHECK(bar["BOOM!"].is_seed()); // error, seqs do not have key lookup

// Note that maps can also use index keys as well as string keys:
CHECK(root["foo"].id() == root[0].id());
CHECK(root["bar"].id() == root[1].id());
CHECK(root["john"].id() == root[2].id());

// IMPORTANT. The ryml tree uses indexed linked lists for storing
// children, so the complexity of `Tree::operator[csubstr]` and
// `Tree::operator[size_t]` is linear on the number of root
// children. If you use `Tree::operator[]` with a large tree where
// the root has many children, you will see a performance hit.
// To avoid this hit, you can create your own accelerator
// structure. For example, before doing a lookup, do a single
// traverse at the root level to fill an `map<csubstr,size_t>`
// mapping key names to node indices; with a node index, a lookup
// (via `Tree::get()`) is O(1), so this way you can get O(log n)
// lookup from a key. (But please do not use `std::map` if you
// care about performance; use something else like a flat map or
// sorted vector).
// As for node refs, the difference from `NodeRef::operator[]` and
// `ConstNodeRef::operator[]` to `Tree::operator[]` is that the
// latter refers to the root node, whereas the former are invoked
// on their target node. But the lookup process works the same for
// both and their algorithmic complexity is the same: they are
// both linear in the number of direct children. But of course,
// depending on the data, that number may be very different from
// one to another.

// Hierarchy:

    ryml::ConstNodeRef foo = root.first_child();
    ryml::ConstNodeRef john = root.last_child();
    CHECK(tree.size() == 6); // O(1) number of nodes in the tree
    CHECK(root.num_children() == 3); // O(num_children[root])
    CHECK(foo.num_siblings() == 3); // O(num_children[parent(foo)])
    CHECK(foo.parent().id() ==; // parent() is O(1)
    CHECK(root.first_child().id() == root["foo"].id()); // first_child() is O(1)
    CHECK(root.last_child().id() == root["john"].id()); // last_child() is O(1)
    CHECK(john.first_sibling().id() ==;
    CHECK(foo.last_sibling().id() ==;
    // prev_sibling(), next_sibling(): (both are O(1))
    CHECK(foo.num_siblings() == root.num_children());
    CHECK(foo.prev_sibling().id() == ryml::NONE); // foo is the first_child()
    CHECK(foo.next_sibling().key() == "bar");
    CHECK(foo.next_sibling().next_sibling().key() == "john");
    CHECK(foo.next_sibling().next_sibling().next_sibling().id() == ryml::NONE); // john is the last_child()

// Iterating:
    ryml::csubstr expected_keys[] = {"foo", "bar", "john"};
    // iterate children using the high-level node API:
        size_t count = 0;
        for(ryml::ConstNodeRef const& child : root.children())
            CHECK(child.key() == expected_keys[count++]);
    // iterate siblings using the high-level node API:
        size_t count = 0;
        for(ryml::ConstNodeRef const& child : root["foo"].siblings())
            CHECK(child.key() == expected_keys[count++]);
    // iterate children using the lower-level tree index API:
        size_t count = 0;
        for(size_t child_id = tree.first_child(root_id); child_id != ryml::NONE; child_id = tree.next_sibling(child_id))
            CHECK(tree.key(child_id) == expected_keys[count++]);
    // iterate siblings using the lower-level tree index API:
    // (notice the only difference from above is in the loop
    // preamble, which calls tree.first_sibling(bar_id) instead of
    // tree.first_child(root_id))
        size_t count = 0;
        for(size_t child_id = tree.first_sibling(bar_id); child_id != ryml::NONE; child_id = tree.next_sibling(child_id))
            CHECK(tree.key(child_id) == expected_keys[count++]);

// Gotchas:
CHECK(!tree["bar"].has_val());          // seq is a container, so no val
CHECK(!tree["bar"][0].has_key());       // belongs to a seq, so no key
CHECK(!tree["bar"][1].has_key());       // belongs to a seq, so no key
//CHECK(tree["bar"].val() == BOOM!);    // ... so attempting to get a val is undefined behavior
//CHECK(tree["bar"][0].key() == BOOM!); // ... so attempting to get a key is undefined behavior
//CHECK(tree["bar"][1].key() == BOOM!); // ... so attempting to get a key is undefined behavior

// Deserializing: use operator>>
    int foo = 0, bar0 = 0, bar1 = 0;
    std::string john;
    root["foo"] >> foo;
    root["bar"][0] >> bar0;
    root["bar"][1] >> bar1;
    root["john"] >> john; // requires from_chars(std::string). see serialization samples below.
    CHECK(foo == 1);
    CHECK(bar0 == 2);
    CHECK(bar1 == 3);
    CHECK(john == "doe");

// Modifying existing nodes: operator<< vs operator=

// As implied by its name, ConstNodeRef is a reference to a const
// node. It can be used to read from the node, but not write to it
// or modify the hierarchy of the node. If any modification is
// desired then a NodeRef must be used instead:
ryml::NodeRef wroot = tree.rootref();

// operator= assigns an existing string to the receiving node.
// This pointer will be in effect until the tree goes out of scope
// so beware to only assign from strings outliving the tree.
wroot["foo"] = "says you";
wroot["bar"][0] = "-2";
wroot["bar"][1] = "-3";
wroot["john"] = "ron";
// Now the tree is _pointing_ at the memory of the strings above.
// That is OK because those are static strings and will outlive
// the tree.
CHECK(root["foo"].val() == "says you");
CHECK(root["bar"][0].val() == "-2");
CHECK(root["bar"][1].val() == "-3");
CHECK(root["john"].val() == "ron");
// WATCHOUT: do not assign from temporary objects:
// {
//     std::string crash("will dangle");
//     root["john"] = ryml::to_csubstr(crash);
// }
// CHECK(root["john"] == "dangling"); // CRASH! the string was deallocated

// operator<< first serializes the input to the tree's arena, then
// assigns the serialized string to the receiving node. This avoids
// constraints with the lifetime, since the arena lives with the tree.
wroot["foo"] << "says who";  // requires to_chars(). see serialization samples below.
wroot["bar"][0] << 20;
wroot["bar"][1] << 30;
wroot["john"] << "deere";
CHECK(root["foo"].val() == "says who");
CHECK(root["bar"][0].val() == "20");
CHECK(root["bar"][1].val() == "30");
CHECK(root["john"].val() == "deere");
CHECK(tree.arena() == "says who2030deere"); // the result of serializations to the tree arena
// using operator<< instead of operator=, the crash above is avoided:
    std::string ok("in_scope");
    // root["john"] = ryml::to_csubstr(ok); // don't, will dangle
    wroot["john"] << ryml::to_csubstr(ok); // OK, copy to the tree's arena
CHECK(root["john"] == "in_scope"); // OK!
CHECK(tree.arena() == "says who2030deerein_scope"); // the result of serializations to the tree arena

// Adding new nodes:

// adding a keyval node to a map:
CHECK(root.num_children() == 3);
wroot["newkeyval"] = "shiny and new"; // using these strings
wroot.append_child() << ryml::key("newkeyval (serialized)") << "shiny and new (serialized)"; // serializes and assigns the serialization
CHECK(root.num_children() == 5);
CHECK(root["newkeyval"].key() == "newkeyval");
CHECK(root["newkeyval"].val() == "shiny and new");
CHECK(root["newkeyval (serialized)"].key() == "newkeyval (serialized)");
CHECK(root["newkeyval (serialized)"].val() == "shiny and new (serialized)");
CHECK( ! tree.in_arena(root["newkeyval"].key())); // it's using directly the static string above
CHECK( ! tree.in_arena(root["newkeyval"].val())); // it's using directly the static string above
CHECK(   tree.in_arena(root["newkeyval (serialized)"].key())); // it's using a serialization of the string above
CHECK(   tree.in_arena(root["newkeyval (serialized)"].val())); // it's using a serialization of the string above
// adding a val node to a seq:
CHECK(root["bar"].num_children() == 2);
wroot["bar"][2] = "oh so nice";
wroot["bar"][3] << "oh so nice (serialized)";
CHECK(root["bar"].num_children() == 4);
CHECK(root["bar"][2].val() == "oh so nice");
CHECK(root["bar"][3].val() == "oh so nice (serialized)");
// adding a seq node:
CHECK(root.num_children() == 5);
wroot["newseq"] |= ryml::SEQ;
wroot.append_child() << ryml::key("newseq (serialized)") |= ryml::SEQ;
CHECK(root.num_children() == 7);
CHECK(root["newseq"].num_children() == 0);
CHECK(root["newseq (serialized)"].num_children() == 0);
// adding a map node:
CHECK(root.num_children() == 7);
wroot["newmap"] |= ryml::MAP;
wroot.append_child() << ryml::key("newmap (serialized)") |= ryml::SEQ;
CHECK(root.num_children() == 9);
CHECK(root["newmap"].num_children() == 0);
CHECK(root["newmap (serialized)"].num_children() == 0);
// When the tree is mutable, operator[] does not mutate the tree
// until the returned node is written to.
// Until such time, the NodeRef object keeps in itself the required
// information to write to the proper place in the tree. This is
// called being in a "seed" state.
// This means that passing a key/index which does not exist will
// not mutate the tree, but will instead store (in the node) the
// proper place of the tree to be able to do so, if and when it is
// required.
// This is a significant difference from eg, the behavior of
// std::map, which mutates the map immediately within the call to
// operator[].
// All of the points above apply only if the tree is mutable. If
// the tree is const, then a NodeRef cannot be obtained from it;
// only a ConstNodeRef, which can never be used to mutate the
// tree.
CHECK(!root.has_child("I am not nothing"));
ryml::NodeRef nothing = wroot["I am nothing"];
CHECK(nothing.valid());   // points at the tree, and a specific place in the tree
CHECK(nothing.is_seed()); // ... but nothing is there yet.
CHECK(!root.has_child("I am nothing")); // same as above
ryml::NodeRef something = wroot["I am something"];
ryml::ConstNodeRef constsomething = wroot["I am something"];
CHECK(!root.has_child("I am something")); // same as above
CHECK(something.is_seed()); // same as above
CHECK(!constsomething.valid()); // NOTE: because a ConstNodeRef
                                // cannot be used to mutate a
                                // tree, it is only valid() if it
                                // is pointing at an existing
                                // node.
something = "indeed";  // this will commit to the tree, mutating at the proper place
CHECK(root.has_child("I am something"));
CHECK(root["I am something"].val() == "indeed");
CHECK(!something.is_seed()); // now the tree has this node, so the
                             // ref is no longer a seed
// now the constref is also valid (but it needs to be reassigned):
ryml::ConstNodeRef constsomethingnew = wroot["I am something"];
// note that the old constref is now stale, because it only keeps
// the state at creation:

// Emitting:

// emit to a FILE*
ryml::emit_yaml(tree, stdout); // there is also emit_json()
// emit to a stream
std::stringstream ss;
ss << tree;
std::string stream_result = ss.str();
// emit to a buffer:
std::string str_result = ryml::emitrs_yaml<std::string>(tree); // there is also emitrs_json()
// can emit to any given buffer:
char buf[1024];
ryml::csubstr buf_result = ryml::emit_yaml(tree, buf);
// now check
ryml::csubstr expected_result = R"(foo: says who
- 20
- 30
- oh so nice
- oh so nice (serialized)
john: in_scope
newkeyval: shiny and new
newkeyval (serialized): shiny and new (serialized)
newseq: []
newseq (serialized): []
newmap: {}
newmap (serialized): []
I am something: indeed
CHECK(buf_result == expected_result);
CHECK(str_result == expected_result);
CHECK(stream_result == expected_result);
// There are many possibilities to emit to buffer;
// please look at the emit sample functions below.

// ConstNodeRef vs NodeRef

ryml::NodeRef noderef = tree["bar"][0];
ryml::ConstNodeRef constnoderef = tree["bar"][0];

// ConstNodeRef cannot be used to mutate the tree, but a NodeRef can:
//constnoderef = "21";  // compile error
//constnoderef << "22"; // compile error
noderef = "21";         // ok, can assign because it's not const
CHECK(tree["bar"][0].val() == "21");
noderef << "22";        // ok, can serialize and assign because it's not const
CHECK(tree["bar"][0].val() == "22");

// it is not possible to obtain a NodeRef from a ConstNodeRef:
// noderef = constnoderef; // compile error

// it is always possible to obtain a ConstNodeRef from a NodeRef:
constnoderef = noderef;    // ok can assign const <- nonconst

// If a tree is const, then only ConstNodeRef's can be
// obtained from that tree:
ryml::Tree const& consttree = tree;
//noderef = consttree["bar"][0];    // compile error
noderef = tree["bar"][0];           // ok
constnoderef = consttree["bar"][0]; // ok

// ConstNodeRef and NodeRef can be compared for equality.
// Equality means they point at the same node.
CHECK(constnoderef == noderef);
CHECK(!(constnoderef != noderef));

// Dealing with UTF8
ryml::Tree langs = ryml::parse_in_arena(R"(
en: Planet (Gas)
fr: Planète (Gazeuse)
ru: Планета (Газ)
ja: 惑星(ガス)
zh: 行星(气体)
# UTF8 decoding only happens in double-quoted strings,\
# as per the YAML standard
decode this: "\u263A \xE2\x98\xBA"
and this as well: "\u2705 \U0001D11E"
// in-place UTF8 just works:
CHECK(langs["en"].val() == "Planet (Gas)");
CHECK(langs["fr"].val() == "Planète (Gazeuse)");
CHECK(langs["ru"].val() == "Планета (Газ)");
CHECK(langs["ja"].val() == "惑星(ガス)");
CHECK(langs["zh"].val() == "行星(气体)");
// and \x \u \U codepoints are decoded (but only when they appear
// inside double-quoted strings, as dictated by the YAML
// standard):
CHECK(langs["decode this"].val() == "☺ ☺");
CHECK(langs["and this as well"].val() == "✅ 𝄞");

// Getting the location of nodes in the source:
ryml::Parser parser;
ryml::Tree tree2 = parser.parse_in_arena("expected.yml", expected_result);
ryml::Location loc = parser.location(tree2["bar"][1]);
CHECK(loc.line == 3u);
CHECK(loc.col == 4u);

The quickstart.cpp sample (from which the above overview was taken) has many more detailed examples, and should be your first port of call to find out any particular point about ryml's API. It is tested in the CI, and thus has the correct behavior. There you can find the following subjects being addressed:

sample_substr();               ///< about ryml's string views (from c4core)
sample_parse_file();           ///< ready-to-go example of parsing a file from disk
sample_parse_in_place();       ///< parse a mutable YAML source buffer
sample_parse_in_arena();       ///< parse a read-only YAML source buffer
sample_parse_reuse_tree();     ///< parse into an existing tree, maybe into a node
sample_parse_reuse_parser();   ///< reuse an existing parser
sample_parse_reuse_tree_and_parser(); ///< how to reuse existing trees and parsers
sample_iterate_trees();        ///< visit individual nodes and iterate through trees
sample_create_trees();         ///< programatically create trees
sample_tree_arena();           ///< interact with the tree's serialization arena
sample_fundamental_types();    ///< serialize/deserialize fundamental types
sample_formatting();           ///< control formatting when serializing/deserializing
sample_base64();               ///< encode/decode base64
sample_user_scalar_types();    ///< serialize/deserialize scalar (leaf/string) types
sample_user_container_types(); ///< serialize/deserialize container (map or seq) types
sample_std_types();            ///< serialize/deserialize STL containers
sample_emit_to_container();    ///< emit to memory, eg a string or vector-like container
sample_emit_to_stream();       ///< emit to a stream, eg std::ostream
sample_emit_to_file();         ///< emit to a FILE*
sample_emit_nested_node();     ///< pick a nested node as the root when emitting
sample_json();                 ///< JSON parsing and emitting
sample_anchors_and_aliases();  ///< deal with YAML anchors and aliases
sample_tags();                 ///< deal with YAML type tags
sample_docs();                 ///< deal with YAML docs
sample_error_handler();        ///< set a custom error handler
sample_global_allocator();     ///< set a global allocator for ryml
sample_per_tree_allocator();   ///< set per-tree allocators
sample_static_trees();         ///< how to use static trees in ryml
sample_location_tracking();    ///< track node locations in the parsed source tree

Using ryml in your project

Package managers

If you opt for package managers, here's where ryml is available so far (thanks to all the contributors!):

Although package managers are very useful for quickly getting up to speed, the advised way is still to bring ryml as a submodule of your project, building both together. This makes it easy to track any upstream changes in ryml. Also, ryml is small and quick to build, so there's not much of a cost for building it with your project.

Single header file

ryml is provided chiefly as a cmake library project, but it can also be used as a single header file, and there is a tool to amalgamate the code into a single header file. The amalgamated header file is provided with each release, but you can also generate a customized file suiting your particular needs (or commit):

[user@host rapidyaml]$ python3 tools/ -h
usage: [-h] [--c4core | --no-c4core] [--fastfloat | --no-fastfloat] [--stl | --no-stl] [output]

positional arguments:
  output          output file. defaults to stdout

optional arguments:
  -h, --help      show this help message and exit
  --c4core        amalgamate c4core together with ryml. this is the default.
  --no-c4core     amalgamate c4core together with ryml. the default is --c4core.
  --fastfloat     enable fastfloat library. this is the default.
  --no-fastfloat  enable fastfloat library. the default is --fastfloat.
  --stl           enable stl interop. this is the default.
  --no-stl        enable stl interop. the default is --stl.

The amalgamated header file contains all the function declarations and definitions. To use it in the project, #include the header at will in any header or source file in the project, but in one source file, and only in that one source file, #define the macro RYML_SINGLE_HDR_DEFINE_NOW before including the header. This will enable the function definitions. For example:

// foo.h
#include <ryml_all.hpp>

// foo.cpp
// ensure that foo.h is not included before this define!
#include <ryml_all.hpp>

If you wish to package the single header into a shared library, then you will need to define the preprocessor symbol RYML_SHARED during compilation.

As a library

The single header file is a good approach to quickly try the library, but if you wish to make good use of CMake and its tooling ecosystem, (and get better compile times), then ryml has you covered.

As with any other cmake library, you have the option to integrate ryml into your project's build setup, thereby building ryml together with your project, or -- prior to configuring your project -- you can have ryml installed either manually or through package managers.

Currently cmake is required to build ryml; we recommend a recent cmake version, at least 3.13.

Note that ryml uses submodules. Take care to use the --recursive flag when cloning the repo, to ensure ryml's submodules are checked out as well:

git clone --recursive

If you omit --recursive, after cloning you will have to do git submodule update --init --recursive to ensure ryml's submodules are checked out.

Quickstart samples

These samples show different ways of getting ryml into your application. All the samples use the same quickstart executable source, but are built in different ways, showing several alternatives to integrate ryml into your project. We also encourage you to refer to the quickstart source itself, which extensively covers most of the functionality that you may want out of ryml.

Each sample brings a script with the sequence of commands required to successfully build and run the application (this is a bash script and runs in Linux and MacOS, but it is also possible to run in Windows via Git Bash or the WSL). Click on the links below to find out more about each sample:

Sample name ryml is part of build? cmake file commands
singleheader yes
ryml brought as a single header file,
not as a library
singleheaderlib yes
ryml brought as a library
but from the single header file
CMakeLists.txt (shared library) (static library)
add_subdirectory yes CMakeLists.txt
fetch_content yes CMakeLists.txt
find_package no
needs prior install or package

CMake build settings for ryml

The following cmake variables can be used to control the build behavior of ryml:

  • RYML_WITH_TAB_TOKENS=ON/OFF. Enable/disable support for tabs as valid container tokens after : and -. Defaults to OFF, because this may cost up to 10% in processing time.
  • RYML_DEFAULT_CALLBACKS=ON/OFF. Enable/disable ryml's default implementation of error and allocation callbacks. Defaults to ON.
  • RYML_STANDALONE=ON/OFF. ryml uses c4core, a C++ library with low-level multi-platform utilities for C++. When RYML_STANDALONE=ON, c4core is incorporated into ryml as if it is the same library. Defaults to ON.

If you're developing ryml or just debugging problems with ryml itself, the following cmake variables can be helpful:

  • RYML_DEV=ON/OFF: a bool variable which enables development targets such as unit tests, benchmarks, etc. Defaults to OFF.
  • RYML_DBG=ON/OFF: a bool variable which enables verbose prints from parsing code; can be useful to figure out parsing problems. Defaults to OFF.

Forcing ryml to use a different c4core version

ryml is strongly coupled to c4core, and this is reinforced by the fact that c4core is a submodule of the current repo. However, it is still possible to use a c4core version different from the one in the repo (of course, only if there are no incompatibilities between the versions). You can find out how to achieve this by looking at the custom_c4core sample.

Other languages

One of the aims of ryml is to provide an efficient YAML API for other languages. JavaScript is fully available, and there is already a cursory implementation for Python using only the low-level API. After ironing out the general approach, other languages are likely to follow (all of this is possible because we're using SWIG, which makes it easy to do so).


A JavaScript+WebAssembly port is available, compiled through emscripten.


(Note that this is a work in progress. Additions will be made and things will be changed.) With that said, here's an example of the Python API:

import ryml

# ryml cannot accept strings because it does not take ownership of the
# source buffer; only bytes or bytearrays are accepted.
src = b"{HELLO: a, foo: b, bar: c, baz: d, seq: [0, 1, 2, 3]}"

def check(tree):
    # for now, only the index-based low-level API is implemented
    assert tree.size() == 10
    assert tree.root_id() == 0
    assert tree.first_child(0) == 1
    assert tree.next_sibling(1) == 2
    assert tree.first_sibling(5) == 2
    assert tree.last_sibling(1) == 5
    # use bytes objects for queries
    assert tree.find_child(0, b"foo") == 1
    assert tree.key(1) == b"foo")
    assert tree.val(1) == b"b")
    assert tree.find_child(0, b"seq") == 5
    assert tree.is_seq(5)
    # to loop over children:
    for i, ch in enumerate(ryml.children(tree, 5)):
        assert tree.val(ch) == [b"0", b"1", b"2", b"3"][i]
    # to loop over siblings:
    for i, sib in enumerate(ryml.siblings(tree, 5)):
        assert tree.key(sib) == [b"HELLO", b"foo", b"bar", b"baz", b"seq"][i]
    # to walk over all elements
    visited = [False] * tree.size()
    for n, indentation_level in ryml.walk(tree):
        # just a dumb emitter
        left = "  " * indentation_level
        if tree.is_keyval(n):
           print("{}{}: {}".format(left, tree.key(n), tree.val(n))
        elif tree.is_val(n):
           print("- {}".format(left, tree.val(n))
        elif tree.is_keyseq(n):
           print("{}{}:".format(left, tree.key(n))
        visited[inode] = True
    assert False not in visited
    # NOTE about encoding!
    k = tree.get_key(5)
    print(k)  # '<memory at 0x7f80d5b93f48>'
    assert k == b"seq"               # ok, as expected
    assert k != "seq"                # not ok - NOTE THIS! 
    assert str(k) != "seq"           # not ok
    assert str(k, "utf8") == "seq"   # ok again

# parse immutable buffer
tree = ryml.parse_in_arena(src)
check(tree) # OK

# parse mutable buffer.
# requires bytearrays or objects offering writeable memory
mutable = bytearray(src)
tree = ryml.parse_in_place(mutable)
check(tree) # OK

As expected, the performance results so far are encouraging. In a timeit benchmark compared against PyYaml and ruamel.yaml, ryml parses quicker by generally 100x and up to 400x:

| style_seqs_blck_outer1000_inner100.yml | count | time(ms) | avg(ms)  | avg(MB/s) |
| parse:RuamelYamlParse                  |     1 | 4564.812 | 4564.812 |     0.173 |
| parse:PyYamlParse                      |     1 | 2815.426 | 2815.426 |     0.280 |
| parse:RymlParseInArena                 |    38 |  588.024 |   15.474 |    50.988 |
| parse:RymlParseInArenaReuse            |    38 |  466.997 |   12.289 |    64.202 |
| parse:RymlParseInPlace                 |    38 |  579.770 |   15.257 |    51.714 |
| parse:RymlParseInPlaceReuse            |    38 |  462.932 |   12.182 |    64.765 |

(Note that the parse timings above are somewhat biased towards ryml, because it does not perform any type conversions in Python-land: return types are merely memoryviews to the source buffer, possibly copied to the tree's arena).

As for emitting, the improvement can be as high as 3000x:

| style_maps_blck_outer1000_inner100.yml | count |  time(ms) |  avg(ms)  | avg(MB/s) |
| emit_yaml:RuamelYamlEmit               |     1 | 18149.288 | 18149.288 |     0.054 |
| emit_yaml:PyYamlEmit                   |     1 |  2683.380 |  2683.380 |     0.365 |
| emit_yaml:RymlEmitToNewBuffer          |    88 |   861.726 |     9.792 |    99.976 |
| emit_yaml:RymlEmitReuse                |    88 |   437.931 |     4.976 |   196.725 |

YAML standard conformance

ryml is close to feature complete. Most of the YAML features are well covered in the unit tests, and expected to work, unless in the exceptions noted below.

Of course, there are many dark corners in YAML, and there certainly can appear cases which ryml fails to parse. Your bug reports or pull requests are very welcome.

See also the roadmap for a list of future work.

Known limitations

ryml deliberately makes no effort to follow the standard in the following situations:

  • Containers are not accepted as mapping keys: keys must be scalars.
  • Tab characters after : and - are not accepted tokens, unless ryml is compiled with the macro RYML_WITH_TAB_TOKENS. This requirement exists because checking for tabs introduces branching into the parser's hot code and in some cases costs as much as 10% in parsing time.
  • Anchor names must not end with a terminating colon: eg &anchor: key: val.
  • Non-unique map keys are allowed. Enforcing key uniqueness in the parser or in the tree would cause log-linear parsing complexity (for root children on a mostly flat tree), and would increase code size through added structural, logical and cyclomatic complexity. So enforcing uniqueness in the parser would hurt users who may not care about it (they may not care either because non-uniqueness is OK for their use case, or because it is impossible to occur). On the other hand, any user who requires uniqueness can easily enforce it by doing a post-parse walk through the tree. So choosing to not enforce key uniqueness adheres to the spirit of "don't pay for what you don't use".
  • %YAML directives have no effect and are ignored.
  • %TAG directives are limited to a default maximum of 4 instances per Tree. To increase this maximum, define the preprocessor symbol RYML_MAX_TAG_DIRECTIVES to a suitable value. This arbitrary limit reflects the usual practice of having at most 1 or 2 tag directives; also, be aware that this feature is under consideration for removal in YAML 1.3.

Also, ryml tends to be on the permissive side where the YAML standard dictates there should be an error; in many of these cases, ryml will tolerate the input. This may be good or bad, but in any case is being improved on (meaning ryml will grow progressively less tolerant of YAML errors in the coming releases). So we strongly suggest to stay away from those dark corners of YAML which are generally a source of problems, which is a good practice anyway.

If you do run into trouble and would like to investigate conformance of your YAML code, beware of existing online YAML linters, many of which are not fully conformant; instead, try using, an amazing tool which lets you dynamically input your YAML and continuously see the results from all the existing parsers (kudos to @ingydotnet and the people from the YAML test suite). And of course, if you detect anything wrong with ryml, please open an issue so that we can improve.

Test suite status

As part of its CI testing, ryml uses the YAML test suite. This is an extensive set of reference cases covering the full YAML spec. Each of these cases have several subparts:

  • in-yaml: mildly, plainly or extremely difficult-to-parse YAML
  • in-json: equivalent JSON (where possible/meaningful)
  • out-yaml: equivalent standard YAML
  • emit-yaml: equivalent standard YAML
  • events: reference results (ie, expected tree)

When testing, ryml parses each of the 4 yaml/json parts, then emits the parsed tree, then parses the emitted result and verifies that emission is idempotent, ie that the emitted result is semantically the same as its input without any loss of information. To ensure consistency, this happens over four levels of parse/emission pairs. And to ensure correctness, each of the stages is compared against the events spec from the test, which constitutes the reference. The tests also check for equality between the reference events in the test case and the events emitted by ryml from the data tree parsed from the test case input. All of this is then carried out combining several variations: both unix \n vs windows \r\n line endings, emitting to string, file or streams, which results in ~250 tests per case part. With multiple parts per case and ~400 reference cases in the test suite, this makes over several hundred thousand individual tests to which ryml is subjected, which are added to the unit tests in ryml, which also employ the same extensive combinatorial approach.

Also, note that in their own words, the tests from the YAML test suite contain a lot of edge cases that don't play such an important role in real world examples. And yet, despite the extreme focus of the test suite, currently ryml only fails a minor fraction of the test cases, mostly related with the deliberate limitations noted above. Other than those limitations, by far the main issue with ryml is that several standard-mandated parse errors fail to materialize. For the up-to-date list of ryml failures in the test-suite, refer to the list of known exceptions from ryml's test suite runner, which is used as part of ryml's CI process.

Alternative libraries

Why this library? Because none of the existing libraries was quite what I wanted. When I started this project in 2018, I was aware of these two alternative C/C++ libraries:

  • libyaml. This is a bare C library. It does not create a representation of the data tree, so I don't see it as practical. My initial idea was to wrap parsing and emitting around libyaml's convenient event handling, but to my surprise I found out it makes heavy use of allocations and string duplications when parsing. I briefly pondered on sending PRs to reduce these allocation needs, but not having a permanent tree to store the parsed data was too much of a downside.
  • yaml-cpp. This library may be full of functionality, but is heavy on the use of node-pointer-based structures like std::map, allocations, string copies, polymorphism and slow C++ stream serializations. This is generally a sure way of making your code slower, and strong evidence of this can be seen in the benchmark results above.

Recently libfyaml appeared. This is a newer C library, fully conformant to the YAML standard with an amazing 100% success in the test suite; it also offers the tree as a data structure. As a downside, it does not work in Windows, and it is also multiple times slower parsing and emitting.

When performance and low latency are important, using contiguous structures for better cache behavior and to prevent the library from trampling caches, parsing in place and using non-owning strings is of central importance. Hence this Rapid YAML library which, with minimal compromise, bridges the gap from efficiency to usability. This library takes inspiration from RapidJSON and RapidXML.


ryml is permissively licensed under the MIT license.