New API is as follows:
* `clixon_xml2file()` - Print internal tree as XML to file
* `clixon_xml2cbuf()` - Print internal tree as XML to buffer
* `clixon_json2file()` - Print internal tree as JSON to file
* `clixon_json2cbuf()` - Print internal tree as JSON to buffer
* `clixon_cli2file()` - Print internal tree as CLI format to file
* `clixon_txt2file()` - Print internal tree as text format to file
* Added `skiptop` parameter, if set only apply to children of a node, skip top node
* default is 0
* Functions are merged, ie removed and replaced with more generic functions
* `xml2json_cbuf()`: Added `skiptop` parameter: `xml2json_cbuf(..., int skiptop)`
* `xml2json()` and `xml2json_cb()` merged into `xml2json_file()` with `skiptop`
* Replace `xml2json(...)` with `xml2json_file(..., stdout, 0)`
* Replace `xml2json_cb(...)` with `xml2json_file(..., 0)`
* `clicon_xml2cbuf()`: Added `skiptop` parameter: `clicon_xml2cbuf(..., int skiptop)`
* `xml2cli()`: Added `skiptop` parameter: `xml2cli(..., int skiptop)`
* Fixed: [very slow execution of load_set_file #288](https://github.com/clicon/clixon/issues/288)
* New `clixon-lib@2021-11-11.yang` revision
* Modified option: RPC stats extended with YANG stats
* Modified `clixon-config@2021-11-11.yang` revision
* Added option:
* CLICON_PLUGIN_CALLBACK_CHECK
* Enable to make plugin context check before and after all callbacks.
* Added statistics for YANG: number of objects and memory used
* Use the treeref no-copy option of CLIgen to reduce memory
* Refactored cli-generation/autocli-start code
* Refactored cligen glue functions to use cligen_eval directly (remove clicon_eval,clixon_cligen_eval)
* `count`, `name`, `contains`, `not`, as defined in [xpath 1.0](https://www.w3.org/TR/xpath-10)
* `deref`, `derived-from` and `derived-from-or-self` from RFC7950 Section 10.
* in particular in augment/when statements
* Improved error handling
* Verification of XPath functions is done at startup when yang modules are loaded, not when XPaths are evaluated.
* Separation of "not found" and "not implemented" XPath functions
* Both give a fatal error (backend does not start).
* `clixon_xml_parse_string()`
* `clixon_xml_parse_file()`
* New JSON parsing API, with same signature as XML parsing:
* `clixon_json_parse_string()`
* `clixon_xml_parse_file()`
* XML YANG binding API have been rearranged as follows:
* `xml_bind_yang_rpc()`
* `xml_bind_yang_rpc_reply()`
* `xml_bind_yang()`
* `xml_bind_yang0()`
* If you have used, `ys`, add `xml_spec_set(x, ys)` after the statement
* If you have `xml_type_set(x, TYPE)` or `xml_prefix_set(x, PREFIX)` immediately after the statement, you can remove those and set them directly as: `xml_new(name, PREFIX, xp, TYPE)`
w-level general upgrades and as a complement to module-specific upgrade.
* Called on startup after initial XML parsing, but before module-specific upgrades
* Enabled by definign the `.ca_datastore_upgrade`
* [General-purpose upgrade documentation](https://clixon-docs.readthedocs.io/en/latest/backend.html#ge
neral-purpose)
* JSON parse error messages change from ` on line x: syntax error,..` to `json_parse: line x: syntax err
or`
* Unknown-element error message is more descriptive, eg from `namespace is: urn:example:clixon` to: `Fai
led to find YANG spec of XML node: x with parent: xp in namespace urn:example:clixon`.
* C-API parse and validation API more capable
* `xml_spec_populate` family of functions extended with three-value return values
* -1: error, 0: parse OK, 1: parse and YANG binding OK.
* `xml_parse` and `json_parse` API changes
* Three value returns: -1: error, 0: parse OK, 1: parse and YANG binding OK.
* Extended `xml_parse_file2` and `xml_parse_string2` extended API functions with all options available.
* New concept called `yang_bind` that defines how XML symbols are bound to YANG after parsing
* Existing API same except `xml_parse_file` `endtag` argument moved to `xml_parse_file2`
* C-API: Added instrumentation: `xml_size` and `xml_stats_get`.
* Fixed: Enabling modstate (CLICON_XMLDB_MODSTATE), changing a revision on a yang, and restarting made the backend daemon exit at start (thanks Matt)
* Also: ensure to load `ietf-yang-library.yang ` if CLICON_XMLDB_MODSTATE is set
Major New features
* New and updated search functions using xpath, api-path and instance-id
* New search functions using api-path and instance_id:
* C search functions: `clixon_find_instance_id()` and `clixon_find_api_path()`
* Binary search optimization in lists for indexed leafs in all three formats.
* This improves search performance to O(logN) which is drastical improvements for large lists.
* You can also register explicit indexes for making binary search (not only list keys)
* For more info, see docs at [paths](https://clixon-docs.readthedocs.io/en/latest/paths.html) and
[search](https://clixon-docs.readthedocs.io/en/latest/xml.html#searching-in-xml)
API changes on existing features (you may need to change your code)
* On failed validation of leafrefs, error message changed from: `No such leaf` to `No leaf <name> matching path <path>`.
* CLI Error message (clicon_rpc_generate_error()) changed when backend returns netconf error to be more descriptive:
* Original: `Config error: Validate failed. Edit and try again or discard changes: Invalid argument`
* New (example): `Netconf error: application operation-failed Identityref validation failed, undefined not derived from acl-base . Validate failed. Edit and try again or discard changes"
Minor changes
* Test framework
* Added `-- -S <file>` command-line to main example to be able to return any state to main example.
* Added `test/cicd` test scripts for running on a set of other hosts
* C-code restructuring
* clixon_yang.c partitioned and moved code into clixon_yang_parse_lib.c and clixon_yang_module.c and move back some code from clixon_yang_type.c.
* partly to reduce size, but most important to limit code that accesses internal yang structures, only clixon_yang.c does this now.
The reason is to be conservative with the API. However, less used functions, such as xpath_vec_bool(), xpath_vec_ctx() and xpath_vec_flag() are changed with a new `nsc`parameter, which should be set to NULL in most cases.
* Names lexically correct (NCName)
* Syntactically Correct handling of '<?' (processing instructions) and '<?xml' (XML declaration)
* XML prolog syntax for 'well-formed' XML
* <!DOCTYPE (ie DTD) is not supported.
