1# zcbor v. 0.9.1 (2024-10-17) 2 3Any new bugs, requests, or missing features should be reported as [Github issues](https://github.com/NordicSemiconductor/zcbor/issues). 4 5## Improvements: 6 7 * README improvements: 8 * pypi_README.md: Add separate README for use in the pypi package 9 * README: Clarify what ZCBOR_STOP_ON_ERROR does 10 * RELEASE_NOTES.md: Move "Unsupported CDDL features" to the README 11 12## Bugfixes: 13 14## Unsupported CDDL features 15 16This section has been moved to the README, and this heading will be removed in subsequent releases. 17 18 19# zcbor v. 0.9.0 (2024-08-27) 20 21Any new bugs, requests, or missing features should be reported as [Github issues](https://github.com/NordicSemiconductor/zcbor/issues). 22 23## Improvements: 24 25 * zcbor.py: 26 * Allow --file-header to accept a path to a file with header contents 27 * Avoid generated names that are C keywords (capitalize them if present) 28 * Whenever generated code uses function pointers, add dummy calls using the function pointers, to verify that the arguments have the right type. 29 * Optimize range checks 30 * Convert function docs in comments to pep-257 docstrings and add some new docstrings 31 * C libraries: 32 * Remove zcbor_simple_*() functions from the header file to discourage their use. 33 * zcbor_decode: Allow deciding at run-time whether to enforce canonical data (via 'enforce_canonical'), instead of via ZCBOR_CANONICAL. 34 * zcbor_print.h: Move function implementations into new file zcbor_print.c 35 * Documentation: 36 * MIGRATION_GUIDE.md: Add a migration guide file 37 * README.md: Clarification on n_flags 38 * ARCHITECTURE.md: Small improvements to bring the document up to date 39 40## Bugfixes: 41 42 * zcbor.py: 43 * Fixes to support code generation on Windows 44 * Handle when files are on different drives in Windows 45 * Fix a few bugs in code generation to do with dereferencing 46 * Fix bug where NINT literals were not given unique names 47 * Fix a bug where empty groups and lists would give an error 48 * Fix bstr names based on size 49 * Don't omit zcbor_union_elem_code() when there is a range check. 50 * Fix bug where paths to generated files could not be absolute 51 * Fix a bug where some decoding function were not called 52 * Fix/improve deeply nested types 53 * Remove outdated assert statement that caused incorrect assert 54 * Fix a bug in CDDL parsing to do with nested brackets 55 * zcbor_decode: 56 * Fix various things related to constant_state 57 * Fix a bug with enforce_canonical (ZCBOR_CANONICAL) and floats 58 * zcbor_encode.c: Fix a length bug in zcbor_bstr_end_decode() 59 * zcbor_tags.h: Typo fix YANK -> YANG 60 61## Unsupported CDDL features 62Not all features outlined in the CDDL specs [RFC8610](https://datatracker.ietf.org/doc/html/rfc8610), [RFC9090](https://datatracker.ietf.org/doc/html/rfc9090), and [RFC9165](https://datatracker.ietf.org/doc/html/rfc9165) are supported by zcbor. 63The following is a list of limitations and missing features: 64 65 * Generated code does not support unordered maps. 66 * Using `&()` to turn groups into choices (unions). `&()` is supported when used with `.bits`. 67 * Representation Types (`#x.y`), except for tags (`#6.y(foo)`) which are supported. 68 * Unwrapping (`~`) 69 * The control operators `.regexp`, `.ne`, `.default`, and `.within` from RFC8610. 70 * The control operators `.sdnv`, `.sdnvseq`, and `.oid` from RFC9090. 71 * The control operators `.plus`, `.cat`, `.det`, `.abnf`, `.abnfb`, and `.feature` from RFC9165. 72 * Generics (`foo<a, b>`). 73 * Using `:` for map keys. 74 * Cuts, either via `^` or implicitly via `:`. 75 * Most of the "Extended Diagnostic Notation" is unsupported. 76 77 78# zcbor v. 0.8.1 (2024-01-26) 79 80Any new bugs, requests, or missing features should be reported as [Github issues](https://github.com/NordicSemiconductor/zcbor/issues). 81 82See also the [release notes for 0.8.0](#zcbor-v-080-2024-01-09) below. 83 84## Improvements: 85 86 * zcbor.py: Use zcbor_uint_decode() when decoding only positive enums 87 * zcbor_common.h: Add ZCBOR_VERSION macro 88 89## Bugfixes: 90 91 * zcbor_common.h: Add forward declaration for strnlen() 92 * zcbor.py: Fix conversion of UINT8_MAX to avoid script errors when using e.g. 255 in CDDL 93 * zcbor_print.h: Fix iterator type and print formats to avoid compiler warnings 94 95## Unsupported CDDL features 96Not all features outlined in the CDDL specs [RFC8610](https://datatracker.ietf.org/doc/html/rfc8610), [RFC9090](https://datatracker.ietf.org/doc/html/rfc9090), and [RFC9165](https://datatracker.ietf.org/doc/html/rfc9165) are supported by zcbor. 97The following is a list of limitations and missing features: 98 99 * Using `&()` to turn groups into choices (unions). `&()` is supported when used with `.bits`. 100 * Representation Types (`#x.y`), except for tags (`#6.y(foo)`) which are supported. 101 * Unwrapping (`~`) 102 * The control operators `.regexp`, `.ne`, `.default`, and `.within` from RFC8610. 103 * The control operators `.sdnv`, `.sdnvseq`, and `.oid` from RFC9090. 104 * The control operators `.plus`, `.cat`, `.det`, `.abnf`, `.abnfb`, and `.feature` from RFC9165. 105 * Generics (`foo<a, b>`). 106 * Using `:` for map keys. 107 * Cuts, either via `^` or implicitly via `:`. 108 * Most of the "Extended Diagnostic Notation" is unsupported. 109 110 111# zcbor v. 0.8.0 (2024-01-09) 112 113Any new bugs, requests, or missing features should be reported as [Github issues](https://github.com/NordicSemiconductor/zcbor/issues). 114 115## Improvements: 116 * C library: Add support for searching for elements in a map. 117 * Overhaul zcbor logging/printing 118 * Refactor printing of errors from generated functions 119 * zcbor_print.h: Improve zcbor_trace() and other tracing 120 * zcbor_print.h: Move from printk to printf 121 * Rename zcbor_print() to zcbor_log() 122 * zcbor_print.h: Move printing code to a new file zcbor_print.h 123 * zcbor_debug.h: Add function for getting error string and printing it 124 * Performance improvements 125 * zcbor.py: Refactor the regex matching to fix label and remove all_types_regex 126 * zcbor.py: Merge regexes for pure types without values 127 * zcbor.py: Add a cache for compiled regex patterns 128 * zcbor.py: Don't regenerate the big list of regexes every call 129 * zcbor_encode: Simplify value_encode() by using the new zcbor_header_len 130 * zcbor_decode.c: Streamline value_extract 131 * zcbor_common: Refactor zcbor_header_len 132 * zcbor.py: Don't use zcbor_present_encode() 133 * zcbor_common: Add a zcbor_entry_function helper 134 * Code generation name improvements: 135 * zcbor.py: Make generated variable names C-compatible. 136 * zcbor.py: Change name of generated choice enum members to add a '_c' 137 * Avoid leading and trailing as well as repeated underscores 138 * API cleanup/improvements: 139 * include: Rearrange and improve zcbor_decode.h and zcbor_encode.h 140 * zcbor_encode: Change zcbor_tag_encode to take a pointer argument 141 * zcbor_common: Rename ZCBOR_FLAG_TRANSFER_PAYLOAD to ZCBOR_FLAG_KEEP_PAYLOAD 142 * zcbor_decode: Add _pexpect() functions 143 * Internal cleanup: 144 * zcbor_encode, zcbor_decode: Move inline functions from header files 145 * zcbor_decode.c: Move MAJOR_TYPE() and ADDITIONAL() macros to zcbor_common.h 146 * src: Abstract float16 conversion and move it to zcbor_common.c 147 * zcbor_common: Add ZCBOR_FLAG_KEEP_DECODE_STATE to zcbor_process_backup() 148 * Move zcbor_array_at_end() from zcbor_common to zcbor_decode 149 * Move from setup.py to pyproject.toml for creation of zcbor packages 150 * zcbor_decode: Add validation that data follows canonical CBOR rules 151 * zcbor.py: Change the label regex to accept non-latin characters 152 * C library: Change usage of strlen to use strnlen 153 154## Bugfixes: 155 * zcbor.py: Fix range_checks for 'OTHER' type 156 * zcbor_decode.c: Fix "'num_decode' may be used uninitialized" 157 * zcbor.py: Generate cmake files with linux paths on Windows 158 * Remove references to uint_fast32_t 159 * Use ZCBOR_BIG_ENDIAN instead of CONFIG_BIG_ENDIAN 160 * zcbor.py: Fix cborhex generation so it generates newlines instead of spaces 161 * zcbor.py: Adjust the lower bound on negative numbers. 162 * zcbor_encode, zcbor_decode: fix double promotion warnings 163 * zcbor_print.h: Add missing errors to zcbor_error_str() 164 * zcbor_print.h: Fix memcmp expression 165 * zcbor.py: DataTranslator: fix type checking 166 167## Unsupported CDDL features 168Not all features outlined in the CDDL specs [RFC8610](https://datatracker.ietf.org/doc/html/rfc8610), [RFC9090](https://datatracker.ietf.org/doc/html/rfc9090), and [RFC9165](https://datatracker.ietf.org/doc/html/rfc9165) are supported by zcbor. 169The following is a list of limitations and missing features: 170 171 * Using `&()` to turn groups into choices (unions). `&()` is supported when used with `.bits`. 172 * Representation Types (`#x.y`), except for tags (`#6.y(foo)`) which are supported. 173 * Unwrapping (`~`) 174 * The control operators `.regexp`, `.ne`, `.default`, and `.within` from RFC8610. 175 * The control operators `.sdnv`, `.sdnvseq`, and `.oid` from RFC9090. 176 * The control operators `.plus`, `.cat`, `.det`, `.abnf`, `.abnfb`, and `.feature` from RFC9165. 177 * Generics (`foo<a, b>`). 178 * Using `:` for map keys. 179 * Cuts, either via `^` or implicitly via `:`. 180 * Most of the "Extended Diagnostic Notation" is unsupported. 181 182 183# zcbor v. 0.7.0 (2023-04-01) 184 185Any new bugs, requests, or missing features should be reported as [Github issues](https://github.com/NordicSemiconductor/zcbor/issues). 186 187## Improvements: 188 189### Code generation 190 * Add support for float16 encoding and decoding 191 * Add support for C++ in generated code 192 * Use (U)INT(8|16|32|64)_(MIN|MAX) macros in generated code 193 * Add the --c-code-columns option for adding newlines to c_code 194 * Refactor CBOR <-> YAML/JSON translation 195 * Allow adding file headers to generated files, e.g. for copyright. 196 197### C libraries 198 * Add support for float16 encoding and decoding 199 * Add helper functions zcbor_compare_strings() and zcbor_header_len() 200 * Rename prim/primitive to simple 201 * Add new APIs zcbor_uint_(encode/decode) 202 * Move tags enum into separate file zcbor_tags.h and add more tags from the IANA list 203 * When generating cmake files, make all paths relative to the cmake file. 204 * Use the zcbor_compare_strings() function in str_expect() 205 * Make the zcbor source files compile with -Wall and -Wconversion 206 * Refactor zcbor_any_skip() 207 * Change all uint_fast32_t usage to use size_t instead 208 209### Other 210 * README: Improve docs, particularly to help new users 211 * Add two samples: a hello world with no code generation, and one with code generation and script usage 212 * Port decode/encode/unit tests to new ztest API 213 * Fix fuzz tests 214 215## Bugfixes: 216 * zcbor.py: Fix 'zcbor --version' so it works again 217 * zcbor.py: Fix cbor bstrs so they properly check unambiguous values 218 * zcbor.py: Fix rendering of entry functions when they have no output 219 * zcbor.py: Don't flag cbor bstrs as unambiguous 220 221## Unsupported CDDL features 222Not all features outlined in the CDDL specs [RFC8610](https://datatracker.ietf.org/doc/html/rfc8610), [RFC9090](https://datatracker.ietf.org/doc/html/rfc9090), and [RFC9165](https://datatracker.ietf.org/doc/html/rfc9165) are supported by zcbor. 223The following is a list of limitations and missing features: 224 225 * Map elements in data must appear in the same order as they appear in the CDDL. 226 * C API doesn't support searching through a map. 227 * Using `&()` to turn groups into choices (unions). `&()` is supported when used with `.bits`. 228 * Representation Types (`#x.y`), except for tags (`#6.y(foo)`) which are supported. 229 * Unwrapping (`~`) 230 * The control operators `.regexp`, `.ne`, `.default`, and `.within` from RFC8610. 231 * The control operators `.sdnv`, `.sdnvseq`, and `.oid` from RFC9090. 232 * The control operators `.plus`, `.cat`, `.det`, `.abnf`, `.abnfb`, and `.feature` from RFC9165. 233 * Generics (`foo<a, b>`). 234 * Using `:` for map keys. 235 * Cuts, either via `^` or implicitly via `:`. 236 * Most of the "Extended Diagnostic Notation" is unsupported. 237 238 239# zcbor v. 0.6.0 (2022-10-12) 240 241Any new bugs, requests, or missing features should be reported as [Github issues](https://github.com/NordicSemiconductor/zcbor/issues). 242 243## Improvements: 244 * Refactor zcbor CLI to add 'verify' and put all options after keyword 245 * Make several C++ related improvements in the C library 246 * Fix Coverity warnings in C library and generated code 247 * zcbor_decode: Add zcbor_int[32|64]_expect_union 248 * zcbor_common.h: Add new zcbor_assert_state() macro 249 * zcbor_common.h: Rename ARRAY_SIZE to ZCBOR_ARRAY_SIZE to avoid #include order problems in Zephyr 250 * .github: Refactor Github action and add QEMU testing on mps2_an521 (ARM) and qemu_malta_be (MIPS, big-endian) 251 * zcbor.py: Raise argparse error on insufficient --output-* args 252 * zcbor_decode.h: Define variables to avoid rvalue address errors 253 * Add release dates to all RELEASE_NOTES entries 254 255## Bugfixes: 256 * zcbor.py: Fix bug where the union_int optimization decodes twice 257 * zcbor_encode.c: Fix error for big-endian architectures 258 259## Unsupported CDDL features 260Not all features outlined in the [CDDL spec](https://datatracker.ietf.org/doc/html/rfc8610) are supported by zcbor. 261The following is a list of limitations and missing features: 262 263 * Map elements in data must appear in the same order as they appear in the CDDL. 264 * C API doesn't support searching through a map. 265 * 16-bit floating point numbers (float16). 266 * Using `&()` to turn groups into choices (unions). `&()` is supported when used with `.bits`. 267 * Representation Types (`#x.y`), except for tags (`#6.y(foo)`) which are supported. 268 * Unwrapping (`~`) 269 * The control operator `.regexp`. 270 * The control operator `.ne`. 271 * The control operator `.default`. 272 * Generics (`foo<a, b>`). 273 * Most of the "Extended Diagnostic Notation" is unsupported. 274 275 276# zcbor v. 0.5.1 (2022-06-21) 277 278Any new bugs, requests, or missing features should be reported as [Github issues](https://github.com/NordicSemiconductor/zcbor/issues). 279 280## Improvements: 281 * requirements.txt: Change PyYAML requirement from >6.0.0 to >5.4.1 to be more widely compatible. 282 * zcbor_encode/zcbor_decode: Add new functions zcbor_int_encode() and zcbor_int_decode() which accept the integer size as an argument. 283 * zcbor.py: Use zcbor_int_encode() and zcbor_int_decode() when operating directly on _choice enums, since enums can be any size, depending on compiler and project. This removes the restrictions that enums must be 32 bits. 284 285## Bugfixes: 286 * tests: Fix an error in manifest3.cddl (encode/test1) (naming collision with 'uri') 287 288## Unsupported CDDL features 289Not all features outlined in the [CDDL spec](https://datatracker.ietf.org/doc/html/rfc8610) are supported by zcbor. 290The following is a list of limitations and missing features: 291 292 * Map elements in data must appear in the same order as they appear in the CDDL. 293 * C API doesn't support searching through a map. 294 * 16-bit floating point numbers (float16). 295 * Using `&()` to turn groups into choices (unions). `&()` is supported when used with `.bits`. 296 * Representation Types (`#x.y`), except for tags (`#6.y(foo)`) which are supported. 297 * Unwrapping (`~`) 298 * The control operator `.regexp`. 299 * The control operator `.ne`. 300 * The control operator `.default`. 301 * Generics (`foo<a, b>`). 302 * Most of the "Extended Diagnostic Notation" is unsupported. 303 304 305# zcbor v. 0.5.0 (2022-06-13) 306 307Any new bugs, requests, or missing features should be reported as [Github issues](https://github.com/NordicSemiconductor/zcbor/issues). 308 309## Improvements: 310 311### Code generation 312 * Add the --short-names option when generating code, which shortens names in struct definitions. 313 * Allow --default-max-qty to accept a string label when generating code so that the value can be configured after generation. 314 * Update versions of dependencies in pip requirements. 315 316### C Libraries 317 * Add const to *ptr argument of cbor_tstr_encode_ptr. 318 * Make it easier to initialize zcbor_state_t objects: Allow single states as input to zcbor_new_state() and remove return value. 319 * Make zcbor_list/map_end_encode() more robust to wrong max_num arguments (when ZCBOR_CANONICAL is enabled). 320 * Add zcbor_peek_error() for reading the error value without clearing it. 321 * Add zcbor_size_put/encode/decode/expect functions, to allow directly encoding and decoding size_t type variables. 322 323### Tests 324 * Test on multiple Python versions 325 * Add tests for using ranges (< or >) with floats in code generation. 326 * Add/enable test for signature check of SUIT manifests. 327 328 329## Bugfixes: 330 * zcbor.py: Fix missing size check for repeated bstrs with their own type 331 * zcbor_encode.c: Fix encoding of floats equal to 0.0 (or any with leading 0s) 332 * zcbor_encode.h: Fix "taking address of rvalue" errors with some compiler configurations, stemming from zcbor_b|tstr_encode_ptr() 333 * zcbor.py: Add forward declarations to functions in generated code. 334 * zcbor.py: Fix bug when adding tagged (#6.x) items to larger types, e.g. lists. 335 * zcbor.py: Fix naming bug for container types with no members. 336 * zcbor.py: "union_int" optimization: If all union members start with an integer, the optimization does a single int_decode() and identifies the correct union member by the result. 337 * zcbor.py: Fix a bug preventing the "union_int" optimization from being applied. 338 * zcbor.py: Expand the "union_uint" optimization to work on all integers, instead of only positive integers. 339 340 341## Unsupported CDDL features 342Not all features outlined in the [CDDL spec](https://datatracker.ietf.org/doc/html/rfc8610) are supported by zcbor. 343The following is a list of limitations and missing features: 344 345 * Map elements in data must appear in the same order as they appear in the CDDL. 346 * C API doesn't support searching through a map. 347 * 16-bit floating point numbers (float16). 348 * Using `&()` to turn groups into choices (unions). `&()` is supported when used with `.bits`. 349 * Representation Types (`#x.y`), except for tags (`#6.y(foo)`) which are supported. 350 * Unwrapping (`~`) 351 * The control operator `.regexp`. 352 * The control operator `.ne`. 353 * The control operator `.default`. 354 * Generics (`foo<a, b>`). 355 * Most of the "Extended Diagnostic Notation" is unsupported. 356 357 358# zcbor v. 0.4.0 (2022-03-31) 359 360**cddl-gen has been renamed to zcbor!** 361 362Any new bugs, requests, or missing features should be reported as [Github issues](https://github.com/NordicSemiconductor/zcbor/issues). 363 364## Improvements: 365 366### CDDL language 367 * Add support for 32-bit and 64-bit floating point numbers. 368 * Add support for `undefined` in CDDL. 369 * Add the CDDL prelude (from RFC 8610 Appendix D) to all builds (unless excluded with `--no-prelude`) 370 * Add support for more complex `.cbor` and `.cborseq` expressions. 371 372 373### Command line and Python module 374 * Rename script/module to zcbor. 375 * Better 64-bit support: 376 * New argument: `--default-bit-size` which controls the assumed size of integers. See zcbor code `--help` for more information. 377 * Add the `--include-prefix` option for path prefix to generated h files. 378 * Allow using both `--encode` and `--decode` together to generate both at the same time. 379 * Make changes to generated code to build with `-Wpedantic` and `-Wconversion`. 380 381 382### C Libraries 383 * All public function names have been prefixed with `zcbor_`. 384 * Many function names have also been changed slightly. 385 * Better API docs. 386 * Add error value to state, which is populated when a function fails (returns `false`). 387 * Add a `stop_on_error` feature that makes functions abort when an error has occurred (to avoid checking the output of all functions). 388 * Add support for 32-bit and 64-bit floating point numbers. 389 * Better 64-bit support: 390 * New 64-bit integer decoding and encoding functions (in addition to existing 32-bit functions). 391 * `uint32_t` usage has been largely transitioned to `size_t` or `uint_fast32_t` which are typically the same size as the architecture (32 or 64 bits). 392 * This means that list/map/string sizes can be 64 bits on 64-bit architectures. Note that having 64-bit sizes on 32-bit architectures rarely make sense since they won't fit in the address space. 393 * New helper macros for instantiating state variables: `ZCBOR_STATE_D()` and `ZCBOR_STATE_E()` for decoding and encoding respectively. 394 * Add support for decoding payloads split over multiple memory chunks. 395 * There are new APIs in zcbor_common.h and zcbor_decode.h to handle this. Look for `zcbor_update_state()` and various functions with `fragment` in the name. 396 * Strings can be split between chunks, there is a new type `struct zcbor_string_fragment` for this purpose. 397 * Data headers can not be split. 398 * Backups to previous chunks cannot be restored. 399 * Split `zcbor_multi_encode()` into `zcbor_multi_encode()` and `zcbor_multi_encode_minmax()`. The former is more useful for manual use, while the latter is for generated code. 400 * Remove `zcbor_any_encode()` since it was just an alias for `zcbor_nil_put()`. 401 * Rename `zcbor_string_type_t` to `struct zcbor_string`. 402 * Rename `zcbor_state_backups_t` to `struct zcbor_state_constant`. 403 * Add support for decoding and encoding `undefined` CBOR values. 404 * `zcbor_any_decode()`: 405 * Rename to `zcbor_any_skip()` 406 * Add support for indeterminate-length arrays. 407 * Have `zcbor_any_skip()` consume any tags that are present. 408 * Introduce `zcbor_bstr_expect_ptr()`, `zcbor_bstr_put_ptr()`, `zcbor_tstr_expect_ptr()`, and `zcbor_tstr_put_ptr()` 409 * zcbor_common.h: Add an enum (`zcbor_rfc8949_tag`) with some tag values. 410 * Move CBOR intro text from zcbor_decode.h to README. 411 * Control asserts separately from verbose (add ZCBOR_ASSERT option) 412 * Introduce `zcbor_list_map_end_force_encode()` and `zcbor_list_map_end_force_decode()`. 413 414 415### Tests 416 * decode: Remove test4 since it hasn't been used or updated. 417 * New unit test directory and project. Tests: 418 * 64-bit vs 32-bit integers. 419 * 64-bit vs 32-bit strings. 420 * String encoding and decoding macros (`_lit()`, `_arr()`, `_term()`). 421 * Payload chunks and string fragments. 422 * Collect boilerplate Cmake code for tests into a new file test_template.cmake. 423 * Add tests for draft-ietf-suit-manifest-16. 424 * Rename "strange" tests to "corner_cases" 425 * Build tests with `-Wpedantic` and `-Wconversion` 426 427 428## Bugfixes: 429 * zcbor.py: Avoid #include-ing the full path of the type file. 430 * zcbor.py: Avoid creating functions with "None" result type. 431 * src: Fix code so it builds correctly with asserts enabled. 432 * zcbor_decode.c: Fix `list_map_start_decode()` to handle elem_count correctly 433 * zcbor_decode.c: Fix `zcbor_any_skip()` so it reverts the state correctly on fail. 434 * zcbor.py: Fix regex for ints/floats to ignore names with hyphens (-) 435 * zcbor.py: Fix error in 'convert' where a bstr dict value was incorrectly translated 436 * setup.py: Fix windows install 437 * zcbor.py: Add suffix for literal values larger than 32-bits. 438 * src: Fix code to build with `-Wpedantic` and `-Wconversion`. 439 * zcbor.py: Fix some code generation bugs to do with range checking (min/max size) of ints and CBOR-encoded bstrs. 440 * zcbor.py: Fix bug where indirection for ints with range checks was wrong. 441 * zcbor.py: Fix bugs in range checks in `bstr`s with `.cbor` statements. 442 443## Unsupported CDDL features 444 445Not all features outlined in the [CDDL spec](https://datatracker.ietf.org/doc/html/rfc8610) are supported by zcbor. 446The following is a list of limitations and missing features: 447 448 * Map elements in data must appear in the same order as they appear in the CDDL. 449 * C API doesn't support searching through a map. 450 * 16-bit floating point numbers (float16). 451 * Using `&()` to turn groups into choices (unions). `&()` is supported when used with `.bits`. 452 * Representation Types (`#x.y`), except for tags (`#6.y(foo)`) which are supported. 453 * Unwrapping (`~`) 454 * The control operator `.regexp`. 455 * The control operator `.ne`. 456 * The control operator `.default`. 457 * Generics (`foo<a, b>`). 458 * Most of the "Extended Diagnostic Notation" is unsupported. 459 460 461# cddl-gen v. 0.3.0 (2021-11-17) 462 463Any new bugs, requests, or missing features should be reported as [Github issues](https://github.com/NordicSemiconductor/cddl-gen/issues). 464 465## Improvements: 466 467### CDDL language 468 * Add more support for hexadecimal/octal/binary numbers in CDDL. E.g. in size ranges or value limits 469 * Add support for ranges with three dots in CDDL files 470 * Change CDDL quantifier syntax from `x**y` to `x*y` 471 * Add support for `$$` and `$` 472 * Add support for the `.bits` operator. 473 474### Command line 475 * Change command line name from `cddl_gen` to `cddl-gen` 476 * Add `--dq` as a shorthand for `--default-max-qty` 477 * Allow passing multiple cddl files via command line 478 * Allow adding the current git sha to generated file headers. 479 * Various changes to the logic for naming in code generation 480 * This can give different names of e.g. struct members compared with previous versions of cddl-gen 481 * Generate a cmake file when generating code. 482 * The `target_cddl_source()` Cmake function has been removed. 483 * See README.md for more info. 484 485### Python module 486 * Improve tags support in DataTranslator 487 * Improve decoding of cbor bstrs in DataTranslator 488 * Provide the raw string along with the decoded object. 489 490### C Libraries 491 * Refactor cbor_common.c 492 * Add support for indefinite length arrays in cbor_decode.c. 493 494### Tests 495 * Move all cddl files into tests/cases 496 * Run tests with twister instead of custom bash scripts 497 * Add tests for draft-ietf-suit-manifest-14 498 499 500## Bugfixes: 501 * cbor_encode.h: Add missing `.` in `tstrx_put` and `tstrx_put_term` 502 * cbor_encode.c: Fix `get_result_len` and add big-endian support 503 * cddl_gen.py: Fix bug where range checks were dropped 504 * cbor_decode.c: Fix bug where a `nil` could be interpreted as `true` 505 * cddl_gen.py: Fix parsing of ranges with no max (`n*` or `n**`) 506 * cbor_encode.c: Fix bug where minimum quantity restrictions were ignored 507 * cddl_gen: Fix generation of map-within-map 508 * cbor_decode.c: Add overflow checks and fix assert in `any_decode()`. 509 510 511## Unsupported CDDL features 512 513Not all features outlined in the [CDDL spec](https://datatracker.ietf.org/doc/html/rfc8610) are supported by cddl-gen. 514The following is a list of limitations and missing features: 515 516 * Map elements in data must appear in the same order as they appear in the CDDL. 517 * Floating point numbers. 518 * Using `&()` to turn groups into choices (unions). `&()` is supported when used with `.bits`. 519 * Representation Types (`#x.y`), except for tags (`#6.y(foo)`) which are supported. 520 * Unwrapping (`~`) 521 * The control operator `.regexp`. 522 * The control operator `.ne`. 523 * The control operator `.default`. 524 * Generics (`foo<a, b>`). 525 * Not all types from the "standard prelude" are supported. 526 * Most of the "Extended Diagnostic Notation" is unsupported. 527
