C18 (also called C17) does not add any new features to C11, but it does incorporate changes and clarifications suggested in the following Defect Reports (DRs). For details, see:
- The last C18 public draft.
- A list of incorporated PRs (marked C17), which is reproduced below.
Below we outline whether a change was required to our semantics of C11 to incorporate the wording of a DR and what changes those were in each case.
| DR # | Summary | Change required? | Notes |
|---|---|---|---|
| 400 | realloc with size zero problems |
Yes | 5 |
| 401 | "happens before" cannot be cyclic | No | 2 |
| 402 | memory model coherence is not aligned with C++11 | No | 2 |
| 403 | malloc() and free() in the memory model |
No | 2 |
| 404 | joke fragment remains in a footnote | No | 3 |
| 405 | mutex specification not aligned with C++11 on total order | No | 2 |
| 406 | Visible sequences of side effects are redundant | No | 2 |
| 407 | SC fences do not restrict modification order enough | No | 1 |
| 410 | ilogb inconsistent with lrint, lround |
No | 1 |
| 412 | #elif |
No | 4 |
| 414 | typos in 6.27 threads.h | No | 3 |
| 415 | Missing divide by zero entry in Annex J.2 | Yes | 6 |
| 416 | Proposed defect report regarding tss_t |
No | 2 |
| 417 | Missing entries in Annex J | Yes | 6 |
| 419 | What the heck is a "generic function"? | No | 7 |
| 423 | underspecification for qualified rvalues | Yes | 8 |
| 426 | G.5.1: -yv and -x/v are ambiguous |
No | 1 (complex arith.) |
| 428 | runtime-constraint issue with sprintf family of routines in Annex K | No | 1 (bounds checking) |
| 429 | Should gets_s discard next input line when (s == NULL) ? |
No | 1 (bounds checking) |
| 430 | getenv_s, maxsize should be allowed to be zero |
No | 1 (bounds checking) |
| 431 | atomic_compare_exchange: What does it mean to say two structs compare equal? |
No | 2 (atomic) |
| 433 | Issue with constraints for wide character function | No | 1 (bounds checking) |
| 434 | Missing constraint w.r.t. Atomic | No | 9 |
| 436 | Request for interpretation of C11 6.8.5#6 | No | 9 |
| 437 | clock overflow | No | 1 |
| 438 | ungetc / ungetwc and file position after discarding push back |
No | 9 |
| 439 | Issues with the definition of "full expression" | TODO | sequence points |
| 441 | Floating-point issues in C11 from PDTS 18661-1 UK review, Issue 2 | No | 1 |
| 444 | Issues with alignment in C11, part 1 | Yes | 10 |
| 445 | Issues with alignment in C11, part 2 | Yes | 10 |
| 447 | Boolean from complex | No | 1 |
| 448 | What are the semantics of a # non-directive? |
No | 1 |
| 450 | tmpnam_s clears s[0] |
No | 1 |
| 452 | Effective Type in Loop Invariant | Yes | 11 |
| 453 | Atomic flag type and operations | No | 1 |
| 457 | The ctime_s function in Annex K defined incorrectly |
No | 1 (bounds checking) |
| 458 | ATOMIC_XXX_LOCK_FREE macros not constant expressions |
No | 1 |
| 459 | atomic_load missing const qualifier |
No | 1 |
| 460 | aligned_alloc underspecified |
Yes | 12 |
| 462 | Clarifying objects accessed in signal handlers | TODO | signals |
| 464 | Clarifying the Behavior of the #line Directive |
No | 4 |
| 465 | Fixing an inconsistency in atomic_is_lock_free |
No | 2 (atomic) |
| 468 | strncpy_s clobbers buffer past null |
No | 1 (bounds checking) |
| 470 | mtx_trylock should be allowed to fail spuriously |
No | 2 (threads.h) |
| 471 | Complex math functions cacosh and ctanh |
No | 1 |
| 472 | Introduction to complex arithmetic in 7.3.1p3 wrong due to CMPLX |
No | 1 (complex arith.) |
| 473 | "A range error occurs if x is too large." is misleading | No | 1 (math.h) |
| 475 | Misleading Atomic library references to atomic types | No | 2 (stdatomic.h) |
| 477 | nan should take a string argument |
No | 1 |
| 480 | cnd_wait and cnd_timewait should allow spurious wake-ups |
No | 2 (threads.h) |
| 481 | Controlling expression of _Generic primary expression |
Yes | 13 |
| 485 | Problem with the specification of ATOMIC_VAR_INIT |
No | 1 |
| 487 | timespec vs. tm |
No | 1 |
| 491 | Concern with Keywords that Match Reserved Identifiers | No | 9 |
-
No change required: DR concerns a feature or part of the C11 standard which our C11 semantics did not cover, e.g., the preprocessor, much of the standard library, and optional extensions. In the future, if we model this feature with our semantics, we will incorporate this DR.
-
No change required: We have very limited support for multi-threaded programs and the C/C++11 shared memory consistency model. In the future, when we implement the C/C++11 memory consistency model, we will incorporate this DR.
-
No change required: DR wording has little or no normative impact (such as a typo fix) and requires no change to our project.
-
No change required: DR concerns the preprocessor, which is not modeled by our semantics.
-
Change required: added a check for the implementation-defined behavior resulting from passing a zero size to
realloc(as well asmalloc,calloc, andaligned_alloc). -
Change required: because we reference the items listed in Appendix J (portability issues) by the order in which they appear, we need to update references to this appendix with the new ordinals.
-
No change required: We have very limited support for
stdatomic.h. In the future, when we implement these features, we will incorporate this DR. -
Change required: changes were made to our semantics of casting values to qualified types. The type of such expressions is the unqualified type.
-
No change required: updated wording perhaps has normative impact or resolves some ambiguity, but conforms to our interpretation of the standard.
-
Change required: semantics of alignment.
-
Change required: effective type of objects with a temporary lifetime.
-
Change required: semantics of
aligned_allocwere updated to allow unsupported alignments to result in returning a null pointer and no undefined behavior. -
Change required: The semantics of calculating the type of the controlling expression in a generic selection were updated.