Systems/C Compiler

What's new in Release 2.30 of Systems/C

C Compiler:

Support for GCC's __attribute__((__transparent_union__)) attribute.
New -fident/-fno-indent options to control generation of .ident strings on platforms that support that.
New -fgcc-version=x.y.z option to specify desired GCC compatility (for Linux and z/TPF environments.)
New -fsyntax-only/-fno-syntax-only options to simply parse the source and not generate the program.
Support for __has_builtin in the C preprocessor.
Support __builtin_mempcpy() and __mempcpy.
The -D and -U options are now processed in the order they appear on the command line.
New __builtin_thread_pointer() builtin for environments that support thread-local-storage (z/TPF and z/Linux).
Support multiple -fpreinclude=filename options on the command line.
Support for GCC's __attribute__((__deprecated__)).
Support for GCC's __attribute__((__unavailable)).
The __attribute__((aligned)) attribute without any specification now indicates the maximum possible alignment.
The alignment of _Decimal128 changed to match GCC's for DCC, z/TPF and Linux environments. For LE environments the alignment will match IBM's.
For C23 mode (-fc23), a type-mismatch error is now generated for a function definition with no prototype when a previous prototype is in scope, e.g:
void f(int); void f() { }
now generates a type mismatch error for the declaration.
New -Wlabel-unused option and warning.
New -Wunused-function option and warning that diagnoses unused defined static functions in a compilation.
New -Wunused-parameter and -Wunused-variable options, and warnings.
New -trigraphs/-no-trigraphs option to enable/disable C trigraphs. C23 removes trigraphs from the language, -trigraphs specified after -fc23 can be used to re-enable them if needed.
Support C23's typeof() and typeof_unqual().
Support GCC's -fshort-enums option. -fshort-enums is equivalent to -fenum=SMALL and -fno-short-enums is equivalent to -fenum=INT.
For z/Linux CFI information, default to using .cfi_xxx instructions the GNU gas assembler. -fno-dwarf2-cfi-asm selects the old manually-generated tables.
If allowed, the __builtin_popcount() function takes advantage of the Miscellaneous-Instruction-Extentions Facility 3 enhancements to the POPCNT instruction to reduce the generated code
The compiler now supports __builtin_abort() and __builtin_trap(). For the z/Arch targets, __builtin_trap() generates 2 bytes of 0x0000 because the Principles of Operation say:
Operation code 00 hex will never be assigned to an instruction implemented in the CPU
Thus, when execution reaches those two bytes, the program will receive an INVALID OPERATION exception.
Support for GCC's -MM option which is similar to -M except that dependencies involving system include files are not included in the output.
The __attribute__((__packed__)) attribute is now recognized and applied to individual structure field members for Linux(z/TPF) and DCC modes as well as IBM compatibility mode. It should have always applied in Linux mode mode, but this represents a change in the ABI for DCC mode. For example, this structure:
struct packed_chars { char a:4; char b:8 __attribute__ ((packed)); char c:4; };
only uses 2 bytes where before it would have been 3.
Support conversion between 128-bit integers and the various floating point formats.
Support conversion between 128-bit integers and packed decimal data.
Support for -Wswitch and -Wswitch-enum. When the controlling expression of a switch statement is an enumerated type, the compiler will verify that all values of the enumeration appear as case labels.
Added -Wswitch-outside-range/-Wno-switch-outside-range which checks that a case label's value is in the range of the controlling switch expression.
Improvements in tracking the setting of the CC in the register allocator for improved code.
Improvements to many messages to include the involved types.
Support for C23 _BitInt(x) bit-precise integers and decimal constants.
Support for C23 apostrophe separators in decimal integer and floating point constants.
The compiler uses relative jumps within regions for performance if the architecture option allows.
New __stpcpy() and __builtin_stpcpy() compiler builtins.
Option names are migrating to the same style as GCC and CLANG, where '-' is used instead of '_'. The old option names continue to be accepted, but the new names will appear in the listing and documentation.
The optimizer now recognizes (s+/-C s) style expressions where 's' is a signed integer value. The assumption is that signed integer arithmetic is undefined when it overflows, so such an expression can be deduced to be TRUE or FALSE . The new -fwrapv and -fno-wrapv options control this assumption. -fwrapv indicates the compiler should assume signed integer overflow "wraps-around" and defeats such optimization. -fwrapv is the default but that may change in the future.
The optimizer now recognizes (ptr+/-C ptr) with the assumption that pointer arithmetic doesn't wrap-around. New -fwrapv-pointer and -fno-wrapv-pointer options to control this.
If -fc23 is specified, _Static_assert's string 's' is optional.
Support -fhosted/-fno-hosted options that controls the definition of __STDC_HOSTED__. Also - define __STDC_HOSTED__ if c99 or c11 is enabled; but only then. -ffreestanding is a synonym for -fno-hosted.
New warning #2055:
ISO C90 forbids mixed declarations and code
which is enabled by -fc90 and disabled by -fc99 or -fc11. This can be helpful if writing code that needs to also be compiled with older compilers.
For GCC compatibility, the compiler now recognizes __volatile as a synonym for volatile.
For MVS or z/OS targets, we properly follow z/OS convention of saving/restoring FPR8-FPR15 if used in a function. There is a new -mafp_volatile option also causes any function call to indicate the registers are altered and thus saved/restored in the calling function.
The code generated for the builtin strcmp() function was improved to 'swap' the comparison addresses for CLC and CLST instructions and take advantage of IPM + shifts to produce a -1, 0, +1 result.
The code generated for the builtin memcmp() function was improved to do no branches for constant lengths when possible. Also, the code was dramatically improved over the previous byte-by-byte comparison code to do the comparison in 256-byte chunks, then a final CLC for the remainder.
Enhanced support for GCC's __FUNCTION__ and support GCC's __PRETTY_FUNCTION__ pre-defined identifiers.
-@ processing improved on MVS to support quoted strings.

z/TPF:

Support for -fno-dwarf2-cfi-asm to allow the use of older GAS assemblers.
New -fasm-listing, -fno-asm-listing to the tpf-dcc and tpf-dcxx scripts to control inclusion of the GAS assembly listings.

C Library:

New pthread_main_np() function added to the library.
New mempcpy() function.
New void *__initial_parm_block_addr(void) function that returns the address of the incoming parm block for DCC MVS 32-bit and 64-bit environments.
Support for pthread-based POSIX async-I/O routines: aio_cancel(2), aio_error(2), aio_read(2), aio_return(2), aio_suspend(2), aio_write(2), lio_listio(2).
clock_gettime() supports CLOCK_MONOTONIC as well as CLOCK_REALTIME.
sigqueue() added to the runtime library.
Support pread() and pwrite() functions for HFS files.
Added the __ibmfd function that returns the underlying IBM HFS or SOCKET file descriptor for a Dignus file descriptor.
The strfmon() function and monetary.h header (POSIX) was added.
The __smf_record() function for writing SMF records was added.
BSAM I/O now supports the Large Block Interface (LBI) and if the system supports it will use it.
BSAM I/O block buffers are now allocated above the 16M line to free up low-mem usage (if the file is not a TSO terminal, where BSAM still requires the block buffer remain below-the-line.)
On most systems, the C library will now allocate a DCBE along with a DCB. The DCBE can often be allocated above the 16M line except for situations where IBM still requires it below (a DD or DSN name that specifies a MEMBER, and thus uses the DFSMS FIND macro will require the DCBE be allocated below-the-line.)
Support for the IBM __MAP_64 flag was added to mmap().
Support for the IBM __MAP_MEGA flag was added to mmap().
x37 and related I/O ABENDs for a BSAM write() will assign ENOSPC to errno instead of EIO.
New "abend=abend" and "abend=recover" attributes for BSAM file opens. If "abend=abend" is specified, then an ignorable I/O ABEND will become a "real" ABEND and need to be handled via normal ABEND methods (e.g. a signal handler for SIGABND.) If "abend=recover" is specified, the library will "ignore" ignorable I/O ABENDs and report the error to the caller and set errno to EIO or other appropriate value. The default is "abend=recover".
POSIX 1 stpcpy() and stpncpy() functions added to the library.
C library runtime options TRAP(ON)/TRAP(OFF) now take precendent over any environment decision about establishing an ESTAE, so that the ESTAE can just be completely turned off.
Signal handling now supports the POSIX SI_USER value in the siginfo_t passed to the signal handler.
The pathconf() and fpathconf() functions now support the newer _PC_ACL and _PC_ACL_ENTRIES_MAX options from IBM.
Added C11's stdalign.h header.
Added support for C11's max_align_t type in stddef.h.
Added POSIX functions strnlen() and wcsnlen().
The BFP round(), roundf() and roundl() functions could raise the INEXACT flag. The ANSI C99 and C11 standards allowed for that behavior, while the C14 standard does not. This function was changed to no longer raise the INEXACT flag in any situation.
The C library is now distributed in GOFF format, retiring the previous XSD format.

Utilities:

PLINK now supports MODE, ALIAS, and CHANGE command cards.
PLINK now generates GOFF output, which is now the default, and new -esd/-xsd/-goff options to control PLINK output format.

What's new in Release 2.25 of Systems/C

Debugger:

Kernel and debugger protocol updated to 1.1.0
The version 2.25 debugger contains new features, including support for a formatted dump of memory (xx command.)
Restarting the program without restarting the debugger.
Function calls (in the Dignus runtime environment) are now allowed in the debugging expressions.
New finish command for running and completing the current function.
Improved "look and feel" to be similar to UNIX debuggers.
Many command fixes and enhancements.

C Compiler:

Support for gcc-style __atomic* builtin functions.
Support for __builtin_fabsd32(), __builtin_fabsd64() and __builtin_fabsd128().
Proper compile-time folding of _DecimalN values.
Support for UTF-8 encoded source files.
Support gcc's __builtin_unreachable() function.
_Noreturn keyword supported when -fc11 enabled.
New -march=z13 flag to enable instructions from the 13th edition of the Principles of Operation.
Support for gcc's __builtin_swap16(), __builtin_swap32() and __builtin_swap64(). These will use hardware instructions if the -march level allows it.
Support for the fp-extensions facility (enabled with -march=z9) to use instructions to convert from IEEE floating-pt to unsigned integers.
If -mmisc-instructions-extensions-facility-2 is allowed, the compiler will now take advantage of the AGH/SGH and MGH instructions for operations involving 64-bit signed values and 16-bit signed values in memory
Support the new DFP conversion instructions made available the -mfp-extensions-facility is enabled (-march=z9 or greater.)
Support for gcc's __builtin_frame_address() and __builtin_return_address() builtin functions.
Support for new -mtpf-skip-trace option which alters z/TPF trace branch targets in the function prologue/expilogue.
Dramatically reduce the memory usage in the compilers, especially when optimization is enabled.
Optimization improvements for constant propagation of string addresses.
Support for the __builtin_isdigit() function.
When allowed by the -march option, conversions of 1_DecimalN values to floating point will now use hardware instructions.
Improvements in live variable analysis to speed up code generation for large functions.
Provides predefined macros __SIZE_TYPE__ and __PTRDIFF_TYPE__.
support _DecimalN types in __builtin_classify_type() (used by the va_arg() expansion.)
Do not convert _DecimalN/_FloatN values to int when applying default promotions. Also, (_Float32) is not promoted to (double) as (float) is as suggested by the standard.
Honor preprocessor "#" directives even inside the arguments to a macro.
Support the __builtin_sqrtf(), __builtin_sqrt() and __builtin_sqrtl() builtins from GCC, for both IEEE and HFP data

Utilities:

Dramatically improve PLINK time for -bonly links.
PLINK now honors -except on long names as well as short ones.
DCCPC now supports the new IPV6-related options to EXEC CICS ASSIGN, EXTRACT TCPIP, EXTRACT WEB, and WEB PARSE. And add STRINGZONE to EXEC CICS FORMATTIME.
Update DCCPC's CVDA (DFHVALUE) table to CICS TS V5.5.
DCCPC now supports all of the new operands for EXEC CICS INQUIRE SYSTEM as of CICS TS V5R4.

C library:

<float.h> now supports _FP_MODE_VARIABLE for the definitions of the various floating-pt constants/sizes, etc...
Add alarm(), setitimer() and getitimer() functions.
Support format-8 DSCBs when invoking OBTAIN for the osddinfo() function.
Added support for the usleep(3) function (based on the recently added nanosleep(2).)
Added support for C11 standard threading (based on the POSIX pthread support): call_once(), cnd_broadcast(), cnd_destroy(), cnd_init(), cnd_signal(), cnd_timedwait(), cnd_wait(), mtx_destroy(), mtx_init(), mtx_lock(), mtx_timedlock(), mtx_trylock(), mtx_unlock(), thrd_create(), thrd_current(), thrd_detach(), thrd_equal(), thrd_exit(), thrd_join(), thrd_sleep(), thrd_yield(), tss_create(), tss_delete(), tss_get(),tss_set().
Added support for POSIX threads (pthreads) (depends on BPX): pthread_atfork(), pthread_cleanup_pop(), pthread_cleanup_push(), pthread_condattr_destroy(), pthread_condattr_init(), pthread_cond_broadcast(), pthread_cond_destroy(), pthread_cond_init(), pthread_cond_signal(), pthread_cond_timedwait(), pthread_cond_wait(), pthread_create(), pthread_detach(), pthread_equal(), pthread_exit(), pthread_getspecific(), pthread_join(), pthread_key_delete(), pthread_kill(), pthread_mutexattr_destroy(), pthread_mutexattr_getprioceiling(), pthread_mutexattr_getprotocol(), pthread_mutexattr_gettype(), pthread_mutexattr_init(), pthread_mutexattr_setprioceiling(), pthread_mutexattr_setprotocol(), pthread_mutexattr_settype(), pthread_mutex_destroy(), pthread_mutex_init(), pthread_mutex_lock(), pthread_mutex_trylock(), pthread_mutex_unlock(), pthread_once(), pthread_rwlockattr_destroy(), pthread_rwlockattr_getpshared(), pthread_rwlockattr_init(), pthread_rwlockattr_setpshared(), pthread_rwlock_destroy(), pthread_rwlock_init(), pthread_rwlock_rdlock(), pthread_rwlock_unlock(), pthread_rwlock_wrlock(), pthread_self(), pthread_setcancelstate(), pthread_setcanceltype(), pthread_setspecific(), pthread_testcancel()
as well as the pthread barrier functions: pthread_barrier_destroy(), pthread_barrier_init() and pthread_barrier_wait().
and the non-POSIX additions pthread_attr_get_weight_np(), pthread_attr_set_weight_np(), pthread_attr_get_synctype_np(), pthread_attr_set_synctype_np() and pthread_set_limit_np().
Added POSIX nanosleep() function (depends on BPX.)
Added the POSIX semaphore functions to the C library, including:sem_destroy(), sem_getvalue(), sem_init(), sem_open(), sem_post(), sem_timedwait(), sem_trywait(), sem_unlink() and sem_wait().
sched_yield() added to the C library.
strchrnul() added to the C library.
unsetenv() now returns (int) to match the POSIX specification.
Added the POSIX clock_gettime(), clock_settime() and clock_getres() functions. Mainframe hardware only supports the REALTIME clock.
Added getdelim() and getline() POSIX functions.
Added wordexp() and wordfree() POSIX functions.
Added __abendcode() and __rsncode() functions to be able to determine these values in a SIGABND signal handler.
Added posix_spawn() and the associated functions to the C library.
Added vfork() to the C library.
New __getprv(2) (@@GETPRV) function to return the current PRV address.
Support for sigsuspend(2), sigwait(2) and pause(3) signal-related functions for POSIX signals.
O_CLOEXEC open(2) flag added.
pipe2() function added.
New __runopt variable that disables parsing parameter strings for runtime options. If the variable is defined, and the value is 0, then don't do runtime options processing in TSO or BATCH.
Support for lrecl=x open(2) attributes; also for a variable-spanned file defined with LRECL=X, the osddinfo() function returns a special LRECL_X value (32-bit MAXINT), the lrecl field in the return __fldata_t structure from the fldata() function also uses that value.
if the LRECL from the DCB for a Variable Spanned file is X (it has the high-order bit set) then we return a __maxreclen of 0x7fffffff in the fldata_t structure returned by the fldata() function.
Corrections to the record management for Variable Spanned record format. IBM subtracts 4 from the LRECL to be the maximum limit of the record length. This doesn't make too much sense since there could be many segment descriptors along for a spanned record; but it fits with the non-spanned Variable record approach and since IBM does it we now do as well. This does not occur with LRECL=X.
In BSAM I/O, when apply defaults for RECFM/LRECL and BLKSIZE, we now only provide an an inferred setting for RECFM if the DCB doesn't have one, the user didn't give us one, and the file is an output file. If after that, the DCBRECFM is still unset, then we end the open exit processing and allow the OPEN to proceed (likely getting an ABEND and driving the ABEND EXIT because the DCBRECFM is unset.) This aides in failing the open() of a FILEDEF for a non-existing file on CMS. (If the DCB is complete, CMS will allow the open for a non-existing file, which should fail.)
Add __reglen() and __regcpy() to the C library for copying s compiled regular expression.
Fix inadvertent forcing of TRAP(ON) even when it wasn't requested

What's new in Release 2.20 of Systems/C

Debugger:

Version 2.20 introduces the new Dignus Systems/DBG cross-platform debugger. See sysdbg.pdf for more information.

C compiler:

Support the __cs1()/__cds1()/__csg()/__cdsg() builtins in LE mode.
Support GCC's __float128 type, which is always 128-bits long no matter the setting of long_double_is_distinct.
Support for ISO/IEC TS 18661-3:2015, the _Float32/_Float32x/_Float64/_Float64x/_Float128 IEEE interchange and extended floating point types.
Support for __attribute__((used)) on a static function definition to indicate the compiler should not elide it, even if it appears unused. This is helpful for referencing the function from in-line assembly source.
For the Dignus environment, alloca() is now defined to be __builtin_alloca() (as Linux and, essentially, LE environments do.) This allows the compiler to be aware when alloca() has been used.
C listing file now specifies LRECL=133,RECFM=FBA on MVS to provide nice defaults if none are provided in JCL or allocated output file.
Add support for gcc-style Transactional-Execution Facility (-mtransaction-facility or -march=z10) builtins: __builtin_tbegin, __builtin_tbegin_nofloat, __builtin_tbegin_retry, __builtin_tbegin_retry_nofloat, __builtin_tend, __builtin_tabort, __builtin_tx_assist, __builtin_tx_nesting_depth, __builtin_non_tx_store).
Add support for -fgnu89_inline and __attribute__((gnu_inline)), which force the "inline" keyword to be treated like the pre-C99 gcc extension instead of the standard C99 meaning. This is compatible with how gcc treats inline since gcc version 4.3.
Support gcc-compatible __attribute__((packed)) on enum types, which uses the smallest container size that fits all of the enum values.
In IBM compatiblity mode, implement XPLINK __callback function pointers so that XPLINK code can call non-XPLINK function pointers passed in from other DLLs.
Improvements to floating-pt code generation/optimiztion to be more conforming and consistent between optimized and non-optimized code.
Duplicate file-scoped string constants now only appear once in the generated code.
Non-re-entrant visible data now goes into its own CSECT (instead of part of the @CODE section.) This allows for duplicate non-rent definitions to be elided at link time.
Correct issue in register allocation that caused very slow compile-time and excessive register spilling when encountering high register contention.

Utilites:

PLINK was internally restructured for dramatic link-time improvements.
Support BINDER comment cards ('*' in column one) in PLINK, the entire card is ignored and can't be continued.
DB2PP was enhanced to add support for several new DB2 V10 SQL statements: ALTER MASK, ALTER PERMISSION, ALTER TRIGGER, ALTER TRUSTED CONTEXT, CREATE MASK, CREATE PERMISSION, CREATE TRUSTED CONTEXT, CREATE TYPE, EXCHANGE, MERGE INTO.

C library:

Don't gratuitously set the normal disposition for files to CATALOG, as that causes the generation of IEF287I for input files. Only set the normal disposition to CATALOG if we are creating a new file for output, and the open() hasn't otherwise specified a disposition in the attributes.
Support for runtime options in non-USS (TSO and BATCH) parameter processing. In TSO and BATCH, anything up to the first '/' is considered a comma-separted list of runtime options. Supported options are:
ENVAR(namevalue-list)
where "namevalue-list" is a comma-separate list of the form
name=value
ENVAR Sets environment variable "name" to the value "val". The value may be single or double-quote string. Spaces are significant in both the name and the value. The name and value are limited to 1024 characters each.
Unrecognized options are silently ignored.
The POSIX sigsetops functions were added (sigset_t manipulation)
Corrected @@PDIV library routine to generate a decimal divide exception if the divisor is 0.
Added sleep() which will invoke the POSIX sleep function in the USS environment.
Added makecontext()/swapcontext()/getcontext()/setcontext() functions (these functions make use of the EXTRACT PSW and RESUME PROGRAM instructions, and thus require hardware that provides those. All z/Arch machines do, as well as later ESA/390 machines.)
New __pfsctl() function to invoke the BPX BPX1PCT (phsyical file system control) interface.
In support of POSIX signals, the setjmp()/longjmp() functions where changed to match the sigaction handling that POSIX specifies (to provide some interoperability with the new sigsetjmp()/siglongjmp() functions.)
For this reason, the size of the jmpbuf was changed to handle the expanded semantics.
Improved support for signal handling. The runtime library now supports sigaction(), sigprocmask() and other signal-related functions.
In the POSIX runtime environment (or if __bpxsig is set) the runtime establishes support for USS signals and can receive POSIX signals sent from other processes by the kill() function (or, other methods, e.g. SIGINT.)
In both POSIX and non-POSIX environments, an ESTAE exit is used to catch SIGSEGV, SIGILL, SIGFPE and SIGABND events and invoke a signal handler if one is defined.
If the runtime option TRAP(OFF) is specified, no ESTAE exit is established (via #pragma prolkey on the main() function.) The default is TRAP(ON).
The ESTAE exit is only established for programs that begin with main(), DCALL environments provide no signal catching (for either the POSIX or non-posix signals.)
Added support for reading/writing VARIABLE (BLOCKED) SPANNED OS files.

What's new in Release 2.10 of Systems/C

C compiler:

Allocation of offsets in the local stack frame was greatly improved to reduce total runtime stack size.
Many improvements to use newer instructions when the appropriate architecture level is selected. The maximum architecture level is now -march=z11 which represents the 11th revision of the z/Architecture Priniciples of Operation.
Loops that involve constant assignments are converted into memset-style operations.
__func__ is now supported in all runtime environments.
Support for __builtin_fpclassify() for HFP and BFP data. This is the 'generic' version; __builtin_bfp_fpclassify() accepts BFP data, and __builtin_hfp_fpclassify() accepts HFP data.
Support for floating-pt builtin functions: __builtin_frexpf, __builtin_frexp, __builtin_frexpl, __builtin_signbitf, __builtin_signbit, __builtin_signbitl, __builtin_isnanf, __builtin_isnan, __builtin_isnanl, __builtin_isinff, __builtin_isinf, __builtin_isinfl, __builtin_isfinitef, __builtin_isfinite, __builtin_isfinitel, __builtin_isnormalf, __builtin_isnormal, __builtin_isnormall These are 'generic' versions that depend on the current setting of the -fieee option. There are also BFP or HFP specific versions of these (with _bfp or _hfp qualifications after the __builtin portion of the name.)
The compiler now defaults to C99 mode (-fc99), there is a new -fc90 option to indicate that C90 mode be used.
Support the __builtin_return_address() builtin in all environments.
Support for bit manipulation builtin functions: __builtin_clz(), __builtin_clzl(), __builtin_clzll(), __builtin_ctz(), __builtin_ctzl(), __builtin_ctzll(), __builtin_ffs(), __builtin_ffsl(), __builtin_ffsll() If extended immediate instructions are allowed, these will use the FLOGR instruction; otherwise a binary search is generated.
Support for universal-character-name escape sequences in character and string literals if -fc99 or -fc11 is enabled.
Support U/u character literal prefixes and u8/U/u string literal prefixes if -fc11 is enabled. If these are specified the constant is comprised of UTF (ASCII) characters regardless of the -fasciiout option. Compile-time concatenation of string literals does not translate the adjacent strings.
New -fc11 option to indicate C11 standard compatibilty.
The C11 __STDC_ISO_10646__ pre-defined macro will be defined by the compiler if wchar_t is larger than 16 bits, except in Linux environments where that definition is provided by the runtime library header files.
Internal restructuring to improve compile-time performance for sources with many/large static initializations.
New support for -febcdic= option that indicates which EBCDIC code page to use for compile-time ASCII <-> EBCDIC translations. Consult the C compiler documentation for the allowed EBCDIC character sets.
The C compiler now predefines the __LP64__ symbol in all environments if -mlp64 is enabled.
Predefine the __NO_INLINE__ macro if inlining disabled.
If optimization is enabled, predefine the __OPTIMIZE__ and __OPTIMIZE_SIZE__ macros.
If the 3rd argument to __builtin_memchr() is 0, then the result is definitionally NULL.

C Library:

Support FD_FREEONCLOSE fcntl() flag for BSAM I/O. If FD_FREEONCLOSE is set, then the FREE option will be added to the BSAM CLOSE step.
Support for FIXED/FIXED BLOCKED output files with ASA control characters in text mode.
Low level BSAM I/O was rewritten to look toward supporting more modern z/OS features. Because of this, the calculations of DCB LRECL, BLKSIZE, RECFM now can be more subtle when values are missing. Consult the C library PDF for a discussion of how these values are now computed.
getaddrinfo() and getnameinfo() now invoke the IBM (BPX) routines according to the host: line setting in nsswitch().
The type (time_t) changed to a 64-bit value for the 32-bit; this may affect structure alignment/compatibility with existing data. This is in anticipation of the 2038 roll-over date for a 32-bit time_t. Note that (time_t) is used in a (struct timeval), which is also used in the select() function.
New dynalloc() and dynfree() functions for z/OS dynamic allocation support.

Utilities:

DB2PP now accepts "--" as a comment-to-end-of-line within an EXEC SQL statement.
DB2PP now accepts underscore ('_') in more DB2 identifiers; such as the name of an EXEC SQL CALL function.
DB2PP now supports SQL TYPE IS XML and BLOB_FILE / CLOB_FILE / DBCLOB_FILE.
DB2PP has a new -fccsid=x to set the default CCSID.
DB2PP supports EXEC SQL TRUNCATE (from DB2 V9).
D2S now has a -ds0xl option, which causes D2S to treat "DS 0XLn" as equivalent to "EQU n", for compatibility with some sources that used that idiom.

What's new in Release 2.00 of Systems/C

C Compiler:

Support __attribute__((visibility("hidden")) for Systems/C shared objects.
Support __attribute__((always_inline)) on functions to bypass compiler heuristics when determining if a function is eligible for inlining.
Support "dignus.inf-relative" paths for "System Include" specified in the license key file.
Support new -Spath option that is similar to IBM's SEARCH option for searching for included files.
New -fsave_dsa_over_call/-fno_save_dsa_over_call option to control saving/restoring bytes of the DSA over linkage-OS or linkage-ASM function calls.
Support range initializers in C99 mode for GCC compatibility.
Optimize __builtin functions that have constant operands.
New __bit_offsetof() and __bit_sizeof() operations similarly to offsetof() and sizeof() but returning the values in bits.
__int8/__int16/__int32/__int64 in all environments.
Recognize __attribute__((noreturn)) and provide warnings if a return statement is encountered.
Support for GCC-style inline assembly.
New -nodiginc option that indicates the system include directory specified in the license file should be ignored.
New -M=fn option to specify an optional file name for dependency information. New -fdep[=fn] option to generate the dependency information during the normal compilation.
Support new __grande and __regpair type modifiers for (long long) data. __grande indicates the (long long) value should be kept in a 64-bit register instead of two 32-bit registers. Also added new -fllgrande option that indicates __grande should be the default for (long long) data types.
New -fpreproc=filename option that writes preprocessor output to a file without halding the compilation.
Add -MT and -MF options compatibly with GCC. -MT filename provides the target name to use in dependency information. -MF names the output (equivalent to '=filename' for -fdep and -M.)
If in-line __asm__ or gcc-style asm("") appears in a function, generate an LTORG in the 4K region in which it appears so that the in-line ASM can reference ASM literals.
The compiler will now recognize a BASER=int specification in a #pragma prolkey() and use that given integer value as the code base register for the specific function. This allows the programmer to change the code-base for individual functions.
Support unsigned long long (64-bit) bitfields in 32-bit compilations.
Compiler input processing on z/OS was enhanced to improve speed reading input from PDS members.
Messages were changed to use single quotes instead of the historical backquotes.
New _Ieee and _Hexadec qualifiers for floating-point types that indicate the value is either using BFP or HFP format. Values in the "alien" format are converted to the default (as specified by the -fieee option) for operations. E.g.
double _Ieee b1; double _Hexadec h1; h1 = b1 + 10;
If the -fieee option is off, the addition is accomplished in HFP format, otherwise it is accomplished in BFP format and the result converted to HFP for the assignment.
New -fhfpfloatpromote option indicating if HFP (float) values should be promoted. For LE these are promoted to (double) when passed directly in parm blocks. For Linux and Systems/C mode, these are not promoted (to be compatible with IEEE values.) The -fnohfpfloatpromote option will disable the promotion, -fhfpfloatpromote enables it. Previous compiler versions did the promotion in Systems/C mode but this was changed.
Support C11's _Generic generic-selection.
New -fatid option that allows '@' characters in C identifiers (not the first character.)
Optimization improvements

PLINK utility:

PLINK now supports linking/autocalling load modules found in TSO TRANSMIT libraries on cross-platform hosts. Also, when creating load modules (in binding mode), it can add the PDS member to an existing TSO TRANSMIT file.
New options to improve creation of TSO TRANSMIT files.
Support 64-bit relocations in VSE output.
Support __attribute__((visibility("hidden")) for Systems/C shared objects.

DCCPC utility:

Updates to support CICS TS V4R1.
Added support for BIGNUM (long long).

DB2PPC utility:

Use new baseless addressing mode (BRAS instead of B) that's used in DB2 V10.
Add -fdb2ver=v8 and -fdb2ver=v10, which make DB2PPC match the IBM preprocessor output at V8 and V10, respectively. The changes are largely aesthetic -- trigraphs and white space.

C library:

Runtime will now default to using BPX-style sockets, can revert to EZASMI interface by declaring int __bpxso = 0 at global (extern) scope.
Dramatically improve _O_BINARY and O_TEXT read performance from FIXED (BLOCKED) files.
Support for LDS data sets in VSAM I/O. Added "unsafeshare" option to kopen() for processing LDS files without SHAREOPTIONS(1,3) on the allocation.
Support RECFM=U reading and writing for BINARY and TEXT files with and without _O_RECIO.
Accept Standard (S) specification on recfm= option in open() for FIXED (F) record-length files and mark the DCB accordingly.
Corrections and improvements in locating TCPIP.TCPIP.DATA file and related files.
Add the offset from the entry point to the corresponding return pointer to each line of __traceback output, so you can see which call point is active.
New poll() function (emulated based on select())
Added new RLSE and NORLSE attributes in open() processing indicating any unused space after a close is to be released or not. RLSE is the new default.
Support for both HFP and BFP values at runtime. The library now supports all the C99 standard math functions for both HFP and BFP (IEEE) floating point as well as I/O for both HFP and BFP values.
Improvements to floating-pt -> char printf conversions for denormalized and long-double HFP values.
Support for creating an LE CEEPIPI environment and invoking LE functions.
New __svc99 function to provide interface to DYNALLOC.
64-bit runtime BPX interface functions changed to no longer "thunk" to 31-bit interfaces - thus 64-bit programs that use the BPX interface now require z/OS 1.6 or later (z/OS 1.6 was GA in Sept. 2004 and went out of service on Sept 30, 2007.) The removal of "thunking" improves 64-bit program performance.

What's new in Release 1.98 of Systems/C

C Compiler:

Improve global optimizations.
Improve generated code for conversion of 64-bit values to IEEE floating point.
Improve code generated for _Decimal(n,p) multiply.
- Support for -fsigned_char/-funsigned_char options, indicating the default "signedness" of the (char) type.
The -M option output was changed to be consistent with among all the Dignus tools.
Improve compile time performance.

PLINK utility:

Improve PLINK's compatibility with newer versions of the IBM EDCPRLK pre-linker.

D2S utility:

No longer emit #pragma anontstruct if -xanon is specified.
New option -style=nestorg that is similar to -style=next, but with a simpler treatment for ORG statements.
New -xcom/-com=single/-com=con/-com=all to control generation of comments taken from ASM source.
Support for multi-line comments and comments that aren't attached to a particular symbol.

C library:

Support @@SARNTI (stand alone re-entrant initialization) in 64-bit environments.

What's new in Release 1.97 of Systems/C

C Compiler:

New optimization calculations to extend the range of global optimzations and greatly improve optimized code.
The -march=z option was extended to become -march=zN, where N indicates the revision number of the z/Architecture Principles of Operation.
New -mhfp-mult-add, -mgeneral-instructions-extension, -mimmediate-extension options to control the use of the z/Architecture instructions available with those hardware extensions.
Loop optimizations were significantly enhanced to provide better loop-invariant and strength-reduction optimizations.
Generation of switch-table selection code was considerably enhanced to employ several methods, or combinations of those methods, including binary search, bit-testing, etc..
The SAS/C alternate values for special characters (left-brace, right-brace, not-sign and backslash) are now recognized on EBCDIC hosts.
The listing was expanded to provide a STATIC MAP section detailing the layout of the static data area.
The listing was expanded to provide a cross-reference.

C library:

New __userid() function that returns the NUL-terminated usedr id value (or NULL if it can't be found)

What's new in Release 1.96 of Systems/C

C compiler:

New -fhlasm option that indicates the IBM HLASM assembler will be used to assemble the source. This defeats some language features that require Systems/ASM-supported extensions.
Added support for -fll=cba (CALL BACK ANY) mode in IBM DLL compatibility.
New -funsigned_bitfields and -fsigned_bitfields to adjust the default signedness of bitfields in the generated code.
New -ffnio (function name in object) that stores the function name in the generated code just prior to the function. This is helpful in IBM compatibility mode when reading LE dumps.
New -fhide_skipped option which removes listing lines which are skipped during preprocessing.
New support for Floating Point Decimal types, including _Decimal32, _Decimal64 and _Decimal128.
Improved constant folding of IEEE and HFP floating-pt values in the optimizer.
Improved generated code for 64-bit shifts.
Generated code takes advantage of NIHH/NIHL/NILH/NILL and OIHH/ OIHL/OILH/OILL when z/Architecture instructions are allowed.
Improved compilation time.
Support for -On option, where n is a the value 0, 1, 2 or 3 that controls the application of optimizations.
New optimizer with many more global optimizations. These new optimizations demonstrate a significant improvement in the generated code. Some test examples showed improvement between 22% - 38% in runtime performance.

Utilities:

PLINK now passes a flag to the external mapping function indicating the symbol's type.
DCCPC: Added support for VSE translations of many EXEC CICS commands.
DCCPC: Added new -fnoedf option to set the common flag X'40' in the translated CICS output.

What's new in Release 1.95 of Systems/C

C Compiler:

Support specification of both -g and -O at the same time; generating some debugging information for optimized code.

Support for use with z/TPF, including z/TPF runtime components.

#pragma linkage(OS) implies __local for function pointers.

#pragma prolkey/epilkey now imply that the function cannot be inlined.

Support #pragma variable(name, RENT) and #pragma variable(name, NORENT)

Inline memset() will now use up to 7 XC instructions when a constant length is given and the value to store is 0x00.

Improvements in compile-time memory use when -O is specified.

Support Microsoft/GNU __COUNTER__ predefined macro extension.

Improvements to the inlining optimization to avoid creating very large executables.

New -march=z option that indicates that the z/Architecture instructions are available even in 31-bit mode.

New -mlp64 option which replaces the now deprecated -march=zarch option. -mlp64 indicates 64-bit long and 64-bit pointer.

Improved global constant propagation and value numbering algorithms in the optimizer.

New -mdistinct-operands option that indicates the compiler can generate code that uses the z196 distinct-operands-facility instructions.

New -mload-store-on-condition option that indicates the compiler can generate code that uses the z196 load-store-on-condition--facility instructions.

New -mmvcle option that indicates the compiler can use the mvcle/clcle instructions for large memory moves/compares.

Improved code generation for builtin strlen().

Builtin strcat()/memchr() will use string instructions if they are allowed.

Block move and builtin memcpy() with a constant length was improved to use a simple MVC loop with only one EXecuted MVC if needed, or the MVCLE instruction if allowed.

Improvements to recognize common BCTR+LTR+BNZ sequences and fold those into a single BCT instruction.

Improvements to generated code when converting _Decimal() values to (long long).

If the length parameter to builtin memcpy() is a (non-constant) unsigned character, the compiler will generate a single EXecuted MVC instruction as the length cannot be longer than 256.

Support GCC's __local__ syntax for declaring labels local to a scope.

Support GCC's __attribute__ syntax for decorating types, variables and functions with various attributes.

Support GCC's conditional expression extension, allowing for empty "true" operands in conditional expressions.

Enhancmetnts to the C preprocessor to support the defined() test appearing in replacement text on #if lines.

New -w option to suppress all warnings.

Full support for C99-style initializers.

Support GCC's feature of empty brace initializers with a new warning (#2405).

Support C99 compound literals.

Support GCC's 0b (binary) constant extension.

Support C99's flexible array member (VLAs).

Removed limit on number of -I options that can be specified.

The listing has been improved to provide the options in sorted order, as well as provide the command line specification.

New -femitdefs option that adds #define values in -E output.

Support GCC's extension to allow structures or unions with no members.

Support GCC-style variadic macros as well as the ANSI C99 style.

Integrated support for building for z/TPF, as well as TPF 4.1 with the same compiler.

Support GCC's __asm__("string") syntax applied to externally visible variables and functions; to basically be the same as #pragma map.

New #pragma pack() facility for providing IBM LE-style, Systems/C-style or GCC-style structure member packing.

The deprecated -fansi_bitfield option has been replaced with be -fansi_bitfield_packing and -fnonint_bitfield to separate the idea of ANSI bitfield offset allocation from ANSI bitfield types.

#pragma export and _Export now also apply to ELF visibility flags (to support #pragma export in z/TPF.)

Support the \a (alert) escape sequence in string constants.

Support A', E', A" and E" as valid tokens in the C preprocessor.

Support #pragma options(inline)/#pragma options(noinline).

New -fwchar_ucs option that causes wide-character string or character constants to be encoded in UCS-2/UCS-4 regardless of the -fasciiout setting.

New -fdfe/-fnodfe options to enable or disable elimination of unreferenced (dead) function.

Improved code generation for bitfields, _Decimal data, 64/31-bit conversions, register usage, etc...

DB2PPC utility:

Improve EXEC SQL INCLUDE to provide search command line options.

DCCPC utility:

Improved support for VSE with new -fvse option that enables VSE-specific translataions

Added support for CICS TS 3.1 CHANNEL/CONTAINER/READ/READNEXT/READPRV commands.

Added support for EXCI mode with new -fexci option.

D2S utility:

Added new #define name_length output to represent the length attribute of EQU symbols when -equ=def is specified.

PLINK utility:

New -pu option, to provide mapping rules that map the the name to a, possibly shortened, upper-case variant of the given name.

Support the ORDER card to place specific CSECTs at the beginning of the output.

C library:

ddnfind() now supported in 64-bit environments

new osddinfo() function to return information about a DD name.

Added C99 functions frexpf(), frexpl(), ldexpf() and ldexpl().

New include files <netinet/tcp.h>

New Keyed File (VSAM) I/O support functions, kopen(), ksearch(), kretrv(), kgetpos(), ksetpos() and kclose().

New __setsockparm() parameter, _SP_SUBTASK, to set the SUBTASK value on the EZASMI INITAPI start.

Improved initial buffer size allocation for DD/DSN-style files in non-text I/o situations to use the file's BLKSIZE.

Added memmem() function.

What's new in Release 1.90 of Systems/C

C Compiler:

Improved codegeneration & optimization.
In IBM compatibility mode, DCC now generates ISD or DWARF debugging information when the -g (debugging) option is specified.
Options listed in compiler listing are sorted in alphabetical order.
Support for z/TPF compilations.
Support more GCC compatibility features, including 'deprecated' and 'packed' attributes, __typeof__(),
New -fstructmaphex option for displaying offsets/lengths in structure map listing in hex.
Default section name length expanded to 1023.
Improved generated code for 1-byte signed __asmval() values.
Support builtin strpbrk() function
Compiler now defines __370__ as the IBM compiler does
Added compiler support for include files in .zip files, as in -Imyheaders.zip
Many codegeneration/optionization improvments, some of which include:
- Use TML instruction for situations like:
  if(x & val) {
  (where 'val' is sufficiently sized to take advantage of the instructions.)
- Improved the codegenerator to eliminate LTRs after certain arithmetic and logical operations.
- Improved value-numbering algorithms for local/block optimizations.
- Improved memory usage in register allocator on many large examples.
- Improvide in-liner algorithms and heuristics.
Support new -fenum= option and #pragma enum() pragma in IBM compatibility mode.
Support -@filename processing discovered when reading a -@file file (i.e. recursive -@).
Support z/OS 1.9 and 1.10 compatibility in IBM compat mode.
Correct result type of __asmval() expressions to be unsigned. Also, support 8-byte __asmval() expressions in a 32-bit compilation. In this situation, the resulting type of the __asmval() is unsigned-long-long instread of unsigned-long.
New message #2147 - no function prototype given for `XXX' disabled by default.
#pragma linkage specifications are not case-specific.
Add support for #pragma runopts(POSIX(ON/OFF)) in IBM compatibility mode.




PLINK:

PLINK properly sets linker-generated date/timestamp in TSO 
     transmit files.

PLINK's definitions of pre-loaded IBM library list updated.

New -fdaranycase option that causes autocall resolution from 
     non-DRANLIB'd DAR archives to be case insensitive.

Support -fmapfunc=name and -fmapdata=value options to
     offer user-provided ESD mapping functions.

Support comma as a token separator in IMPORT and
     RENAME statements when the values are not quoted (per newer
     IBM binder documentation.)

Add new -I option which adds search paths for INCLUDE
     files (in a manner analagous for -S used for AUTOCALL
     files.)

Improvements to the -bvse output to generate VSE "unloaded"
     phases.

Support -@filename processing discovered when
     reading a -@file file (i.e. recursive -@).


  D2S:


Add -prefixmap=old,new option to manage common prefixes on 
     DSECT member names.

No longer generate useless filler structs for CSECTs.

@, $, and # in 
      DSECT names are now translated to a, 
      n, and d
     respectively for the struct name, just like member names.

Add support for -char=[char,signed,unsigned] to control the 
     type generated for "DS C" members.

Replace '*' with '.' in ASM comments so that a C-style comment embedded in the ASM source won't break our output.
Correct handling of HLASM V1R5-compatible ADATA records.

DCCPC:

Add support for CICS TS 3.1 CHANNEL/CONTAINER commands.
Added support for EXCI mode (-fexci), which is used to translate a special version of the EXEC CICS LINK command.
Improved support for "browse mode" EXEC CICS INQUIRE commands.
Accept lowercase "exec cics" commands.
Support DSIDERR, END, TASKIDERR conditions for use with IGNORE/HANDLE CONDITION.
Add support for browse mode (START/NEXT/END keywords) for several INQUIRE commands
Added -fquiet and -fmsgtostdout.
Corrected the translations for INQUIRE TERMINAL() and INQUIRE NETNAME() commands in some situations.
Correctly recognize commas in assembler mode.

DB2PPC:

New DB2 preprocessor utility (DB2PPC) that processes DB2 commands, creating the preprocessed output and DBRM file.

LIBC:

New library functions, ctermid(), fmtcheck(), err()- and warn()- family, fmtmsg().
New multi-byte character functions: mblen(), mbrlen(), mbrtowc(), mbsinit(), mbsnrtowcs(), mbsrtowcs(), mbstowcs(), mbtowc(), wcrtomb(), wcsftime(), wcsnrtombs(), wcsrtombs(), wcstod(), wcstof(), wcstold(), wctomb()
New btowc() and wctob() POSIX functions.
Added <timeconv.h> include file, and related time_to_...() and ..._to_time() functions.
z/OS 1.8 and later trash the upper bits of the 64-bit R15 after GETMAIN (previous versions did not.) We used an LTGR instruction to check the GETMAIN return code and need to be mindful of this change in z/OS behavior.
Corrected strtold() to round to the nearest value instead of rounding to zero.

What is new in Release 1.85 of Systems/C

C Compiler:

Improved support for volatile data references, forcing them to be in memory and loaded/stored at each reference.
Improved compile time for large programs with large amounts of constant data.
New optimizations, including improved constant propogation and basic-block related optimizations.
Improved code generation, particularly when -march=esa390 or later is enabled.
Added support for IBM-style "events file" for use with the IBM TPF Toolkit.
Support for XPLINK linkages in IBM compatibility mode.
Support for later versions of IBM compatibility, including support through z/OS 1.7. This also includes newer IBM predefined macros.
Support Microsoft's __int8/__int16/__int32 and __int64 builtin types.

C Library:

@crt0 labels were renamed to be upper-case @CRT0. Lower-case aliases are provided for compatibility.
New errno values - ENOATTR, EDOOFUS, EBADMSG, EMULTIHOP, ENOLINK and EPROTO.

PLINK enhanced with -singleesd option (for the BS2000 linker) which guarantees only one ESD entry per ESD card.
PLINK now has an AC=n option used to set the authorization code in the PDS directory when generating TSO XMIT files.

What is new in Release 1.80 of Systems/C

Library

Support double-quotes in TSO and BATCH parm strings. The double-quotes allow for arbitrary strings to become an argv value and can include commas, as in PARM='"my,parm1",parm2'. In this case, argv[] would be { "my,parm1", "parm2" }.
Support for setting the argv delimiter via the new __argvc variable.
The scanf() family of functions now supports more C99-specified conversions, including the hh, j, ll, t and z modifiers. Also, many of the older compatibility conversions were removed. %E and %F now follow ANSI rules.
New popen()/pclose() functions for POSIX environments.

Utilities

PLINK now has a new -uu option. -uu operates exactly as -u, except that the return code will be unaffected by unresolved references.
PLINK will now generate sequence numbers on XSD or ESD-style output files. For some situations (but not all) the binder will include those sequence numbers in error messages, which can help in diagnosing binder (or PLINK) issues.
PLINK now has a -bvse option, which causes it to do full binding of the modules and produce a VSE "unloaded" PHASE file.
D2S has a new -dint option that treats 'D'-type fields as 8-byte integers instead of 8-byte floating point values.

C Compiler

New -fpack option to specify per-compilation structure member alignment.
Support for reading #include files from PDS members under OpenEdition.
Support for -frsa=NNN to provide the user a mechanism for defining the per-function register-save-area. Support for 128-bit (long double) floating point data type.
#pragma linkage values can now also apply to function typedefs, to support building function pointers that have different linkage conventions.
New -march=esa390 option; causes the compiler to use halfword immediate instructions, use string instructions and to assume 16 floating-pt registers. Enabled by default with -flinux is specified.
New state-of-the-art register allocation algorithms, greatly improving register use in the generated code.

What is new in Release 1.76 of Systems/C

C Compiler:

New message #2399 - non-constant member-designator in offsetof

PLINK:

New -showctl option that causes PLINK to display any control cards encountered during linking.

Library:

When the AMODE is 24, POSIX functions now graciously fail instead of ABENDing.
OPEN processing dynamically moved below-the-line for MSP support.
O_CREAT and O_EXCL flags in open(2) now supported for dynamically allocating MVS file-system files. Also - several new attribute values pertinent to specifying sizes and file attributes when creating files.
New binary tree creation/deletion/manipulation routines, tdelete(3), tfind(3), tsearch(3), twalk(3).
New BSD 4.3 compatibility functions, lsearch(3) and lfind(3) for linear searching.
New XPG4.2 hcreate(3)/hsearch(3) and hdestroy(3) hash-table manipulation functions.
New fldata() function which returns low-level information for open file descriptors.
New fseeko() and ftello() functions from Version 2 of the Single UNIX Specification (SUSv2).
New re-entrant time routines - asctime_r(), ctime_r(), gmtime_r() and localtime_r(). Also added timegm(). (See ctime(3).)
New POSIX.2 functions glob(3) and fnmatch(3).
MVS 3.8 objects were moved to their own directories/PDSs, so linking for MVS 3.8 simply becomes a matter of choosing the right PDS concatenation or PLINK search order.

What is new in Release 1.75 of Systems/C

C Compiler:

If -march=zarch is specified, the compiler will predefine the macro _LP64 to 1; otherwise, it defines the _ILP32 macro to 1.
New -fcodepage500 option on EBCDIC platforms to support source in IBM-500 codepage. New #pragma filetag supported on EBCDIC platforms to indicate the source code code-page.
Constant modulus by power-of-two (either signed or unsigned) is now optimized in the code-generator.
When -fc370 is not specified, the compiler recognizes #pragma pack(pop) as a equivalent to #pragma pack(reset).
Compiler now supports the IBM/Microsoft __FUNCTION__ predefined identifier. This is replaced with a string constant that is the name of the current function.
Compiler was changed to define the name of the function (in Systems/C mode) in a compiler-generated character string that preceeds the function prologue. Then, a new LNAMEADDR= parameter was added to DCCPRLG which provides the label for this string. This allows the string to be as long as necessary, and contain any required bytes, and also be properly demangled in the C++ case. This significantly enhances the __traceback function. Improved code generation for passing constant parameters to functions, particularly in Linux mode.
New -fc370=z1r2 option, to indicate compatibility with the IBM V1R2 z/OS compiler.
New messages # 2356, #2357, #2097, #2098, #2168, #2169, #2371.
Improvements in Linux compatibilty to be compatible with GCC 3.2 and later.

Library:

The getprogname() and setprogname() functions were added.
Support for new __jobname(), __stepname(), __procname(), __get_cpuid(), __isPosixOn() functions.
Support for POSIX/UNIX functions: _exit(), getpid(), getppid(), pipe(), getcwd(), getlogin(), uname(), fork(), exec*(), execve(), wait(), waitpid(), wait3(), getgid(), setgid(), getegid(), setegid(), getuid(), setuid(), geteuid(), seteuid(), setreuid(), setregid(), getpwent(), setpassent(), setpwent(), endpwent(), getpwnam(), getpwuid(), getgrent(), setgroupent(), setgrent(), endgrent(), getgrnam(), getgruid(), getpid(), getppidd(), getcwd(), getwd(), getgroups(), setgroups(), getpriority(), setpriority(), nice(), getpgrp(), setpgrp(), getpgid(), setpgid(), getsid(), setsid(), getrusage(), kill(), times(), __passwd(), realpath(), mmap(), mprotect(), msync(), munmap(), ftok(), shmat(), shmdt(), shmget(), shmctl(), msgget(), msgctl(), msgrcv(), msgsnd(), system(), semctl(), semop(), semget(), ptsname(), grantpt(), unlockpt(), tcsendbreak(), tcdrain(), tcflow(), tcflush(), tcgetattr(), tcsetattr(), tcgetpgrp(), tcsetpgrp(), cfgetispeed(), cfsetispeed(), cfgetospeed(), cfsetospeed(), cfsetspeed(), and cfmakeraw().
Support for HFS file system I/O functions: open(), read(), write(), close(), lseek(), unlink(), rename(), isatty(), fcntl(), rmdir(), mkdir(), access(), fsync(), umask(), fstat(), stat(), lstat(), link(), symlink(), readlink(), mknod(), mkfifo(), opendir(), readdir(), rewinddir(), closedir(), chown(), fchown(), lchown(), chmod(), fchmod(), chdir(), fchdir(), chroot(), truncate(), ftruncate(), pathconf(), fpathconf() utime(), utimes() and futimes().
Support for executing 31-bit and 64-bit Systems/C programs under OpenEdition (via execve().)
New C99 <inttypes.h> file, and the imaxabs(), imaxdiv(), stroimax(), stroumax(), wcstoimax() and wcstoumax() functions from section 7.8.2 of the ANSI C99 standard.
New C99 <wctype.h> and <wchar.h> and the associated functions from there (section 7.25 of the C99 standard), including iswalnum(), iswalpha(), iswascii(), iswblank(), iswcntrl(), iswdigit(), iswgraph(), iswhexnumber(), iswideogram(), iswlower(), iswnumber(), iswphonogram(), iswprint(), iswpunct(), iswrune(), iswspace(), iswspecial(), iswupper(), iswxdigit(), towlower(), towupper(), wctrans(), towctrans(), wcwidth(), wctype(), iswctype(), wcstoul(), wcstol(), wcstoll() and wcstoull().
The default buffer size for STDIO files was increased from 1024 bytes to 8192 bytes, to improve I/O performance. //HFS:-style files use the fstat() function to determine optimimum buffer sizes, particularly in network situations.
optimizations for multiple SEEK_ENDs on the same read-only non-//HFS: files. Also, further optimizations to avoid redundant NOTEs.
New signal-related external variables, sys_signame[], sys_siglist[] and sys_nsig;
New single/linked/circular list and queue functions (see QUEUE(3) in the Systems/C library manual).
The library has been updated to be "thread-aware", although a general pthreads library has not been implemented.
The resolver routines now support the DOMAIN statement as well as DOMAINORIGIN in the TCPIP.DATA file.
New sysconf() function to provide partial POSIX compatibility in non-POSIX environments.
New TCP/IP related functions - getipnodebyname(), getipnodebyaddr(), and freehostent().
The ANSI C99 standard strtoll(), strtoull() are provided (the previous non-standard strtoq() and strtouq() are still provided for backwards compatibility.)
New wide-char and string functions: strcasestr(), wcscat(), wcslcpy(), wcsrchr(), wmemcpy(), strlcat(), wcschr(), wcslen(), wcsspn(), wmemmove(), strlcpy(), wcscmp(), wcsncat(), wcsstr(), wmemset(), strmode(), wcscpy(), wcsncmp(), wcswidth(), strnstr(), wcscspn(), wcsncpy(), wmemchr(), strsignal(), wcslcat(), wcspbrk() and wmemcmp().
New ctype related functions: digittoint(), ishexnumber(), isideogram(), isnumber(), isphonogram(), isrune() and isspecial().
New ddnfind()/ddnnext() functions which determine the DSN associated with a DD.

Utilities:

PLINK is now able to directly produce a load module on cross-platform hosts, using the TSO transmit format.
New PLINK option -DD=name=template which specifies a search pattern template to use processing an INCLUDE card with a DDNAME(MEM)-style name.
PLINK now supports pre-linking/producing IBM DLLs in IBM compatibility mode.
New -elc/-xelc options in D2S to enable lower-casing of EQU/SECTION names.
New -bfstruct/-xbfstruct option in D2S to control the generation of bitfield-populated structures for odd-sized data (e.g. XL3 will be translated to struct { unsigned BF : 24; }.
D2S now generates comments to show the original type.

What is new in Release 1.70 of Systems/C

Library:

A rename() function has been added which can rename DSN-style files.
New ASCII<->EBCDIC translation functions (see the ATOE(3) section in the Systems/C C library manual.)

Utilities:

PLINK has been rewritten to do more "linking" on its own. In particular, the GOFF2XSD utility is now no longer required, as PLINK will accept GOFF input.
Also - by default, PLINK will process all defined "dummy" sections creating a PRV table. This allows reentrant programs to be generated for systems the use older versions of the IBM linker that don't support a Psuedo Register Vector larger than 4K (e.g. VSE and older CMS.)
PLINK will also eliminate duplicately defined weak functions in Systems/C programs (mostly used for C++ templates.) PLINK will remove the bytes from the generated TXT sections, signifcantly reducing the resulting load-module size.
Furthermore, PLINK (and DAR/DRANLIB) can now handle IBM C objects. When an IBM compatible object is discovered, PLINK mimics the IBM Pre-Linker. Thus, all IBM pre-linking functions can be accomplished on a cross-platform host.

Compiler:

Support for defining #pragma weak functions in Systems/C mode. (The new version of PLINK will automatically remove duplicate weak definitions and eliminate the duplicate code from the resulting load module.) This is also employed for proper handling of C++ automatic template instantiation.
New -finstrument_functions option. For Linux (-flinux), this decorates function entry/exit points compatibly with the normal Linux utilities. In Systems/C mode, this causes macros to be generated a function entry/exit points.
New -fstructmap/-fno_structmap option to enable/disable the "structure map" portion of the listing.
New -fasmcomm option in the C compiler, which describes how the original C lines should appear in the generated assembler output.
New -fasmlnno option in the C compiler. This causes the codegenerator to generate comments detailing the line number in the generated assembler source.
New -fflex, -fmin_lm_reg=X and -fmin_stm_reg=X options in the C compiler. These control peepholing of consecutive Load/STore instructions. On the FLEX/ES(tm) system, LM and STM instructions can perform slower than the sequence of L/ST instructions.
New -fat and -fsascdigraphs in the C compiler. -fat enables support for the SAS/C(tm) @-operator; and -fsascdigraphs enables support for SAS/C(tm)-style digraphs.
New -fopts[=macro] and -fendmacro[=text] in the C compilers.
New -fdll and -fexportall in C compiler (in IBM mode)
New -fdollar option in the C compiler - allows '$' in identifiers.
AR moves of structures increased from 256 to 4095 bytes.
Improved code generation in testing "flag" bytes, unsigned char operations and unsigned int operations.
Reduction in SAR/SAC/CPYA instructions for __far/__based pointers.
Code generation improvements to add new options to the prologue/epilogue to not save/restore as many registers (or, possibly no saving/restoring of registers.) Also, improvements to reduce the local DSA requirements.
Improvements in constant multiplication evaluation to avoid unnecessary register copies.
Linux prologue/epilogue improvements.

What is new in Release 1.65 of Systems/C

Compiler:

Support for #pragma ident and #ident which places text characters into the generated object.
-O option to engage optimizations.
C compiler now tokenizes #pragma statements, to allow comments within the pragmas.
Support for C99 inline functions, including improved inliner for C and C++ inline functions. The compilers will now inline functions explicitly via the "inline" keyword, or based on command-line heuristics.
Improved register tracking in the code generator, providing faster runtime performance.
C99 compliance for main() functions that have no explicit return value. The compiler will provide a default value of 0 when appropriate.

Library:

New <bitstring.h> header file that defines bitstring manipulation macros.

What is new in Release 1.60 of Systems/C

Compiler:

-fquiet option, which suppresses the banner messages
-fignore_case, causes the compiler to no treat upper-case letters "special" in generated HLASM compatible source
-march=zarch option:
True 64-bit compiler using "Grande" instructions
#pragma linkage(identifier,ALIGN4) to allow for older 4-byte aligned (ESA) parameter lists.
extern "align4" linkage for C++
Uses all 16 floating point registers for improved math code.
Takes advantage immediate instructions
New -famode=any to support using Grande instructions and 64-bit integers/pointers when the execution AMODE may not be 64-bits.
__ptr31/__ptr64 pointer type qualifiers, for specifically declaring 32 (31) bit or 64 bit pointers, no matter what type of code is being compiled.
-fc99 option to optionally enabled ANSI C99 features:
enables _Bool data type and recognition of C99 keywords. declarations anywhere - including declarations in for-statements
#pragma STDC FENV_ACCESS which suppresses floating pointer optimizations
long-long integer constants supported following C99 rules (no warnings)
switch() statements extended to support long-long integral values when -fc99 not enabled
-fc370 now supports the following IBM builtin functions: alloca, tsched, cs, cds. Thus, all of the documented IBM C builtin functions are now supported.
#pragma csect ({CODE,STATIC,TEST), "name") for setting sname.
New complete ANSI C 99 C preprocessor with support for c99 digraphs, variadic macros, _Pragma, guarded headers not re-read, etc...
#pragma export supported in IBM compatiblity mode
C++ compiler now includes a structure map in the listing file.
Enhanced debugging information support for C and C++ compilers in -flinux mode. C and C++ code compiled with -flinux is now directly debuggable by the Linux debugger gdb.
Support for -fpic and -fPIC for Linux/390 and z/Linux.
Continued "general" code-generation improvements.
z/Linux and OS/390 USS (OpenEdition) are now supported platforms.

PLINK:

-r (re-plink) option, allows for PLINK produced objects to be passed through PLINK again.
-allow_ref option, allows for PLINK to ignore particular unresolved references, which will be resolved later.

C Library:

New ANSI C99 header files < stdbool.h>, < stdint.h >, < iso646.h >.
New z/Architecture rent and non-rent libraries, allowing stack/heap/PRV above-the-bar, user-configurable.
Non z/Architecture library now supports stack/heap/PRV above-the-line, user-configurable.
New __malloc31/__free31 and __malloc24/__free24 functions providing memory allocation management within a known addressing range.
< stdarg.h > now defines va_copy as specified in the ANSI C99 standard. Furthemore, stdarg related macros now support passing integral types smaller than `int'-sized values, which is a common extension on other platforms.
Runtime support for executing C programs on MVS 3.8.
The library now honors the timezone offset (CVTTZ) from the operating system, and assumes the clock is set to UTC. The TZ environment variable will be appropriately set at program startup for non-DCALL programs.

DCCPC:

New CICS preprocessor for C. Supports CICS/ESA V4 R1 and CICS TS V2.

D2S:

New DSECT to C structure converter. Supports 64-bit data types, and takes advantage of Systems/C extensions to produce a much more readible C structure definition.

Documentation:

Documentation PDFs now include a table of context with links into the documentation sections, and links within the document to interesting sections.

What is new in Release 1.50 of Systems/C

Compiler:

Support for __attribute__((__constructor__)) and __attribute__((__destructor__)) in function declarations. The __attribute__((__constructor__)) specification causes the declared C function to be marked as a "constructor" function - i.e. one that is executed by the C library before main() is invoked.
Similarly, __attribute__((__destructor__)) causes the function to be executed after main() returns, or by the exit() function.
Improved code for comparisons of double constants to double values.
Improved code generation for testing bit-field members of structures.
New #pragma weak for indicating weak function and non-rent data references.

Library:

_exit() (which is invoked by exit()) no longer does a direct SVC 3 to end the program. Instead, Systems/C programs now properly release resources and simply return to the caller.
The timezone() function has been added to the C library.
The cabs() and hypot() functions have been added to the C math library.

What is new in Release 1.41 of Systems/C

Compiler:

Improved code generation
Improved diagnostics, particularly warnings which denote non-ANSI conformance.
Support for #pragma export in LE370 mode (requires the Systems/ASM assembler.)
Compiler will no longer create an empty @DATA section, which eliminates warnings from the VSE linker.
Unsigned long-long addition was changed to use the AL instruction instead of A instruction to avoid any overflow exceptions.

Pre-Linker (PLINK):

PLINK error/warning messages can now be directed to STDOUT only, which can be helpful on Windows platforms.
Improved PLINK messages when duplicate definitions are encountered.

GOFF2XSD utility:

GOFF2XSD now properly handles GOFF logical records which span multiple records.

Systems/C C Library:

New __setsockparm() function which allows the program to define IBM TCP/IP initialization values.
A replacement malloc() implementation was implemented which should avoid the issues of over-allocating space for large allocations. This should result in a much smaller memory "footprint" for programs.
Autocall issues with some of the math library functions were corrected.

Documentation:

The rather large single book is now broken up into 3 smaller books, The Systems/C C Compiler Manual, The Systems/C C Library Manual and The Systems/C Utilities Manual.

What is new in Release 1.40 of Systems/C

Compiler:

New #pragma pack to specify maximum structure element alignment.
New -fpatch and related options for providing patch areas in the generated code.
Improved code generation, resulting in anywhere from 10% to greater than 27% runtime improvement, depending on the application. The compiler typically generates between 10% to 6% fewer instructions.
Improved compile time (up to 20% improvement in some cases.)
Corrections for ANSI conformance.
Improved where line numbers are located for debugging information.
The split-bar (in EBCDIC) is now also recognized as the |-bar, allow compilation of source which contain that character (e.g. codepage 037.)
Improvements to __far pointer support.
Improvements to Linux/390 debugging information.
Linux/390 code generator now supports gcc's __builtin_alloca().

Library:

DCALL now correctly terminates the socket API when the DCALL'd function returns (iff sockets were used.)
fcntl() now supports the F_GETFL (get flags) command.
fdopen() now supported.

Utilities:

DAR,DRANLIB now operate better on MVS and handle PDS file names better.
New GOFF2XSD utility which converts HLASM GOFF to XSD format which PLINK can handle.

What's new in Verion 1.31 of Systems/C

Improved code generation in the Systems/C Compiler.
Systems/C compiler augmented with many more diagnostics.
Support for #pragma eject, #pragma page(n) and #pragma page(), #pragma pagesize(n) and the new -fpagesize=n option which allow the user to tailor the compiler listing.
#pragma options() is now implemented in the Systems/C compile, and supports some of the options available on the command line.
New option (-fmrc) which will cause the compiler to return mainframe-style return codes on non-mainframe platforms.
New __far pointer flexibility, including options to change the alignment of __far pointers and to specify the order of ALET and OFFSET.
Calling a #pragma linkage-OS function now saves and restores the first three words of the local save area.
When -g is specified in combination with -flinux, the Systems/C compiler will now include gcc-compatible debugging information in the generated assembler source. This allows the programmer to use the Linux debugger, gdb.
The Systems/C C library now supports the O_APPEND flag in open() for sequential files.
getcliendid(), selectex(), givesocket() and takesocket() are now provided in the library - for implementing multi-tasking socket programs.
The Direct-CALL support in the Systems/C library is now extended to allow for the creation and re-use of Systems/C environment, saving considerable run-time overhead.
PLINK now has improved messages, particularly when an autocalled reference cannot be resolved; or when duplicate definitions are discovered. PLINK messages also not include the name of the object where a symbol is found.

What's new in Verion 1.30 of Systems/C

The Systems/C library now supports lseek() to any offset within a text or binary read-only file. lseek() to offset 0 is also supported for both input and output files.
Improved the memcmp() builtin function to use a CLC when the operand length is constant and less than 256.
PLINK now supports the ARLIBRARY and INSERT control cards.
Support for UNIX ar-style archive libraries and linking with them, the DAR and DRANLIB utilities as well as improvements to PLINK. Allowing for resolving (AUTOCALLING) long-name references.
New -fnosname option which leaves section name selection to PLINK. PLINK can now automatically give each Systems/C object a unique name for the load module.
long-long multiply is now generated in-line instead of an out-of-line function call.
Support for Linux/390 (#include_next, #warning, several GCC builtins, compound statements, as style generated assembly, Linux/390 entry/exit, etc...)
Support for IEEE format floating point constants, compiler constant folding and code generation.
PLINK now supports the -@filename option on command lines for long command lines on Windows and OS/390
__STDC__ is now defined.
support user-selectable wide-character size, either 2 or 4 bytes.
support for Direct-CALL in the Systems/C library; which allows a Systems/C environment to be directly instantiated.
Improved parser recovery from several error situations
__register variables are now supported at file scope; which globally reserves a register.
New character escape sequences for in-line assembler source to support single/double quotes, provide the CSECT name.
More warning/error messages for improved programmer productivity
Improved matching of C source to generated assembler source
Support for ASCII character and string constants with the new -fasciiout option.
Support for 64-bit integer constants
Improved code generation
Improved __far pointer support.

Learn more about Systems/C here.

Learn more about Dignus, LLC here.