What's changed

The Dinkum Compleat Libraries is a new product that replaces our long popular Dinkum Unabridged Library V4.02, which we continue to maintain for our existing supported customers. Here is a summary of what's changed between products.

Major additions

• added hardware header for TR18015
• added iohw.h header for TR18037
• added stdfix.h header for TR18037
• added numerous headers, special math functions, overloads for TR19768
• added uchar.h header for TR19769
• added Corex libraries (allocators, code conversions, threads)

Major changes to the Standard C++ Library

• algorithm: added bounding for random_shuffle with function
• algorithm: added is_heap, is_sorted
• algorithm: changed int to size_t in _Median
• algorithm: corrected _Mid1 in random search_n
• algorithm: corrected prev_permutation for equal elements
• algorithm: corrected random_shuffle for large indices
• algorithm: improved search_n
• algorithm: reduced temp buffer request for stable_sort
• algorithm: sped up _Debug_heap
• algorithm: tidied count logic in stable_partition
• bitset: added to_string overloads
• bitset: dropped redundant tests from at
• bitset: improved unsigned long constructor
• bitset: moved _Trim outside loop for left shift
• deque: fixed overlap for insert
• deque: improved pop error checking
• deque: protected against self swap
• deque: simplified _Off calculation in iterators
• exception: added throw() on destructor
• exception: added throw() on uncaught_exception
• fstream: corrected close logic
• functional: added identity, project1st, project2nd, select1st, select2nd
• hash_map: fixed overlap for insert
• hash_set: fixed overlap for insert
• ios.cpp: deferred _Ploc initialization for robustness
• ios.cpp: dropped negative index code in _Findarr
• iosfwd: changed streamoff to _Longlong
• iostream.cpp: protected flush calls
• istream: changed seekg to clear eofbit on success
• list: added destroy counter to _Buynode with pointer constructors
• list: added knot check to splice
• list: changed remove to store test value
• list: corrected _LOCK_DEBUG locks
• list: corrected _Splice calls in sort
• list: corrected erase all return
• list: corrected iterator invalidation for sort
• list: eliminated orphaning of end() iterators
• list: protected against self swap
• locale0.cpp: added _Locinfo(const string&)
• map: added extra erase for _HAS_STRICT_CONFORMANCE
• memory: changed auto_ptr stored pointer from void* to _Ty*
• memory: changed autp_ptr_ref stored pointer from const void ** to void *
• memory: corrected auto_ptr looping
• memory: corrected auto_ptr_ref assignment
• memory: corrected handling of negative arg to get_temporary_buffer
• memory: delayed allocation in _Temp_buffer
• memory: simplified auto_ptr operator*/->
• newop.cpp: protected zero count
• newop2.cpp: protected zero count
• ostream: added _Addit to basic_ostream constructor
• ostream: added exception handling to _Osfx
• set: added extra erase for _HAS_STRICT_CONFORMANCE
• slist: added knot check to splice
• slist: changed remove to store test value
• slist: corrected _LOCK_DEBUG locks
• slist: corrected _Orphan_ptr
• slist: corrected _Splice calls in sort
• slist: corrected iterator invalidation for sort
• slist: corrected previous
• slist: corrected swap for different allocators
• slist: eliminated orphaning of end() iterators
• slist: protected against self swaps
• sstream: added ios::ate support
• sstream: corrected overflow logic for reallocated buffer
• stack: added const_/reference types
• utility: protected against self swaps
• valarray: added explicit default construction to valarray(size_t) in _Grow
• valarray: changed delete[] to delete
• valarray: changed valarray::resize to discard old contents
• valarray: corrected _Grow initialization for default constructors
• valarray: corrected valarray(size_t) initialization
• valarray: protected min/max from macros
• vector: added _Vector_const_iterator
• vector: added std:: to copy calls
• vector: added swap overload for vector elements
• vector: changed clear to not free storage
• vector: corrected _LOCK_DEBUG locks
• vector: corrected iterator+= for vector
• vector: fixed overlap for insert
• vector: protected against self swaps
• vector: reduced use of _Tmp in insert_n
• vector: simplified operator[] for iterator debugging performance
• xcomplex: corrected tanh
• xhash: added U to large decimal numbers
• xhash: corrected rehash for small bucket argument
• xhash: improved hash functions
• xhash: protected against self swaps
• xhash: simplified swap to fix end() bug
• xiosbase: added hexfloat
• xiosbase: corrected _USE
• xlocale: added -3 return from _Mbrtowc
• xlocale: added std:: to bad_cast
• xlocale: clobbered locale name in combine for strict conformance
• xlocale: corrected _Dowiden for null character
• xlocale: dropped name option for combine
• xlocale: weakened codecvt::do_length for strict conformance
• xlocmes: corrected do_open return
• xlocmon: added numpunct overrides for _HAS_STRICT_CONFORMANCE
• xlocmon: clipped negative _Fracdigits in _Putmfld
• xlocmon: made _Fracdigits an unsigned int in _Putmfld
• xlocmon: tidied group checking
• xlocnum: added hexfloat for C99
• xlocnum: added hexfloat input
• xlocnum: added numpunct overrides for _HAS_STRICT_CONFORMANCE
• xlocnum: changed ',' to '\0' for num_put grouping
• xlocnum: changed ::memchr to _CSTD memchr
• xlocnum: changed default _Kseparator to comma
• xlocnum: corrected I64 fixup for MS+Dinkum native
• xlocnum: corrected ll/l64 fixup in _Ifmt
• xlocnum: corrected strict _Grouping construction
• xlocnum: corrected trailing precision logic
• xlocnum: disabled group checking for _Getffld
• xlocnum: disabled group checking for empty grouping string
• xlocnum: skipped zero counting for Inf and zero
• xloctime: added -3 return from _Mbrtowc
• xloctime: corrected do_get_date eof logic
• xmemory: changed _Destroy_range arguments for nonstandard allocators
• xstring: added _Chassign
• xstring: added _String_const_iterator
• xstring: changed compare to always call traits::compare
• xstring: protected against self swaps
• xstring: tweaked single-element append
• xtree: corrected _Orphan_ptr
• xtree: corrected _Orphan_ptr declaration and call
• xtree: dropped nonmutable _Lmost/_Rmost/_Root
• xtree: eliminated orphaning of end() iterators
• xtree: made _Dec and _Inc more alike for invalid case
• xtree: protected against self swaps
• xutility: added _STD qualifier to mismatch calls in equal
• xutility: added non-const overloads to _Debug_lt*
• xutility: corrected bad function name for _Debug_order with pred
• xutility: eliminated orphaning of all iterators on assign
• xutility: made first argument a ref for default _Debug_pointer
• xutility: protected min/max from macros
• xutility: templatized compares, subtract for reverse_iterator
• yvals.h: added __STDC_VERSION__ definition
• yvals.h: dropped _MIN/_MAX
• yvals.h: increased _MAX_SIG_DIG to 48

Major changes to the Standard C Library:

• atoll.c: corrected cast
• fclose.c: added sanity check
• fdopen.c: corrected open logic
• feholdexcept.c: corrected FE_ALL_EXCEPT spelling in IA64
• fesetexceptflag.c: corrected p type for software-only version
• fesetround.c: changed movl to mov for Linux
• fgetwc.c: added -3 return from _Mbtowc
• fileno.c: corrected _FD_VALID call
• fwrite.c: added nleft logic
• nexttoward.c: corrected y Nan and x Inf handling
• nexttoward.c: dropped INEXACT raises
• nexttoward.c: added UNDERFLOW raise for all denorms
• nexttowardf.c: corrected y Nan and x Inf handling
• nexttowardf.c: dropped INEXACT raises
• nexttowardf.c: added UNDERFLOW raise for all denorms
• nexttowardl.c: corrected y Nan and x Inf handling
• nexttowardl.c: corrected decrementing _L1 for 16 byte
• nexttowardl.c: corrected Inf fixup for 8/16 byte
• nexttowardl.c: dropped INEXACT raises
• nexttowardl.c: added UNDERFLOW raise for all denorms
• remove.c: added _Rmdir for Posix
• setvbuf.c: changed size == 0 logic to better match FreeBSD
• tmpfile.c: corrected lock logic
• tmpnam.c: added remove
• xctype.c: made NUL a control char for Posix
• xdscale.c: corrected for large lexp, rounding denormalized
• xdscale.c: corrected round to even logic
• xdtest.c: added _DENORM for fpclassify
• xfdscale.c: corrected for large lexp, rounding denormalized
• xfdscale.c: corrected round to even logic
• xfdtest.c: added _DENORM for fpclassify
• xfenv0.c: corrected _FPP_NONE case
• xferaise.c: added _FE_INEXACT on over/underflow for IEC 60559
• xfrprep.c: enabled buffer allocation for line buffering
• xgentime.c: corrected default case
• xgentime.c: corrected isleapyr logic
• xgetfld.c: added k, K, r, and R qualifiers for fixed-point
• xgetfld.c: added AltiVec stuff
• xgetfloa.c: added pointer argument for AltiVec
• xgetint.c: added pointer argument for AltiVec
• xgetstr.c: added -3 return from _Mbtowc
• xgetstr.c: suppressed argument pickup for noconv
• xgetstr.c: corrected error return logic
• xgetstr.c: corrected range-detection logic
• xldscale.c: corrected _DLONG==1 denormalized
• xldscale.c: corrected for large lexp, rounding denormalized
• xldscale.c: corrected _DLONG==2 denormalized, round to even
• xldscale.c: corrected zero detection for _DLONG==1
• xldtest.c: added _DENORM for fpclassify
• xldtob.c: converted extra digit to fix rounding
• xldunsca.c: corrected 80-bit denormalized exponent
• xmbtowc.c: added -3 return
• xprintf.c: added _HAS_FIXED_POINT stuff for %v
• xprintf.c: corrected separator logic
• xprintf.c: added argno
• xputfld.c: added _HAS_FIXED_POINT k, K, r, and R specifiers
• xputfld.c: corrected bad code handling
• xputfld.c: added _HAS_FIXED_POINT v specifier
• xputfld.c: added argno
• xputstr.c: checked _Puttxt returns
• xscanf.c: changed nconv to detect unstored conversions
• xscanf.c: added secure/prec for _HAS_SECURE_LIB
• xscanf.c: corrected overflow test
• xscanf.c: corrected read failure logic
• xscanf.c: corrected matching failure returns
• xscanf.c: added AltiVec stuff
• xstoflt.c: reworked to avoid excess digits
• xstoflt.c: corrected value for bad parse
• xstoflt.c: added rollback logic to exponent parse
• xstopfx.c: corrected NaN) and hex parsing
• xstoxflt.c: reworked to avoid excess digits
• xstoxflt.c: corrected incomplete exponent parse
• xstoxflt.c: regularized predicate on fraction accumulator
• xstoxflt.c: reconciled maxsig compares with xstoflt.c
• xthread.c: corrected thread start logic
• xwgetfld.c: added k, K, r, and R qualifiers for fixed-point
• xwgetfld.c: added AltiVec stuff
• xwgetflo.c: added pointer argument for AltiVec
• xwgetint.c: added pointer argument for AltiVec
• xwgetstr.c: suppressed argument pickup for noconv
• xwgetstr.c: corrected error return logic
• xwgetstr.c: corrected range-detection logic
• xwldtob.c: converted extra digit to fix rounding
• xwprintf.c: corrected x.qual logic
• xwprintf.c: corrected separator logic
• xwprintf.c: added argno
• xwputfld.c: added k, K, r, and R qualifiers for fixed-point
• xwputfld.c: corrected bad code handling
• xwputfld.c: added _HAS_FIXED_POINT v specifier
• xwputfld.c: added argno
• xwputfld.c: reconciled with xputfld.c
• xwputstr.c: checked _WPuttxt returns
• xwrename.c: corrected parens in wcstombs test
• xwscanf.c: changed nconv to detect unstored conversions
• xwscanf.c: corrected overflow test
• xwscanf.c: corrected matching failure return
• xwscanf.c: added AltiVec stuff
• xwstoflt.c: reworked to avoid excess digits
• xwstoflt.c: corrected value for bad parse
• xwstoflt.c: added rollback logic to exponent parse
• xwstoflt.c: reconciled maxsig compares with xstoflt.c
• xwstopfx.c: corrected NaN) and hex parsing
• xwstopfx.c: corrected Inf parsing
• xwstoxfl.c: reworked to avoid excess digits
• xwstoxfl.c: corrected incomplete exponent parse
• xwstoxfl.c: repaired damaged predicate on fraction accumulator
• xwstoxfl.c: reconciled maxsig compares with xstoflt.c
• xxacosh.h: changed zero divide to invalid
• xxasinh.h: corrected negative x test
• xxatanh.h: corrected negative test
• xxcacos.h: corrected Inf+I*NaN logic
• xxcacos.h: corrected special cases
• xxcacosh.h: corrected special cases
• xxcacosh.h: corrected final atan2 first argument
• xxcasin.h: rewrote in terms of casinh
• xxcasinh.h: corrected special cases
• xxcatan.h: rewrote in terms of catanh
• xxcatanh.h: corrected special cases
• xxccos.h: rewrote in terms of ccosh
• xxccosh.h: corrected special cases
• xxcexp.h: added over/underflow tests
• xxcexp.h: moved trap setting to Exp
• xxcexp.h: corrected special cases
• xxclog.h: reduced cm precision for larger Hypot scaling
• xxclog.h: corrected precision near unit circle
• xxclog.h: added zero-divide reporting
• xxcsin.h: rewrote in terms of csinh
• xxcsinh.h: corrected special cases
• xxcsqrt.h: added special cases for C99
• xxcsqrt.h: corrected (inf, x) logic
• xxcsqrt.h: corrected (0 or finite, -0) logic
• xxctan.h: eliminated divide by Inf
• xxctan.h: rewrote in terms of ctanh
• xxctanh.h: eliminated divide by Inf
• xxctanh.h: corrected special cases
• xxerf.h: changed return for small x
• xxerfc.h: added underflow reporting
• xxerfc.h: added more intervals, extra precision exp
• xxerfc.h: corrected interval logic
• xxerfc.h: moved trap setting to Exp
• xxerfc.h: improved fits
• xxerfc.h: improved zero representations
• xxerfc.h: added erf declaration
• xxexp.h: moved trap setting to Exp
• xxexp2.h: split off exponent to improve accuracy
• xxexp2.h: moved trap setting to Exp
• xxexpm1.h: changed return for small x
• xxfdim.h: eliminated -0 return
• xxfma.h: corrected rounding logic
• xxfma.h: corrected accumulation
• xxfma.h: corrected exp test for huge+2*huge-huge, xp*yp add
• xxfma.h: separated out z=Inf case
• xxfma.h: added _Xp stuff
• xxfmax.h: added -0 handling
• xxfmin.h: added -0 handling
• xxfmod.h: corrected for large exponent difference
• xxlgamma.h: moved in stuff from xxxlgamma.h
• xxlgamma.h: added overflow report to lgamma_big
• xxlgamma.h: added patches for zeros
• xxlgamma.h: added [1.2, 1.5] approximation
• xxlgamma.h: improved root approximations
• xxllrint.h: simplified
• xxllrint.h: added FE_INEXACT logic
• xxllround.h: simplified
• xxllround.h: added round declaration
• xxlog1p.h: changed return for small x
• xxlog1p.h: replaced rational approx with _Log1p calls
• xxlog1p.h: replaced _Log1p call with extra precision approx
• xxlrint.h: simplified
• xxlrint.h: added FE_INEXACT logic
• xxlround.h: simplified
• xxnearbyint.h: simplified
• xxnearbyint.h: moved finite case to xxxrint.h
• xxnearbyint.h: corrected return syntax
• xxpow.h: corrected -0 return
• xxpow.h: corrected over/underflow logic, tidied
• xxpow.h: corrected (1+eps)^huge
• xxpow.h: made zexp long for safety
• xxpow.h: added _Xp stuff
• xxpow.h: moved trap setting to Exp
• xxpow.h: corrected xpx1 load, revised logic
• xxpow.h: added ln(x+a) logic
• xxpow.h: added _Pow for tgamma
• xxpow.h: corrected ln(1) to log2(1) conversion
• xxremquo.h: added qneg
• xxrint.h: simplified
• xxrint.h: added inexact check
• xxround.h: simplified
• xxsqrt.h: corrected -0 case
• xxstod.h: renamed, added perr
• xxstod.h: simplified accumulation
• xxtan.h: enhanced, moved Quad to xxxquad.h
• xxtan.h: corrected 1/eps sign
• xxtanh.h: added half argument step
• xxtanh.h: eliminated intermediate overflow
• xxtgamma.h: added over/underflow reporting
• xxtgamma.h: eliminated Lgamma call
• xxtgamma.h: moved large x handling to Tgamma
• xxtgamma.h: changed Tgamma_big to Gamma_big
• xxxatan.h: added inexact logic
• xxxcaddcc.h: corrected arithmetic
• xxxcaddcc.h: added infinity fixup
• xxxcaddcc.h: reconciled with multiply
• xxxcaddcc.h: removed infinity fixup
• xxxcaddcr.h: dropped assigns
• xxxcdivcc.h: corrected NaN tests
• xxxcdivcc.h: added infinity fixup
• xxxcdivcc.h: removed infinity fixup
• xxxcdivcr.h: dropped assigns
• xxxcmulcc.h: added infinity fixup
• xxxcmulcc.h: removed infinity fixup
• xxxcmulcr.h: dropped assigns
• xxxcosh.h: corrected -Inf return
• xxxcosh.h: moved trap setting to Exp
• xxxcsubcc.h: corrected arithmetic
• xxxcsubcc.h: added infinity fixup
• xxxcsubcc.h: reconciled with multiply
• xxxcsubcc.h: removed infinity fixup
• xxxcsubcr.h: dropped assigns
• xxxdtent.h: renamed, added perr
• xxxexp.h: changed exponent argument to long
• xxxexp.h: corrected c* precision
• xxxexp.h: corrected Inf sign and size of xexp
• xxxexp.h: added special code checking and trap setting
• xxxexp.h: corrected y==0 logic
• xxxexp.h: corrected -0 logic
• xxxhypot.h: improved scaling for csqrt
• xxxhypot.h: corrected zero arg case
• xxxlog.h: generalized Inf return
• xxxlog.h: added Logpoly for pow
• xxxprec.h: added new file
• xxxquad.h: enhanced, moved out of xxxsin.h
• xxxquad.h: increased c* precision
• xxxquad.h: added Xp stuff for 128-bit precision
• xxxquad.h: corrected inexact return
• xxxquad.h: made full precision
• xxxquad.h: added Quadph
• xxxquad.h: corrected overflow in 53-bit values
• xxxquad.h: added tables of large moduli
• xxxquad.h: corrected XSIZE usage with xpx
• xxxrint.h: added new file, copying from xxnearbyint.h
• xxxsin.h: enhanced, moved out Quad to xxxquad.h
• xxxsinh.h: corrected small x case
• xxxsinh.h: moved trap setting to Exp
• xxxtgamma.h: moved in Gamma_big from xxxlgamma.h
• xxxtgamma.h: moved in large x handling from xxtgamma.h
• xxxtgamma.h: changed Tgamma_big to Gamma_big