ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/libecb/Changes
Revision: 1.47
Committed: Wed Apr 13 15:43:27 2022 UTC (2 years, 1 month ago) by root
Branch: MAIN
CVS Tags: HEAD
Changes since 1.46: +1 -0 lines
Log Message:
*** empty log message ***

File Contents

# Content
1 TODO: naming of cutnions, index_to_coord, xy_to_s, xy_to_linear/seq, 32/64
2 TODO: returning packed coordinates, really
3 TODO: support order=0?
4 0x0001000c
5 - added ecb_hilbert2d function family.
6 - use __builtin_popcountl on 64 bit clang and gcc.
7 - use 64 bit builtins, not 128 bit builtins, for ctz/clz on 64 bit cpus.
8 - add ecb_clz* functions,
9 - renamed ecb_gray* functions to have trailing bit width.
10 - use ecb_function_ consistently for newly added functions.
11 - say goodbye to gnu-style function definitions - they were not
12 used consistently anyway, and imho make the code harder to read,
13 and didn't even work for indented definitions.
14 - windows/ms versions of clz returned the wrongt value (reported by b_honas).
15
16 0x0001000b
17 - added ecb_gray*_{de,en}code functions.
18
19 0x0001000a
20 - added ecb_ptrmix.
21
22 0x00010009
23 - added ecb_i2a family of functions.
24 - added ECB_64BIT_NATIVE.
25
26 TODO:
27 08:30:06 <b_jonas> I think it could be worth to add a macro that works like alignof or _Alignof on sane
28 compilers, and like __alignof on MS compilers that support it, see
29 http://msdn.microsoft.com/en-us/library/45t0s5f4.aspx
30 08:30:24 <b_jonas> even if you can't support it on all the old compilers
31 08:31:17 <b_jonas> I'd also like a macro for alignas, but sadly, that seems impossible in general, because
32 the MS compiler only has some half-attempt to do something similar but with different and
33 more broken semantics, see http://msdn.microsoft.com/en-us/library/83ythb65.aspx
34 08:31:35 <b_jonas> but I wonder if some special case could still be worth to support
35 08:32:23 <b_jonas> probably not, because it would just account to making a union with a highly aligned type,
36 which is something I can do on any compiler portably
37
38 TODO: #define ECB_IS_INTEGRAL(x) !((1 ? 1 : (x)) / 2)
39 #define ECB_IS_INTEGRAL(x) (sizeof ((x) + 1.0f) != sizeof((x) + 1ULL))
40
41 TODO: ecb_minpot, either using bit tricks or ecb_ldXX
42
43 TODO: __builtin_popcountll exists...
44
45 TODO: __builtin_powi
46
47 TODO: https://gustedt.wordpress.com/2010/06/08/detect-empty-macro-arguments/
48
49 implement is_constant for c11: https://gustedt.wordpress.com/2013/08/22/testing-compile-time-constness-and-null-pointers-with-c11s-_generic/
50
51 #ifdef _MSC_VER
52
53 #include <stdlib.h>
54 #define bswap_32(x) _byteswap_ulong(x)
55 #define bswap_64(x) _byteswap_uint64(x)
56
57 #elif defined(__APPLE__)
58
59 // Mac OS X / Darwin features
60 #include <libkern/OSByteOrder.h>
61 #define bswap_32(x) OSSwapInt32(x)
62 #define bswap_64(x) OSSwapInt64(x)
63
64 #elif defined(__sun) || defined(sun)
65
66 #include <sys/byteorder.h>
67 #define bswap_32(x) BSWAP_32(x)
68 #define bswap_64(x) BSWAP_64(x)
69
70 #elif defined(__FreeBSD__)
71
72 #include <sys/endian.h>
73 #define bswap_32(x) bswap32(x)
74 #define bswap_64(x) bswap64(x)
75
76 #elif defined(__OpenBSD__)
77
78 #include <sys/types.h>
79 #define bswap_32(x) swap32(x)
80 #define bswap_64(x) swap64(x)
81
82 #elif defined(__NetBSD__)
83
84 #include <sys/types.h>
85 #include <machine/bswap.h>
86 #if defined(__BSWAP_RENAME) && !defined(__bswap_32)
87 #define bswap_32(x) bswap32(x)
88 #define bswap_64(x) bswap64(x)
89 #endif
90
91 #else
92
93 #include <byteswap.h>
94
95 #endif
96
97 TODO: generic poprcount etc., also fast_t for them?
98
99 - allow any rotate count in rcb_rot*, at the expense
100 of bad optimisation results on some platforms or with
101 some word sizes.
102 - try to use inttypes.h on mingw.
103
104 0x00010008
105 - aligned/unaligned load/store, bswap, host order
106 conversion.
107 - generic C++ ecb_rot[lr], ecb_popcount, ecb_ctz,
108 ecb_is_pot32 and ecb_bitrev functions.
109
110 0x00010007
111 - new ECB_OPTIMIZE_SIZE symbol.
112
113 0x00010006
114 - new ECB_MEMORY_FENCE_RELAXED memory fence.
115 - use acquire/eelease memory barriers on sun workshop pro, not read/write.
116 - rely on c++ compiler barriers to do the right thing in gcc/clang.
117 - change release memory fence to memory barrier on ia32/ia64.
118 - apply ctz/ld patch for msc by Zsbán Ambrus.
119 - ECB_PTRSIZE erroneously was 8 on most 32bit systems (
120 found by Zsbán Ambrus).
121 - improved compiletime detection of endianness, also, allow
122 runtime detection to indicate other-than-big/little endianness.
123 - no memory barrier neded on arm < 6.
124
125 0x00010005
126 - improve ecb_binary16_to_float.
127 - add ecb_float_to_binary16.
128 - add ecb_binary16_to_binary32 and ecb_binary32_to_binary16 pair.
129
130 0x00010001
131 - add ecb_is_pot32/64.
132 - add intptr_t/uintptr_t.
133 - add ECB_PTRSIZE.
134 - more macros for C/C++ version checks.
135 - support C11 atomics for memory fences.
136 - support gcc-4.7 atomics for memory fences.
137 - support m68k, m88k and sh (patch by Miod Vallat).
138 - add ecb_binary16_to_float.
139
140 TODO: ecb_restrict_array etc. http://ue.tst.eu/5093eafd713ec5fda776d8065070aa4c.txt
141 TODO: ffs
142 64 bit variants of everything
143 TODO: examples from X for clz/ctz
144 TODO: arithmetic right shift
145 TODO: template/generic functions for x32/x64 and so on
146 TODO: #define ecb_integer_multiples_of(n,d) ((char (*)[d])(n) - (char (*)[d])0)
147 TODO: generalised shift
148 TODO: #define ECB_FAST_UNALIGNED_ACCESS
149 unsigned long gensh(unsigned long v, int x) {
150 int a, b;
151 a = (v << x) & -(((unsigned int)x) < 32);
152 x = -x;
153 b = (v >> x) & -(((unsigned int)x) < 32);
154 return a|b;
155 }
156
157 TODO: export(=dllexport) & hidden
158 TODO: flatten
159 TODO: warning(msg)
160 TODO: error(msg)
161 TODO: leaf (uh), noclone (hmmm)
162 TODO: nonnull, returns_nonnull
163 TODO: nothrow
164 TODO: used
165 TODO: trap
166 TODO: http://llvm.org/docs/doxygen/html/Compiler_8h_source.html
167
168 TODO: read/write unaligned macros
169 TODO: htonl and friends
170 TODO: macro to convert from unsigned to signed "the natural way"
171 TODO: ecb_static_assert, with message (just like boost), or somesuch, using array-declaration
172 TODO: alignof
173
174