1 This file is the short summary of the API changes:
3 02.01.2018 - Non-backward compatible
4 Immediate source argument has not been supported
5 for NOT, NEG, and CLZ opcodes anymore. No CPU
6 supports immedate arguments for these opcodes.
8 26.12.2017 - Non-backward compatible
9 The MOVU opcodes are removed because the emulation
10 is inefficient. The sljit_emit_mem() operation is
13 18.10.2017 - Non-backward compatible
14 The SLJIT_CALL0 - SLJIT_CALL3 jump types are
15 replaced by sljit_emit_call and sljit_emit_icall
16 function calls. These functions allows declaring
19 06.05.2017 - Non-backward compatible
20 Src argument is removed from sljit_emit_op_flags.
22 24.04.2017 - Non-backward compatible
23 The sljit_is_fpu_available function is replaced
24 by sljit_has_cpu_feature.
26 20.04.2017 - Non-backward compatible
27 x86 specific cmov is changed to a general function
29 27.03.2017 - Non-backward compatible
30 JIT stack is changed from bottom-up to top-town.
32 15.01.2017 - Non-backward compatible
33 Move with update may modifiy flags, the base register
34 can only be used once and [reg+reg<<shift] addressing
35 mode where shift > 0 is not supported anymore.
37 12.01.2017 - Non-backward compatible
38 Introducing a new flag mechanism which provides better
39 compatibility with CPUs without flags. Only two flags
40 remain: zero and variable. The current type of the
41 variable flag is specified by the arithmetic operator.
42 The SLJIT_KEEP_FLAGS is removed.
44 29.02.2016 - Non-backward compatible
45 Several types and instructions are renamed to improve
46 readability. In general byte, half, and int are renamed
47 to 8, 16, and 32. Floating point types are also renamed
48 from d and s to f64 and f32.
50 [s|u]b -> [s|u]8 (8 bit values)
51 [s|u]h -> [s|u]16 (16 bit values)
52 [s|u]i -> [s|u]32 (32 bit values)
53 d -> f64 (64 bit floating point value)
54 s -> f32 (32 bit floating point value)
56 18.05.2015 - Non-backward compatible
57 SLJIT_[I|]L[U|S]DIV is renamed to SLJIT_[I|][U|S]DIVMOD
59 29.09.2014 - Non-backward compatible
60 The sljit_create_compiler, sljit_allocate_stack, and
61 sljit_free_stack functions have an allocator_data
64 19.09.2014 - Non-backward compatible
65 Using I, D, S prefixes in conditional and floating
66 point operations. And an L prefix to long multiplication
67 and division (op0 opcodes).
69 11.08.2014 - Non-backward compatible
70 A currently unused options parameter is added to sljit_emit_enter
71 and sljit_set_context.
73 06.07.2014 - Non-backward compatible
74 SCRATCH registers are renamed to Rx and SAVED registers
75 are renamed to Sx. See the explanation of these registers
78 31.05.2014 - Non-backward compatible
79 SLJIT_TEMPORARY_EREGx registers were not renamed to
80 SLJIT_SCRATCH_EREGx when the change was done on 08.11.2012
82 05.03.2014 - Backward compatible
83 The sljit_set_target now supports those jumps, which
84 does not created with SLJIT_REWRITABLE_JUMP flag.
85 Reason: sljit_emit_ijump does not support conditional
88 03.03.2014 - Non-backward compatible
89 SLJIT_MOV_UI cannot be combined with SLJIT_INT_OP.
90 Reason: SLJIT_INT_OP flag is not recommended to use
91 directly, and SLJIT_IMOV has no sign bit.
93 29.01.2014 - Backward compatible
94 Bits assigned to SLJIT_MEM and SLJIT_IMM flags are changed.
95 Reason: the most common cases are fits into one byte now,
96 and more registers can be supported in the future.
98 08.11.2012 - Non-backward compatible
99 SLJIT_TEMPORARY_REGx registers are renamed to SLJIT_SCRATCH_REGx.
101 07.11.2012 - Non-backward compatible
102 sljit_emit_cond_value is renamed to sljit_emit_op_flags. An
103 extra source argument is added which will be used in the future.
105 05.11.2012 - Backward compatible
106 sljit_emit_cond_value now supports SLJIT_AND and SLJIT_INT_OP
107 flags, which makes this function complete.
109 01.11.2012 - Non-backward compatible
110 SLJIT_F* opcodes are renamed to SLJIT_*D to show that
111 they are double precision operators. Furthermore
112 SLJIT_*S single precision opcodes are added.
114 01.11.2012 - Non-backward compatible
115 Register arguments of operations with SLJIT_INT_OP flag
116 must be computed by another operation with SLJIT_INT_OP flag.
117 The same way as SLJIT_SINGLE_OP flag works with floating point
118 numbers. See the description of SLJIT_INT_OP.
120 01.11.2012 - Backward compatible
121 All operations whose support the SLJIT_INT_OP flag, have an
122 alternate name now, which includes the SLJIT_INT_OP. These
123 names starting with I.
125 31.10.2012 - Non-backward compatible
126 Renaming sljit_w to sljit_sw, sljit_i to sljit_si, sljit_h
127 to sljit_sh, and sljit_b to sljit_sb. Reason: their sign
128 bit is part of the type now.
130 20.10.2012 - Non-backward compatible
131 Renaming SLJIT_C_FLOAT_NAN to SLJIT_C_FLOAT_UNORDERED.
132 Reason: all architectures call these unordered comparions.