1 Attribute Usage
2 ===============
3
4 Consider attribute assignment observations, along with the possibility of class attribute
5 assignment.
6
7 Note direct assignments as usual, indirect assignments via the attribute usage
8 mechanism. During attribute collection and inference, add assigned values to all
9 inferred targets.
10
11 Since class attributes can be assigned, StoreAttrIndex would no longer need to reject
12 static attributes, although this might still be necessary where attribute usage analysis
13 has not been performed.
14
15 Potentially consider changing static attribute details to use object-relative offsets in
16 order to simplify the instruction implementations. This might allow us to eliminate the
17 static attribute flag for attributes in the object table, at least at run-time.
18
19 Consider attribute usage observations being suspended inside blocks where AttributeError
20 may be caught (although this doesn't anticipate such exceptions being caught outside a
21 function altogether).
22
23 Consider type deduction and its consequences where types belong to the same hierarchy
24 and where a guard could be generated for the most general type.
25
26 Consider permitting multiple class alternatives where the attributes are all identical.
27
28 Support class attribute positioning similar to instance attribute positioning, potentially
29 (for both) based on usage observations. For example, if __iter__ is used on two classes,
30 the class attribute could be exposed at a similar relative position to the class (and
31 potentially accessible using a LoadAttr-style instruction).
32
33 **** Constant attribute users need not maintain usage since they are already resolved. ****
34
35 Loop entry points should capture usage to update later assignments in the loop.
36 The continue and break statements should affect usage propagation.
37
38 Consider handling CallFunc in micropython.inspect in order to produce instances of specific classes.
39 Then, consider adding support for guard removal/verification where known instances are involved.
40 Consider handling branches of values within namespaces in order to support more precise value usage.
41
42 Frame Optimisations
43 ===================
44
45 Stack frame replacement where a local frame is unused after a call, such as in a tail call
46 situation.
47
48 Local assignment detection plus frame re-use. Example: slice.__init__ calls
49 xrange.__init__ with the same arguments which are unchanged in xrange.__init__. There is
50 therefore no need to build a new frame for this call.
51
52 Function Specialisation
53 =======================
54
55 Specialisation of certain functions, such as isinstance(x, cls) where cls is a known
56 constant.
57
58 Structure and Object Table Optimisations
59 ========================================
60
61 Fix object table entries for attributes not provided by any known object, or provide an
62 error, potentially overridden by options. For example, the augmented assignment methods
63 are not supported by the built-in objects and thus the operator module functions cause
64 the compilation to fail. Alternatively, just supply the methods since something has to do
65 so in the builtins.
66
67 Consider attribute merging where many attributes are just aliases for the same underlying
68 definition.
69
70 Consider references to defaults as occurring only within the context of a particular
71 function, thus eliminating default value classes if such functions are not themselves
72 invoked.
73
74 Scope Handling
75 ==============
76
77 Consider merging the InspectedModule.store tests with the scope conflict handling.
78
79 Consider labelling _scope on assignments and dealing with the assignment of removed
80 attributes, possibly removing the entire assignment, and distinguishing between such cases
81 and unknown names.
82
83 Check name origin where multiple branches could yield multiple scope interpretations:
84
85 ----
86 try:
87 set # built-in name
88 except NameError:
89 from sets import Set as set # local definition of name
90
91 set # could be confused by the local definition at run-time
92 ----
93
94 Object Coverage
95 ===============
96
97 Support __init__ traversal (and other implicit names) more effectively.
98
99 Other
100 =====
101
102 Check context_value initialisation (avoiding or handling None effectively).
103
104 __getitem__ could be written in Python, using a native method only to access fragments.
105
106 Consider better "macro" support where new expressions need to be generated and processed.
107
108 Detect TestIdentity results involving constants, potentially optimising status-affected
109 instructions:
110
111 TestIdentity(x, y) # where x is always y
112 JumpIfFalse(...) # would be removed (never false)
113 JumpIfTrue(...) # changed to Jump(...)
114
115 Status-affected blocks could be optimised away for such constant-related results.
