111
|
1 ------------------------------------------------------------------------------
|
|
2 -- --
|
|
3 -- GNAT COMPILER COMPONENTS --
|
|
4 -- --
|
|
5 -- S E M _ R E S --
|
|
6 -- --
|
|
7 -- S p e c --
|
|
8 -- --
|
131
|
9 -- Copyright (C) 1992-2018, Free Software Foundation, Inc. --
|
111
|
10 -- --
|
|
11 -- GNAT is free software; you can redistribute it and/or modify it under --
|
|
12 -- terms of the GNU General Public License as published by the Free Soft- --
|
|
13 -- ware Foundation; either version 3, or (at your option) any later ver- --
|
|
14 -- sion. GNAT is distributed in the hope that it will be useful, but WITH- --
|
|
15 -- OUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY --
|
|
16 -- or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License --
|
|
17 -- for more details. You should have received a copy of the GNU General --
|
|
18 -- Public License distributed with GNAT; see file COPYING3. If not, go to --
|
|
19 -- http://www.gnu.org/licenses for a complete copy of the license. --
|
|
20 -- --
|
|
21 -- GNAT was originally developed by the GNAT team at New York University. --
|
|
22 -- Extensive contributions were provided by Ada Core Technologies Inc. --
|
|
23 -- --
|
|
24 ------------------------------------------------------------------------------
|
|
25
|
|
26 -- Resolution processing for all subexpression nodes. Note that the separate
|
|
27 -- package Sem_Aggr contains the actual resolution routines for aggregates,
|
|
28 -- which are separated off since aggregate processing is complex.
|
|
29
|
|
30 with Types; use Types;
|
|
31
|
|
32 package Sem_Res is
|
|
33
|
|
34 -- As described in Sem_Ch4, the type resolution proceeds in two phases.
|
|
35 -- The first phase is a bottom up pass that is achieved during the
|
|
36 -- recursive traversal performed by the Analyze procedures. This phase
|
|
37 -- determines unambiguous types, and collects sets of possible types
|
|
38 -- where the interpretation is potentially ambiguous.
|
|
39
|
|
40 -- On completing this bottom up pass, which corresponds to a call to
|
|
41 -- Analyze on a complete context, the Resolve routine is called which
|
|
42 -- performs a top down resolution with recursive calls to itself to
|
|
43 -- resolve operands.
|
|
44
|
|
45 -- Since in practice a lot of semantic analysis has to be postponed until
|
|
46 -- types are known (e.g. static folding, setting of suppress flags), the
|
|
47 -- Resolve routines also complete the semantic analysis, and call the
|
|
48 -- expander for possible expansion of the completely type resolved node.
|
|
49
|
|
50 procedure Ambiguous_Character (C : Node_Id);
|
|
51 -- Give list of candidate interpretations when a character literal cannot
|
|
52 -- be resolved, for example in a (useless) comparison such as 'A' = 'B'.
|
|
53 -- In Ada 95 the literals in question can be of type Character or Wide_
|
|
54 -- Character. In Ada 2005 Wide_Wide_Character is also a candidate. The
|
|
55 -- node may also be overloaded with user-defined character types.
|
|
56
|
|
57 procedure Analyze_And_Resolve (N : Node_Id);
|
|
58 procedure Analyze_And_Resolve (N : Node_Id; Typ : Entity_Id);
|
|
59 procedure Analyze_And_Resolve
|
|
60 (N : Node_Id;
|
|
61 Typ : Entity_Id;
|
|
62 Suppress : Check_Id);
|
|
63 procedure Analyze_And_Resolve
|
|
64 (N : Node_Id;
|
|
65 Suppress : Check_Id);
|
|
66 -- These routines combine the effect of Analyze and Resolve. If a Suppress
|
|
67 -- argument is present, then the analysis is done with the specified check
|
|
68 -- suppressed (can be All_Checks to suppress all checks). These checks are
|
|
69 -- suppressed for both the analysis and resolution. If the type argument
|
|
70 -- is not present, then the Etype of the expression after the Analyze
|
|
71 -- call is used for the Resolve.
|
|
72
|
|
73 procedure Check_Parameterless_Call (N : Node_Id);
|
|
74 -- Several forms of names can denote calls to entities without parameters.
|
|
75 -- The context determines whether the name denotes the entity or a call to
|
|
76 -- it. When it is a call, the node must be rebuilt accordingly and
|
|
77 -- reanalyzed to obtain possible interpretations.
|
|
78 --
|
|
79 -- The name may be that of an overloadable construct, or it can be an
|
|
80 -- explicit dereference of a prefix that denotes an access to subprogram.
|
|
81 -- In that case, we want to convert the name into a call only if the
|
|
82 -- context requires the return type of the subprogram. Finally, a
|
|
83 -- parameterless protected subprogram appears as a selected component.
|
|
84 --
|
|
85 -- The parameter T is the Typ for the corresponding resolve call.
|
|
86
|
|
87 procedure Preanalyze_And_Resolve (N : Node_Id; T : Entity_Id);
|
131
|
88 -- Performs a preanalysis of expression node N. During preanalysis, N is
|
111
|
89 -- analyzed and then resolved against type T, but no expansion is carried
|
131
|
90 -- out for N or its children. For more info on preanalysis read the spec
|
111
|
91 -- of Sem.
|
|
92
|
|
93 procedure Preanalyze_And_Resolve (N : Node_Id);
|
|
94 -- Same, but use type of node because context does not impose a single type
|
|
95
|
131
|
96 procedure Preanalyze_With_Freezing_And_Resolve (N : Node_Id; T : Entity_Id);
|
|
97 -- Same, but perform freezing of static expressions of N or its children.
|
|
98
|
111
|
99 procedure Resolve (N : Node_Id; Typ : Entity_Id);
|
|
100 procedure Resolve (N : Node_Id; Typ : Entity_Id; Suppress : Check_Id);
|
|
101 -- Top-level type-checking procedure, called in a complete context. The
|
|
102 -- construct N, which is a subexpression, has already been analyzed, and
|
|
103 -- is required to be of type Typ given the analysis of the context (which
|
|
104 -- uses the information gathered on the bottom-up phase in Analyze). The
|
|
105 -- resolve routines do various other processing, e.g. static evaluation.
|
|
106 -- If a Suppress argument is present, then the resolution is done with the
|
|
107 -- specified check suppressed (can be All_Checks to suppress all checks).
|
|
108
|
|
109 procedure Resolve (N : Node_Id);
|
|
110 -- A version of Resolve where the type to be used for resolution is taken
|
|
111 -- from the Etype (N). This is commonly used in cases where the context
|
|
112 -- does not add anything and the first pass of analysis found the correct
|
|
113 -- expected type.
|
|
114
|
|
115 procedure Resolve_Discrete_Subtype_Indication
|
|
116 (N : Node_Id;
|
|
117 Typ : Entity_Id);
|
|
118 -- Resolve subtype indications in choices (case statements and aggregates)
|
|
119 -- and in index constraints. Note that the resulting Etype of the subtype_
|
|
120 -- indication node is set to the Etype of the contained range (i.e. an
|
|
121 -- Itype is not constructed for the actual subtype).
|
|
122
|
|
123 procedure Resolve_Entry (Entry_Name : Node_Id);
|
|
124 -- Find name of entry being called, and resolve prefix of name with its
|
|
125 -- own type. For now we assume that the prefix cannot be overloaded and
|
|
126 -- the name of the entry plays no role in the resolution.
|
|
127
|
|
128 function Valid_Conversion
|
|
129 (N : Node_Id;
|
|
130 Target : Entity_Id;
|
|
131 Operand : Node_Id;
|
|
132 Report_Errs : Boolean := True) return Boolean;
|
|
133 -- Verify legality rules given in 4.6 (8-23). Target is the target type
|
|
134 -- of the conversion, which may be an implicit conversion of an actual
|
|
135 -- parameter to an anonymous access type (in which case N denotes the
|
|
136 -- actual parameter and N = Operand). Returns a Boolean result indicating
|
|
137 -- whether the conversion is legal. Reports errors in the case of illegal
|
|
138 -- conversions, unless Report_Errs is False.
|
|
139
|
|
140 private
|
|
141 procedure Resolve_Implicit_Type (N : Node_Id) renames Resolve;
|
|
142 pragma Inline (Resolve_Implicit_Type);
|
|
143 -- We use this renaming to make the application of Inline very explicit to
|
|
144 -- this version, since other versions of Resolve are not inlined.
|
|
145
|
|
146 end Sem_Res;
|