| ------------------------------------------------------------------------------ | |
| -- -- | |
| -- GNAT COMPILER COMPONENTS -- | |
| -- -- | |
| -- L I B . L O A D -- | |
| -- -- | |
| -- S p e c -- | |
| -- -- | |
| -- Copyright (C) 1992-2026, Free Software Foundation, Inc. -- | |
| -- -- | |
| -- GNAT is free software; you can redistribute it and/or modify it under -- | |
| -- terms of the GNU General Public License as published by the Free Soft- -- | |
| -- ware Foundation; either version 3, or (at your option) any later ver- -- | |
| -- sion. GNAT is distributed in the hope that it will be useful, but WITH- -- | |
| -- OUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY -- | |
| -- or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License -- | |
| -- for more details. You should have received a copy of the GNU General -- | |
| -- Public License distributed with GNAT; see file COPYING3. If not, go to -- | |
| -- http://www.gnu.org/licenses for a complete copy of the license. -- | |
| -- -- | |
| -- GNAT was originally developed by the GNAT team at New York University. -- | |
| -- Extensive contributions were provided by Ada Core Technologies Inc. -- | |
| -- -- | |
| ------------------------------------------------------------------------------ | |
| -- This child package contains the function used to load a separately | |
| -- compiled unit, as well as the routine used to initialize the unit | |
| -- table and load the main source file. | |
| package Lib.Load is | |
| ------------------------------- | |
| -- Handling of Renamed Units -- | |
| ------------------------------- | |
| -- A compilation unit can be a renaming of another compilation unit. | |
| -- Such renamed units are not allowed as parent units, that is you | |
| -- cannot declare a unit: | |
| -- with x; | |
| -- package x.y is end; | |
| -- where x is a renaming of some other package. However you can refer | |
| -- to a renamed unit in a with clause: | |
| -- package p is end; | |
| -- package p.q is end; | |
| -- with p; | |
| -- package pr renames p; | |
| -- with pr.q .... | |
| -- This means that in the context of a with clause, the normal fixed | |
| -- correspondence between unit and file names is broken. In the above | |
| -- example, there is no file named pr-q.ads, since the actual child | |
| -- unit is p.q, and it will be found in file p-q.ads. | |
| -- In order to deal with this case, we have to first load pr.ads, and | |
| -- then discover that it is a renaming of p, so that we know that pr.q | |
| -- really refers to p.q. Furthermore this can happen at any level: | |
| -- with p.q; | |
| -- package p.r renames p.q; | |
| -- with p.q; | |
| -- package p.q.s is end; | |
| -- with p.r.s ... | |
| -- Now we have a case where the parent p.r is a child unit and is | |
| -- a renaming. This shows that renaming can occur at any level. | |
| -- Finally, consider: | |
| -- with pr.q.s ... | |
| -- Here the parent pr.q is not itself a renaming, but it really refers | |
| -- to the unit p.q, and again we cannot know this without loading the | |
| -- parent. The bottom line here is that while the file name of a unit | |
| -- always corresponds to the unit name, the unit name as given to the | |
| -- Load_Unit function may not be the real unit. | |
| ----------------- | |
| -- Subprograms -- | |
| ----------------- | |
| procedure Initialize; | |
| -- Initialize internal tables | |
| procedure Initialize_Version (U : Unit_Number_Type); | |
| -- This is called once the source file corresponding to unit U has been | |
| -- fully scanned. At that point the checksum is computed, and can be used | |
| -- to initialize the version number. | |
| procedure Load_Main_Source; | |
| -- Called at the start of compiling a new main source unit to initialize | |
| -- the library processing for the new main source. Establishes and | |
| -- initializes the units table entry for the new main unit (leaving | |
| -- the Unit_File_Name entry of Main_Unit set to No_File if there are no | |
| -- more files. Otherwise the main source file has been opened and read | |
| -- and then closed on return. | |
| function Load_Unit | |
| (Load_Name : Unit_Name_Type; | |
| Required : Boolean; | |
| Error_Node : Node_Id; | |
| Subunit : Boolean; | |
| Corr_Body : Unit_Number_Type := No_Unit; | |
| Renamings : Boolean := False; | |
| With_Node : Node_Id := Empty; | |
| PMES : Boolean := False) return Unit_Number_Type; | |
| -- This function loads and parses the unit specified by Load_Name (or | |
| -- returns the unit number for the previously constructed units table | |
| -- entry if this is not the first call for this unit). Required indicates | |
| -- the behavior on a file not found condition, as further described below, | |
| -- and Error_Node is the node in the calling program to which error | |
| -- messages are to be attached. | |
| -- | |
| -- If the corresponding file is found, the value returned by Load is the | |
| -- unit number that indexes the corresponding entry in the units table. If | |
| -- a serious enough parser error occurs to prevent subsequent semantic | |
| -- analysis, then the Fatal_Error flag of the returned entry is set and | |
| -- in addition, the fatal error flag of the calling unit is also set. | |
| -- | |
| -- If the corresponding file is not found, then the behavior depends on | |
| -- the setting of Required. If Required is False, then No_Unit is returned | |
| -- and no error messages are issued. If Required is True, then an error | |
| -- message is posted, and No_Unit is returned. | |
| -- | |
| -- A special case arises in the call from Rtsfind, where Error_Node is set | |
| -- to Empty. In this case Required is False, and the caller in any case | |
| -- treats any error as fatal. | |
| -- | |
| -- The Subunit parameter is True to load a subunit, and False to load | |
| -- any other kind of unit (including all specs, package bodies, and | |
| -- subprogram bodies). | |
| -- | |
| -- The Corr_Body argument is normally defaulted. It is set only in the | |
| -- case of loading the corresponding spec when the main unit is a body. | |
| -- In this case, Corr_Body is the unit number of this corresponding | |
| -- body. This is used to set the Serial_Ref_Unit field of the unit | |
| -- table entry. It is also used to deal with the special processing | |
| -- required by RM 10.1.4(4). See description in lib.ads. | |
| -- | |
| -- Renamings activates the handling of renamed units as separately | |
| -- described in the documentation of this unit. If this parameter is | |
| -- set to True, then Load_Name may not be the real unit name and it | |
| -- is necessary to load parents to find the real name. | |
| -- | |
| -- With_Node is set to the with_clause or limited_with_clause causing | |
| -- the unit to be loaded, and is used to bypass the circular dependency | |
| -- check in the case of a limited_with_clause (Ada 2005, AI-50217). | |
| -- | |
| -- PMES indicates the required setting of Parsing_Main_Extended_Unit during | |
| -- loading of the unit. This flag is saved and restored over the call. | |
| -- Note: PMES is false for the subunit case, which seems wrong??? | |
| procedure Change_Main_Unit_To_Spec; | |
| -- This procedure is called if the main unit file contains a No_Body pragma | |
| -- and no other tokens. The effect is, if possible, to change the main unit | |
| -- from the body it references now, to the corresponding spec. This has the | |
| -- effect of ignoring the body, which is what we want. If it is impossible | |
| -- to successfully make the change, then the call has no effect, and the | |
| -- file is unchanged (this will lead to an error complaining about the | |
| -- inappropriate No_Body spec). | |
| function Create_Dummy_Package_Unit | |
| (With_Node : Node_Id; | |
| Spec_Name : Unit_Name_Type) return Unit_Number_Type; | |
| -- With_Node is the Node_Id of a with statement for which the file could | |
| -- not be found, and Spec_Name is the corresponding unit name. This call | |
| -- creates a dummy package unit so that compilation can continue without | |
| -- blowing up when the missing unit is referenced. | |
| procedure Make_Child_Decl_Unit (N : Node_Id); | |
| -- For a child subprogram body without a spec, we create a subprogram | |
| -- declaration in order to attach the required parent link. We create | |
| -- a Units_Table entry for this declaration, in order to maintain a | |
| -- one-to-one correspondence between compilation units and table entries. | |
| procedure Make_Instance_Unit (N : Node_Id; In_Main : Boolean); | |
| -- When a compilation unit is an instantiation, it contains both the | |
| -- declaration and the body of the instance, each of which can have its | |
| -- own elaboration routine. The file itself corresponds to the declaration. | |
| -- We create an additional entry for the body, so that the binder can | |
| -- generate the proper elaboration calls to both. The argument N is the | |
| -- compilation unit node created for the body. | |
| -- | |
| -- If the instance is not the main program, we still generate the instance | |
| -- body even though we do not generate code for it. In that case we still | |
| -- generate a compilation unit node for it, and we need to make an entry | |
| -- for it in the units table, so as to maintain a one-to-one mapping | |
| -- between table and nodes. The table entry is used among other things to | |
| -- provide a canonical traversal order for context units for CodePeer. | |
| -- The flag In_Main indicates whether the instance is the main unit. | |
| procedure Version_Update (U : Node_Id; From : Node_Id); | |
| -- This routine is called when unit U is found to be semantically | |
| -- dependent on unit From. It updates the version of U to register | |
| -- dependence on the version of From. The arguments are compilation | |
| -- unit nodes for the relevant library nodes. | |
| end Lib.Load; |