JPH0778087A - Compiler device - Google Patents

Abstract

PURPOSE:To obtain the compiler device which can omit the attribute definitions of common symbols when an execution type program is generated from plural object programs as to the process of a compiler on a computer. CONSTITUTION:An object generation part 1 generates a common symbol section 7 as to the common symbols in a source program 2 and sets a data area and attribute definition information as to the respective common symbols. An execution type program generation part 4 generates the execution type program by using the common symbol section 7 that a common symbol management part 6 selects when there is the common symbol section 7 of the object program 3. The common symbol management part 6 selects the common symbol section 7 having the merging result of all pieces of attribute definition information when there are no contradictions between all items of pieces of attribute definition information of plural common symbol sections 7 by the common symbols.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to the processing of a compiler in a computer, and more particularly to a compiler capable of efficiently processing a common symbol when a plurality of object programs are combined and edited to generate an executable program. Regarding the device.

[0002]

2. Description of the Related Art In a computer, a target program is generated from a source program,
There is a compiler device that performs a process of a compiler that generates a single executable program by performing so-called combined edit processing on a plurality of target programs.

In such a combined editing process, common symbols declared in each source program of a plurality of target programs to be processed are processed such that each same common symbol represents the same set of data, It is well known that for each common symbol, an executable program is generated so that the plurality of target programs can access the same data on the executable program.

For the processing, in the object generation unit of the compiler, the analysis unit uses the common symbol declaration statement of the source program (for example, in the case of FORTRAN programming language,
When the COMMON statement illustrated in FIG. 4 (a) or (b) is analyzed, the data of the target program to be generated and the data attribute information are set. For example, in the data section, the common symbol and the program are included as part of the data attribute information. The information indicating the COMMON statement is set by associating it with the data name in.

In the example of the COMMON statement in FIG. 4, (a) and (b) show respective parts of a program unit in which individual object programs are to be generated. In the source program of (a), data in the program is written. Common symbol “BL
"K" is defined, and in the source program (b), a common symbol "BLK" is defined for the data of the data name "B" in the program.

The COMMON statement is used to combine and edit a target program generated from both source programs.
Generates an executable program so that the data with the data name "A" specified by one program and the data with the data name "B" specified by the other program are treated as the same data, using the common symbol "BLK" as an intermediary. Indicates that you need to.

In the program example of FIG. 4, as is well known, "REAL A (1000)" is a data attribute definition indicating "the data of data name A is a real number type and the number of array elements is 1000 in a one-dimensional array". is there.

[PARTITION P = (number of processors = 2, division range = 1: 1000, division method = equal)] and “LOCATE A (P)”
In a multi-processor environment, "allocate data A to a processor according to allocation method P" in the LOCATE statement, and allocation method P uses "P =" in the PARTITION statement to "allocate two processors, elements 1 to 1000, This is a data attribute definition that indicates "equally", and by this, the target program is generated so that the array data A is divided into elements 1 to 500 and elements 501 to 1000 and shared by the two processors. Will be.

Next, when a plurality of target programs are combined and edited to generate an executable program, in the processing of the executable program generator, the target program has common symbol information, and two or more different target programs are stored. When there is the same common symbol, the executable program is generated so that the data corresponding to the same common symbol in those programs is accessed as the data of the same storage area.

To this end, the executable program generator unconditionally selects the prepared common symbol, for example, the data prepared for the data name corresponding to the common symbol of the one target program that is the first in the input order. Then, all programs to be combined proceed with the combined edit processing so that the access destinations are adjusted so that the selected same data is accessed by each data name corresponding to the common symbol.

Therefore, there is no problem if the same data attribute definitions are defined in all the object programs to be combined, but if the attributes that are inconsistent with each other are defined by mistake, the processing result is inconvenient. (When combining the target programs generated from (a) and (b) in the example of Fig. 4, the partition statement is only in (a) and not in (b), so it can be combined correctly. Or not guaranteed).

Further, when the program to be combined and edited has a relationship between the main program and the subprogram, it is desired to create a program by defining different attributes for each variable to be combined with a common symbol. Sometimes.

For example, regarding the data attribute which is used in the main program but not used in the sub program, from the viewpoint of ease of programming, the sub program does not need to be aware of it (thus not defining it).

Alternatively, if the subprogram defines only the basic data attributes necessary for enabling the target program generation process in subprogram units, and the main program can define more detailed attributes, 1
Since the individual subprograms can be commonly used by programs that define various different data attributes in detail, the subprogram can be created as a general-purpose component.

Therefore, in the case where data is made to be the same through the common symbol in the combined editing process, it is checked that there is no contradiction between the data attribute definitions of the related programs, and the data attribute definition can be omitted in each program. Is desired.

It is an object of the present invention to provide a compiler device capable of omitting attribute definition for common symbols.

[0017]

FIG. 1 is a block diagram showing the configuration of the present invention. The figure shows the configuration of a compiler device, in which an object generation unit 1 analyzes a source program 2 written in a predetermined programming language to generate a target program 3, and an execution format program generation unit 4 generates a plurality of target programs 3. Combined edit and execute program 5
And a common symbol management unit 6.

The object generation unit 1 generates a common symbol section 7 in the object program 3 for a predetermined declaration regarding a common symbol in the source program 2,
In the common symbol section 7, a data area and attribute definition information regarding each common symbol are set.

When the target program 3 to be combined and edited has a common symbol section 7, the executable program generator 4 uses the common symbol section 7 selected by the common symbol manager 6 to use the common symbol section 7. Process the generation of the executable program related to.

For each of the common symbols, the common symbol management section 6 determines, for each common symbol, the common symbol section 7 of the plurality of object programs 3 having the common symbol, the attribute definition information regarding the common symbol of all the common symbol sections 7. Are compared for each item, and when there is no contradiction among all the items, the common symbol section 7 having the attribute definition information which is the result of merging all the attribute definition information is selected to execute the executable program. Notify the generation unit 4.

[0021]

With the compiler apparatus of the present invention, when a program in which a common symbol is defined is combined and edited so as to be an executable program, a contradiction of attribute definitions between programs is checked. The data of the program having the attribute definition with the most definition items as a result of merging is selected as the data corresponding to the common symbol.

Therefore, regarding the common symbol, if an inconsistency in the attribute definitions between the programs is detected and appropriate action can be taken, and at least one of the programs to be combined and edited has a complete attribute definition, As long as the definition does not conflict with the attribute definition, other programs can omit the attribute definition or define it so that it can be merged with different attribute definitions.

[0023]

FIG. 2 (a) shows a conceptual structure of an object program having a common symbol section according to the present invention, in which a data section is provided between a data section and an execution section of a conventional object program. The common symbol section is added following.

As is well known, the data section is a part for setting the attribute relating to the data and the information of the data body, and the execution part section is the part for setting the information of the processing procedure. Information will be placed in the common symbol section separately from the data section.

In the common symbol section, the attribute definition information of the data and the data body (necessary initial value is set for each common symbol) are similar to the contents set in the data section for each data name in the case of normal data. Has) is set.

Each attribute definition information set in the common symbol section is, for example, as shown in FIG. 2B, control information such as the length of the attribute information and common symbol information such as the common symbol name, followed by the attribute definition. Allow information to be set.

The attribute definition information is composed of a plurality of items, for example, basic attribute 1, basic attribute 2, ... Basic attribute m, additional attribute 1, additional attribute 2 ,. Then, each item position is made to correspond to each specific type of attribute definition.

Here, each basic attribute is an attribute whose definition is indispensable (including the case where it is determined as a default value), such as a real number, a data type such as an integer, an array or a simple variable, the configuration and size of the array, Others. Further, the additional attribute is an attribute that is arbitrarily defined, and for example, "PARTITION" of the program shown in FIG. 4A is an example thereof.

When the object generator 1 analyzes the source program 2 to generate the target program 3, and identifies a COMMON statement that declares a common symbol to the source program 2, the target program 3 receives the common symbol section 7
To provide.

The object generation unit 1 stores the local data name in the common symbol section 7 according to the correspondence between the common symbol specified in the COMMON statement and the local data name (BLK and A in the example of FIG. 4 (a)). Information of data according to the attributes defined for the data is generated in the common symbol section, and the attribute definition information is set in the format of FIG.

The object generation unit 1 sends the attribute definition information to
The contents determined by each item position are sequentially set in each item of FIG. 2 (b), and valid information is generated as information up to the last item to be set. If so, those items are set with predetermined null symbols, for example.

When a plurality of target programs are specified and the target program has a common symbol section when a target program has a common symbol section when a plurality of target programs are combined and edited to generate the executable program, the execution-form program generator 4 If the same common symbol exists in the common symbol sections of a plurality of different target programs, the common symbol management unit 6 is requested to perform the processing.

In this case, the common symbol management unit 6 selects one common symbol section from a plurality of common symbol sections having the same common symbol, and the execution format program generation unit 4 selects the result as described later. Or it is detected that the attribute definitions are inconsistent among a plurality of common symbol sections, and as a result, the execution format program generation unit 4 is notified as selection failure.

In response to this notification, the executable program generator 4 receives the data information corresponding to the common symbol of the selected common symbol section from the data to be accessed by all the data names associated with the common symbol. As, proceed with the merge editing.

FIG. 3 is a diagram showing an example of the processing flow of the common symbol management section of the present invention. In this case, the common symbol section to be processed is set to the configuration as in the above example, and FIG. It is assumed that the attribute definition information has the configuration as illustrated in ().

In FIG. 3, the common symbol management unit 6 is activated by receiving a processing request by designating a common symbol section and a common symbol from the execution format program generation unit 4, and starting from the common symbol section to be processed in processing step 10. Fetch attribute definition information for the specified common symbol.

It is assumed that the extracted items of the attribute definition information are sequentially processed from the first item (the item of "basic attribute 1" in FIG. 2 (b)), and the same item is set for all the attribute definition information in processing step 11. If there is, compare whether the contents are the same.

In processing step 12, it is identified whether or not there are a plurality of attribute definition information items having the same item in the comparison processing, and if there is a plurality, it is identified in processing step 13 whether or not all of their contents match, and if they match. Identifies the next item in process step 14 and if there is a next item, process step
Return to 11.

Therefore, as long as a plurality of attribute definition information have the same item and the contents thereof match, the above processing from processing step 11 to processing step 14 is repeated to sequentially check each item.

The processing proceeds in this way, and the processing steps
In step 12, it is detected that only one of the attribute definition information has an item to be currently processed, or in processing step 14, it is detected that the item of any attribute definition information has ended The iterative process for successfully ends.

That is, any of the above means that
If there is a match in the definition content between all attribute definition information for all items, or if the number of items differs among the attribute definition information, the definition matches among the attribute definition information with a large number of items, and the number of items This is the case when the item is not defined with a small amount of attribute definition information.

Since the object attribute generator 1 checks the data attribute definition for each program during the object program generation process, the basic attribute must be correctly defined in each program. The success of the above process means that all items of the basic attribute match among all the attribute definition information.

If successful as described above, in processing step 15, one of the attribute definition information items whose items have not been completed until the final round of the repeated comparison process is selected. That is, the attribute definition information with the longest number of items is selected, and the other attribute definition information having items corresponding to each item of the attribute definition information is selected because the corresponding definition contents of all items are the same. The attribute definition information is equal to the result of merging all the attribute definition information.

After the above, in the processing step 16, the common symbol management section 6 notifies the execution format program generation section 4 of the common symbol section having the selected attribute definition information as a selection result and ends the processing.

If it is detected in step 13 during the comparison process that the definition contents of a certain item do not match, it is determined that the common symbol section has not been selected, and the process proceeds to step 17. The execution format program generation unit 4 is notified that there is a definition error, and the processing ends.

By the above processing, when the object program of the program illustrated in FIGS. 4A and 4B is combined and edited, the same PARTITON statement and LOCATE statement as in (a) are also used in (b). As in the case where is described, the combining process is performed normally.

Needless to say, the program of FIG. 4 (b) is shown in the figure when the program of FIG. 4 (a) has a simple data attribute definition without a PARTITON statement and a LOCATE statement. It is possible to deal with the same description content as it is, and it is possible to deal with the case where another additional attribute is defined.

[0048]

As is apparent from the above description, according to the present invention, when a program in which a common symbol is defined is combined and edited to be an executable program, a contradiction of attribute definition between programs is checked, As long as there is no contradiction, the data of the program that has the attribute definition with the most definition items as a result of merging multiple attribute definitions is selected as the data corresponding to the common symbol. If a definition conflict is detected and appropriate action can be taken, and if at least one of the programs to be combined and edited has a complete attribute definition, the other attribute will not conflict with the attribute definition. In some programs, attribute definitions can be omitted or defined so that they can be merged with multiple different attribute definitions. There is a significant industrial effect that kill.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of the present invention.

FIG. 2 is a diagram illustrating a common symbol section.

FIG. 3 is a flow chart of processing of the present invention.

FIG. 4 is a diagram showing an example program.

[Explanation of symbols]

 1 Object Generator 2 Source Program 3 Object Program 4 Executable Program Generator 5 Executable Program 6 Common Symbol Management 7 Common Symbol Sections 10-17 Processing Steps

Claims (1)

[Claims] 1. An object generation unit (1) analyzes a source program (2) written in a predetermined programming language to generate a target program (3), and an execution format program generation unit (4) includes a plurality of the programs. An apparatus for combining and editing an object program to generate an executable program, comprising a common symbol management unit (6), wherein the object generation unit (1) is a predetermined symbol related to a common symbol in the source program (2). , A common symbol section (7) is generated in the target program (3), a data area and attribute definition information regarding each common symbol are set in the common symbol section, and the execution format program generation unit (4) uses the common symbol section selected by the common symbol management unit (6) when the target program (3) to be combined and edited has the common symbol section (7). Processing the production of said executable program relating to the common symbol, the common symbol managing unit (6), for each said common symbol, a plurality of said purpose program having the common symbol
Regarding the common symbol section (7) of (3), the attribute definition information about the common symbol of all the common symbol sections is compared for each item, and if there is no contradiction between all the items, all And selecting the common symbol section that has the result of merging the attribute definition information of
A compiler device configured to notify (4).