JPH06282442A - Compiler device - Google Patents

Abstract

PURPOSE:To obtain the compiler device which improves translation processing for an input/output statement which is specified in array format in the translation processing for a source program of a computer. CONSTITUTION:An analytic part 1 analyzes the source program and specifies elements within a range determined by the execution result of programs in a desired array as input/output data as to the input/output instruction statement of the source program; when the specification of an error branch destination in the case of error detection by input/output processing is discriminated, the analysis result of the input/output statement is sent to an insertion part 2, which inserts an insert program for branching to an error branch destination after processing for releasing a storage area obtained as a work area for holding the input/output data as to the analysis result of the input/output statement and changes the branch destination at the time of the error detection of the input/output statement to branching to the entrance of the insert program. Then a translation part 3 performs translation processing as to the result of processing of the insertion part 2.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a compiler device for improving the translation process of an input / output command statement in which an array expression is designated as input / output data in the process of translating a source program of a computer into a target program.

[0002]

2. Prior Art and Problems to be Solved by the Invention FIG. 2 (a)
Shows an example of a WRITE statement, which is one of the input / output command statements, using the FORTRAN programming language as an example, and the example of the figure shows “element A (I) of array A and element B (I of array B ) Is output to the device having the device reference number “1” ”.

When this statement is analyzed by the compiler for translation, the compiler provides a working area, which is identified by the so-called generated name S0001, as shown conceptually in FIG. 2 (b). Substituting (I) + B (I) into S0001, S0001
Output the contents of. Here, the size of the area of the generated name S0001 is obviously the size of one element of the array A or B.

[ER] specified by the ERR specifier in the WRITE statement
R = 999 "specifies that if an error is detected in the I / O processing by the WRITE statement, as is well known, the WRITE statement is not terminated and the processing proceeds to the next statement, and the processing branches to the label" 999 ". To do.

When the data to be output as in the above example is continuous as A (1) + B (1), A (2) + B (2) .... , Write a source program so that I is changed sequentially by a DO loop or the like and repeatedly executed, and in that case, the compiler creates a generated name S0 as a work area.
Make sure to use the area 001 repeatedly.

[0006] However, a language specification is provided in FORTRAN or the like that enables a description that directly specifies a continuous range of array elements without using such a DO loop. For example, "A (1:
3) ”makes it possible to specify the elements A (1), A (2), and A (3) of the array A, which are used to specify the data of the input / output statement, and You can now write programs like (a).

The line of this program is a declarative statement of the sub-program, the sub-program name is SUB, and the dummy arguments are variables A, B, I, and J. A line is an array declaration statement, A and B are arrays, and * indicates that the number of array elements is determined at the time of execution. The data type is implicitly A, B
Is a 4-byte real type, and I and J are 4-byte integer types.

The line is a WRITE statement similar to the above, but since the data to be output is specified in the form of an array expression, the elements of arrays A and B are from A (I) + B (I) to A (J ) + B (J) are output targets. That is, if I = 1 and J = 3 are specified in the call of this subprogram, A (1) + B (1),
A (2) + B (2) and A (3) + B (3) will be output.

A line is an instruction statement for outputting a character string to a display or the like by a PRINT statement. In this example, the character string "OK"
!! Output is specified. The line is the branch destination when an error occurs in the WRITE statement, and since it is a well-known CONTINUE statement in this example, nothing is done in this statement itself, and it is placed so as to skip the PRINT statement of the line in the case of an error. Yes, but put error handling in this position if necessary. The last line is an END statement that marks the end of the subprogram.

The compiler uses the WR of such a program.
When the analysis result of the ITE sentence is conceptually shown, for example, FIG. 3 (b) is obtained. The line in Fig. 3 (b) is a GET statement that acquires a storage area.
Acquire the area with the generated name SOOO1.

~ Line is A (I) + B in each area of S0001
(I), A (I + 1) + B (I + 1) ..., A (J) + B (J) assignment statement for storing, line is a WRITE statement that outputs the contents of S0001 , Line is the branch destination in case of error in WRITE statement. By doing so, even when the output data is specified by the array expression, the WRITE statement may have the same execution content as in the case of FIG. 2, and the interface with the WRITE statement is the same as the conventional one.

However, unlike the example of FIG. 2 described above, FIG.
In the case of the example of FIG. 3, the area of the generated name S0001 is not provided at the time of translation and is acquired at the time of execution without providing the area of the generated name S0001. Because the number of data elements is determined, the size of S0001 is determined only at the time of execution.

Therefore, in this case, unless the area that has been acquired must be returned to the system at the end of the WRITE statement, the acquired area is accumulated every time such a WRITE statement is executed, and the system manages it. If the storage area is compressed and the effect is severe, the program may not be able to be executed, and it is necessary to provide the area release processing at the required location.

However, it is of course not necessary for the programmer to be aware of this, but for changing the interface with the existing input / output statements such as the WRITE statement prepared as a so-called language library and the execution contents thereof. It must be avoided as it will affect the program.

According to the present invention, the compiler device for dynamically releasing an area for an input / output statement as described above is processed in a translation process while leaving the existing input / output statement as it is. With the goal.

[0016]

FIG. 1 is a block diagram showing the configuration of the present invention. The figure shows a configuration of a compiler apparatus, which is an apparatus for translating a source program written in a predetermined programming language into a target program, and has an analysis unit 1, an insertion unit 2, and a translation unit 3.

The analysis unit 1 analyzes the source program, and the input / output command statement of the source program specifies, as input / output data, an element in a range determined by the execution result of the program in a required array, and outputs the input / output. When it is identified that the error branch destination is specified when an error is detected by the processing of 1, the analysis result of the input / output statement is passed to the insertion unit 2.

The inserting section 2 inserts the analysis result of the input / output statement so as to branch to the error branch destination after executing the processing of releasing the storage area acquired as the work area for holding the input / output data. The program is inserted, the branch destination at the time of error detection of the input / output statement is changed to the branch to the entrance of the inserted program, and the translation unit 3 executes the translation process on the result processed by the insertion unit 2.

[0019]

With the compiler apparatus of the present invention, the dynamic area acquisition and release processing for an input / output statement is automatically incorporated without the need for the programmer to describe, and the processing inserted for that is the input / output statement's For external I / O statements, the branch destination specified by the ERR specifier etc. of the I / O statement is only modified so as to go through the area release processing. Does not need to be changed at all, so its use of the library does not affect other programs at all.

[0020]

DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 4 is a diagram showing an example of the processing flow of the compiler apparatus of the present invention. The analysis unit 1 in FIG. 1 inputs each command statement in a processing step 10 for a given source program input, and in the processing step 11, executes the analysis processing as in the conventional case.

The analysis result is passed to the translation unit 3. However, in processing step 12, it is identified whether the analyzed statement specifies input / output data by an array expression and whether dynamic area acquisition is required. The result is passed to the inserting unit 2, and otherwise it is directly passed to the translating unit 3.

When the inserting section 2 receives the analysis result as conceptually shown in (b) as the analysis result of the WRITE statement of FIG. 3 (a), it generates the area to be acquired by the GET statement in the processing step 13. The name is fetched, the ERR specifier specification of the WRITE statement is fetched in processing step 14, and the designation destination of the ERR specifier is changed to the entry point of the insertion program.

Next, in processing step 15, after the area release processing of the area of the extracted generated name is performed, an insertion program that branches to the original designated destination of the ERR specifier is inserted into the analysis result and translated. Hand over to Part 3.

As a result of the above processing, the analysis result of FIG. 3B is changed as illustrated in FIG. Line in Figure 3 (c)
The lines correspond to the same line in (b), lines to lines are insert programs, and the line in (c) corresponds to the line in (b).
This is the specifier specification destination and the branch destination at the time of error.

If there is an unprocessed statement that is identified in processing step 16, the process returns to processing step 10 and when the analysis of the source program is completed, the translation unit 3 translates the analysis result into the target program in processing step 17. , Output in process step 18.

[0026]

As is apparent from the above description, according to the present invention, in the translation process of the source program of the computer,
Since the dynamic area acquisition and release processing for I / O statements is automatically incorporated, and because there is no need to change the contents and interface of the I / O statements of the library, there is no need to use the library. This has the effect of not affecting other programs.

[Brief description of drawings]

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

FIG. 2 is a diagram illustrating an example of an input / output statement.

FIG. 3 is a diagram illustrating a program example.

FIG. 4 is a flow chart of processing of the present invention.

[Explanation of symbols]

 1 Analysis part 2 Insertion part 3 Translation part 10-18 Processing steps

Claims (1)

[Claims] 1. An apparatus for translating a source program written in a predetermined programming language into a target program, comprising: an analysis unit (1), an insertion unit (2), and a translation unit (3). The analysis unit (1) analyzes the source program, the input / output command statement of the source program specifies, as input / output data, a range of elements determined by the execution result of the program in a required array, and When it is identified that an error branch destination is specified when an error is detected by the processing of, the analysis result of the input / output statement is passed to the inserting section (2), and the inserting section (2) For the analysis result of the I / O statement, insert an insertion program that branches to the error branch destination after executing the process of releasing the storage area acquired as the work area for holding the I / O data, Branch destination when an error is detected in the output statement Change to branch to the inlet of the insertion program, the translation unit (3), the compiler apparatus characterized by being configured to perform the translation processing for a result of processing of the insertion section (2).