JPS62171036A - Module sharing system for compiler - Google Patents

Abstract

PURPOSE:To simplify a system structure and to substantially reduce the linking and editing time by unifying functional modules necessary for a compiler and a user program so as to share it. CONSTITUTION:At the time of initialization, the compiler 20 loads a shared I/O library 23 in a memory. In order to input and output a card, the syntax analyzing and processing part 21 of the compiler main body issues an I/O request to an I/O control routine 22. It knows the address of an I/O processing module corresponding to an input or output from an address control table 24 in the library 23, calls and executes one corresponding I/O processing module. Upon the completion of the I/O processing, the routine 22 returns the I/O processed result to the syntax analyzing and processing part 21 in a format that said part 21 easily processes.

Description

[Detailed Description of the Invention] [Summary] When translating a user program, if the translated user program itself requires functions such as input/output processing and function calculations required by the compiler, a module that performs the processing is required. Unify and share.

[Industrial application field]

The present invention relates to a program processing technique in an information processing system, and particularly to a module sharing method when translating and generating an executable program from a source program.

[Conventional technology]

As shown in FIG. 4(a), a conventional compiler is equipped with a dedicated input/output processing function for manually inputting a user program in the form of a source program and outputting a list of translation results.

On the other hand, the READ and WRITE statements included in the user program translated by the compiler are shown in Figure 4 (
As shown in b), when editing the linkage editor, the input/output library (module group for input/output processing)
It is made executable by combining it with

In addition, the compiler uses known calculations such as square roots and trigonometric functions in the source program as one of the optimization techniques, and calculates the values necessary for those calculations.

When an argument is given as a constant, a method is used in which calculation is performed at the translation stage and the resulting value is replaced with the constant. This is effective in making the program more efficient.

For this purpose, the compiler has modules for calculating functions such as square roots and trigonometric functions. In addition, each function calculation that could not be executed at the translation stage (for example, one whose argument is a variable) is combined with an automatic call library containing those calculation modules by combination editing and executed. FIG. 5 shows the case of a square root (expressed as SQRT).

[Problem that the invention seeks to solve]

Conventional compilers have dedicated modules for input/output processing, functional calculations, etc.

Modules with the same functions that are necessary for processing the user program itself had to be combined and edited from a separate input/output library or automatic call library.

However, in general, combined editing requires a considerable amount of time for the opening process that makes the library file available for use, and modules with the same functionality are prepared twice for the two stages of translation and user program execution. For,
There were problems in that the configuration became redundant and complex, maintainability deteriorated, and the development burden on the compiler increased.

[Means for solving problems]

The present invention solves the above-mentioned problems of the conventional system by unifying and sharing modules of functions necessary for both the compiler and the user program.

The present invention will now be described in detail with reference to the exemplary configuration shown in FIG.

FIG. 1 shows an example in which input/output processing function modules are unified, where ① is a source program, 2 is a compiler, 3 is a translation process, 4 is an object program, and 5 is an input/output library which is a shared module.

6 is an address management table, and 7 is a READ processing unit.

8 represents a WRITE processing unit, and 9 represents an execution process.

The input/output library 5 is configured as one load module, in which an address management table 6 is placed at the head thereof, and a READ processing section 7 and a WRITE processing section 8 each having a manual processing function and an output processing function. Contains modules.

The address management table 6 is an R
The address of the EAD processing section 7 and the address of the WRITE processing section 8 corresponding to the WRITE statement are set.

When the compiler 2 outputs the manual input or list of the source program 1, it is written in the same format as the source program 1, that is, in the same way as before, using READ statements or WRIT statements.
E sentence is written.

When these READ or WRITE statements are issued, the address management table 6 of the input/output library 5 is referred to and the intended READ or WRITE processing section is activated.

Compiler 2 again. When generating an object code string, that is, an object program 4.

READ processing unit 7 or WRIT of input/output library 5 as definition data of READ statement or WRITE statement.
The address of the E processing section 8 is set and output.

After this, the object program 4 is executed.

Although the above configuration is related to the READ/WRITE input/output processing function, it can be similarly applied to square root calculation and other arbitrary functions.

[Effect]

According to the present invention, in the object program generated by the compiler, the address of the module of the function shared with the compiler is directly given, so that the library can be transferred via the name to a member on the DASD, which is conventionally called a V-type atcon. Compared to calling the link, the linkage editor's combined editing process is extremely simple, and the processing time can be shortened.

(Example〕 Next, the present invention will be explained in detail using examples.

FIG. 2 shows the flow of processing when the input/output library is used as a shared module, similar to the example shown in FIG.

In FIG. 2, 20 is a compiler, 21 is a syntax analysis processing section of the compiler main body, and 22 is an input/output control routine that is one of the common routines that support the compiler main body.
23 is a shared input/output library, 24 is an address management table, and 25 is a group of input/output processing modules. Also, ■ to ■ indicate the processing order.

(1) First, the compiler 20 loads the shared input/output library 23 into memory during initialization.

(2) The syntax analysis processing unit 21 of the compiler body issues an input/output request to the input/output control routine 22 for card input (READ) or list output (WRITE).

■ The human output control routine 22 selects READ or WRI from the address management table 24 of the input/output library.
The address of the input/output processing module corresponding to the TE is known, and one corresponding input/output processing module is called and executed.

(2) Upon completion of the input/output processing, the input/output control routine 22 returns the input/output processing results to the syntax analysis processing section 2L in a form that is easy to process.

Furthermore, if the square root module is a shared module, similar operations will be performed in response to a square root calculation request.

For example, if there is a square root calculation with a constant argument such as cubic in the source program.

=SQRT(2,0) As shown in FIG. 3, the address of the square root module 33 is obtained from the address management table 32 of the shared module 31, the square root module 33 is called, and /2.
We get 0=1.414.

On the other hand, if the source program contains a square root calculation of the 5th degree with variables as arguments, A = 2.0 = SQRT (A) The compiler will not branch to the shared square root module 33 in the generated object program. Let them do it. In this case, the address of the square root module 33 is directly set as shown below.

L12,5QRT BALR14,15 SQRT DCA (Square root module address) [Effects of the invention] According to the invention of the method, the system configuration is simplified and the combined editing time is significantly reduced (almost negligible). Moreover, the content of the object code remains the same as before (just set the direct address instead of the V type atcon), and the compiler has dedicated input/output processing functions, square root calculation functions, etc. Because there is no need to prepare.

Compiler creation and maintenance become easier.

[Brief explanation of drawings]

FIG. 1 is an explanatory diagram of the principle of the present invention, FIG. 2 is a flow explanatory diagram of one embodiment of the present invention, FIG. 3 is an explanatory diagram of an example of sharing the square root module according to the embodiment of the present invention, and FIG. 4 is a diagram of the conventional method. An explanatory diagram of a compiler's input/output processing method. FIG. 5 is an explanatory diagram of the conventional square root module combination method. In Figure 1. l; User program 2: Compiler 4: Object program 5: Shared module input/output library 6: Address management table 7: READ processing section 8: WRITE processing section

Claims (1)

[Claims] In an information processing system equipped with a compiler (2), when the compiler (2) and the user program (1) translated by the compiler (2) use the same processing function, The processing functions are provided as a shared module, the shared module has an address management table (6) in which addresses for accessing the individual processing functions are set, and the compiler (2) is configured to handle the common processing functions. When using the above shared module address management table (6)
If the user program (1) requires the above-mentioned common processing function, the object program ( 4) A compiler module sharing method characterized by generating.