JPH06266595A - Data dictionary constituting method - Google Patents

Abstract

PURPOSE:To rationalize data analysis or programming by using a dictionary at the time of the development or maintenance of a large-scaled computer system. CONSTITUTION:The data dictionary is divided at every sub-system 10 and 20 further divided into information before preliminary compile(dictionary source) and information after the preliminary compile(dictionary load), and respectively held in a dictionary source storage file 111 and a dictionary load storage file 112. The registration and correction of the dictionary source is operated by using an editor 113. Also, reference to a data item registered in the dictionary of its own sub-system or the other sub-system is included in the content of the dictionary source. A program 114 preliminarily compiles the dictionary source, and prepares the dictionary load. At that time, the program 114 refers to the dictionary load in the case that the reference to the dictionary of the other sub-system is available. A program 115 refers to only the dictionary load at the time of using the dictionary.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method of constructing a data dictionary for holding data items used in the system, their attributes and meaning contents in the development and maintenance of an information processing system, and in particular, the number of systems The present invention relates to a data dictionary construction method suitable for development and maintenance of a large-scale system that is divided into sub-systems and separately developed.

[0002]

2. Description of the Related Art The importance of previously analyzing and managing data items used in a system has been recognized. Take Susumu Yamada, "Introduction to Information Resource Management" (Ohmsha,
Its importance is also described in Section 2.3 of (1987), and a specific example of constructing a data dictionary (data item database) is described in Chapter 6 thereof.

However, the conventional idea including the above-mentioned documents is based on the premise that all the data used in the system are preliminarily and unitarily analyzed in units of items. It is intended to identify the data item used. In other words, data items of the same meaning accessed in the program should be designated with the same system name throughout the system (hereinafter,
"Name uniqueness"). Many data dictionaries support the function of registering an alias in addition to the real name for a data item, but this is merely for convenience of handling and does not violate the uniqueness of the name.
As described above, conventionally, attributes, meanings, and the like have been registered in a dictionary and searched for uniformly defined data items.

On the other hand, when a dictionary is used, various measures are taken in order to give unique names to all data items in the system in this way, and therefore, it is common to set rules for giving names. Is. There are several ways to give this convention, for example, the main word,
It is most common to give a name in three classifications, a modifier and a classification word, and some data dictionaries also presume naming based on these three classifications.

Information about how data items are arranged to form one aggregate or record is not stored in the dictionary but is separately managed by a copy library or the like.

[0006]

In the conventional data dictionary, all the data used in the system are unitarily analyzed in advance in item units and given a unified name, which causes the following problems.

(1) Difficulty in data analysis In a large-scale system, the number of data items registered in the dictionary is tens of thousands, which exceeds the limit of management by one person in charge of the dictionary. Also, in a large-scale system, it is common to divide the system into several subsystems and launch them separately, and each subsystem has its own purpose and is caused by each business requirement. There are data structure systems and naming conventions. Under such circumstances, it is extremely difficult to preliminarily and centrally analyze all the data items used in the system and give them unique names.

(2) Difficulty in coping with distributed development In recent years, workstations have made remarkable progress, and in the future, it is considered that system development will also be carried out in a distributed manner at a plurality of development sites by utilizing workstations. However, since the conventional data dictionary centrally manages all the data used in the system, it is difficult to distribute the dictionary to each development place. Therefore, if the central host computer has the dictionary all at once, the development environment depends on the host computer, and the merit of the distributed development environment is diminished.

(3) Redundancy of data name Since a unique name is given throughout the system, the data name becomes redundant. Although the rule of names using main words, modifiers, and classification words is effective for making data names unique, data names become long, and coding becomes complicated and difficult to see. Furthermore, because many compilers do not allow Japanese data names, it is necessary to understand such redundant names without the help of Japanese, and it becomes difficult to understand the data items.

(4) Restriction on securing a plurality of work areas In programming, a plurality of areas are often secured for input, output, work, etc. in order to make data having the same attributes and meaning contents. . However, in the conventional data dictionary, it is assumed that all data items with the same meaning are specified with the same name (uniqueness of the name), but if you try to secure multiple areas, the names will be duplicated. I will end up. Therefore, for example, it is necessary to take an unfavorable countermeasure such as excluding the work area from the dictionary management target or registering another item for the work in the dictionary.

(5) Lack of ability to express correspondence with aggregates In the conventional dictionary, the aggregate (C
There is a lack of thinking about the set item in the OBOL language. Therefore, the meaning content that is determined only after being taken into the aggregate cannot be registered in the dictionary. For example, SHAIN-SHIMEI
-The meaning of the KANJI data item can only be registered as "employee name displayed in kanji". This is UR
If it is included in the IAGE record, it actually means "the employee in charge of this sales",
Such meaning content cannot be registered. In addition, the check condition for the data item is often registered in the data dictionary, but in the conventional dictionary, only the check condition closed in the item can be registered, and the check condition between data items determined only after being incorporated in the aggregate is registered. The check conditions for association cannot be registered. For example, it is possible to register that the delivery date should be valid as a date (for example, the month is 01 to 12), that it should be after 1990, but not that it should be after the order date. . Further, since each data item has a unique name, a part of the name of the data item often overlaps with the name of the aggregate. For example, S
If the data item HAIN-SHIMEI-KANJI is in the SHAIN record, the whole name including the aggregate modification is SHAIN-SHIMEI-KAN.
It becomes JI, and the SHAIN notation overlaps.

(6) Separate management from COPY library Many compilers can specify copying a part of text from another library. CO in COBOL language
The PY function, the% INCLUDE function of the PL / I language, and the macro function of the assembler language correspond to this. This function is also usually used to centrally store information about data outside the program, and in that sense, it has a purpose similar to that of a data dictionary. However, since the conventional data dictionary focuses on each data item, it cannot have information about the arrangement of the data items. Therefore, the double management is inevitable that the information enclosed in the data item is in the data dictionary, and the information regarding the arrangement of the data is held separately in the COPY member.

An object of the present invention is to solve each of the problems described above, and alleviates the constraint of uniqueness of names within a range that does not hinder the understanding of meaning contents, that is, data items having the same meaning contents. Even so, it is to provide a data dictionary construction method that allows different names to be called in a program under certain conditions.

[0014]

In order to achieve the above-mentioned object, the invention of claim 1 does not directly specify the attribute of a data item by the data item, but another data item registered in the dictionary. The fact that the attribute of should be referred to can be specified. If the data dictionary also supports registered contents (semantic contents, check conditions, etc.) other than attributes, the same shall apply to them. Track other specified data items to obtain the required information. Since it is possible that another data item may refer to another data item, the data item that is finally referred to is traced in order. At this time, the attribute is specified only in the finally referenced data item, but the semantic content and the check condition can be specified in the data item at each stage of tracking.

According to the second aspect of the present invention, a plurality of independently managed dictionaries are provided, and the reference to the attribute of the data item registered in the other dictionary is designated as the reference to the attribute of the other data item. That is to make it possible. In a large-scale system, it is usual to divide the system into several subsystems and develop them separately, so it is convenient that the dictionary can be managed independently for each subsystem. The plurality of independently managed dictionaries referred to here mean dictionaries managed by each subsystem in such a situation.

According to the third aspect of the invention, when a reference to a data item registered in another dictionary is designated, necessary information is fetched from another dictionary by precompilation of the data item, and when the program is compiled, the other information is fetched. That is, the reference to the dictionary is suppressed. Pre-compilation may be simply a matter of gathering the required information, or it may involve some parsing.

According to the invention of claim 4, when the data item is accessed in the program, the procedure division of the program can use a program access name different from the dictionary search name instead of directly using the dictionary search name. is there.
This is because the construction of the dictionary and the creation of the program are performed by different persons with different purpose meanings. That is, the dictionary search name is given from the perspective of the entire system or subsystem because the dictionary is built by a systems engineer or a senior programmer, but the program access name is the program access name because the program is created by a general programmer. It is given from the standpoint of readability and readability. Information on the correspondence between the program access name and the dictionary search name is given in the data declaration section of the program.
In order to use this information in the procedure division as well, the compiler may have the function or a precompiler may be used.

The invention of claim 5 is such that information about an aggregate can also be registered. That is, the data items registered in the data dictionary may be aggregates instead of individual items. At this time, the registered content for the data item has a function similar to that of the COPY member. In the present invention, the order of arrangement of the constituent elements that make up the data item is registered, and the name and attribute of each constituent element, as well as the semantic content and check conditions are registered. The name registered here is not the dictionary search name but the program access name. That is, the programmer uses the name registered here to refer to a specific component in the aggregate. You can also directly specify the attributes registered here.
Often, references to other data items are specified. At this time, if the other specified data item is an aggregate, nesting of the aggregate is designated.

[0019]

How the present invention solves the above problem will be described below.

(1) Data analysis In the invention of claim 1, by referring to other data items registered in the dictionary for attributes and meaning contents of the data item, the uniqueness of the name is relaxed. , Data having the same meaning can be referred to by different names.
For example, if the data item A is referenced by the data item B, the data having the same meaning can be referenced by both the name A and the name B, and the uniqueness of the name is relaxed. . Even in this case, since the mutual relationship between these data items is held in the dictionary, the relationship between the attributes and the meaning contents will not become unclear. Further, according to the invention of claim 2, a data item registered in another dictionary (that is, a dictionary managed by another subsystem) can be designated as a reference to another data item. It is possible to do it by the subsystem independently, and then refer to the dictionaries of other subsystems only for the data items (usually 10 to 30% of all dictionary items) that need to be shared. In this way, it is possible to limit the range managed by one person in charge of the dictionary, and to respect the system structure and naming conventions of the data structure caused by the subsystem-specific business requirements. This alleviates the data analysis difficulties inherent in large systems.

(2) Correspondence to distributed development If a reference to an attribute of a data item registered in another dictionary is specified, when a dictionary is used (for example, when compiling a program), it should be originally. You will also need other dictionaries specified. However, according to the third aspect of the present invention, since necessary information is fetched from another dictionary by precompiling the data item, the dictionary of the own subsystem is all that is required to use the dictionary. Since it is known which data item of which other subsystem the data item of its own subsystem should refer to, ask the dictionary person in charge of the corresponding other subsystem in advance to send the contents of the data item, Precompile using the contents. The dictionary of the local subsystem is updated for the first time when the dictionary person in charge of the local subsystem performs preliminary compilation using the contents sent. This can increase the independence of each subsystem.

(3) Simplification of data name According to the invention of claim 4, instead of directly using the dictionary search name in the procedure part of the program, a different program access name is used, so that the program tends to be redundant. It is released from the dictionary search name. Even in this case, since the information regarding the correspondence between the program access name and the dictionary search name is given to the data declaration part of the program, the relationship between the attribute and the meaning content is not unclear.
This allows the program access name to be given a simple name. Moreover, since the dictionary search name is not used directly in the program, it is not subject to naming restrictions by the compiler. So to help you understand,
You can give a name in Japanese.

(4) Securing a plurality of work areas Further, instead of directly using the dictionary search name in the procedure part of the program, a different program access name is used, so that the program access name has a one-to-one correspondence with the meaning content of the data item. There is no need to deal with. That is,
In the data declaration part of the program, the same dictionary reference name may be associated with a plurality of program access names. The program access name has a role of uniquely designating an area in the program, and the dictionary reference name has a role of designating attributes and meaning contents of data in the area.
In this way, even if a plurality of areas having the same attributes and meanings are secured, the names do not overlap.

(5) Ability to express correspondence with aggregates According to the invention of claim 5, the information about the aggregates can be registered in the dictionary, so that the representation of the aggregates can be performed accurately. For example, in the example given above, registration is performed as shown in FIG. In the same figure, the fact that the attribute is surrounded by [] means that other data item should be referred to for the attribute of the data item. As a result, "the employee in charge of this sales"
Meaning meaning can be registered. Further, when the data item is an aggregate, since the check condition can be specified not only for each component but also for the entire data item, it can be registered that the delivery date should be after the order date. Further, even if the SHAIN-KANJI data is captured in the SHAIN record, the whole name seen from the program is SHAIN. Because it will only be SHIMEI, SHA
There is no duplication of the IN notation. In addition, URIAGE
Both the "employee responsible for this sales" and the "employee name displayed in kanji" are necessary to accurately grasp all the meanings and contents of the SHAIN data captured in the record. To this end, the meaning and the like can be designated by the data items at each stage, as shown in the figure, and they can be referred together.

(6) Management Separately from COPY Library Furthermore, since the information on the aggregate can be registered in the dictionary, the dictionary according to the present invention can also have the function of the COPY library. This makes it possible to unify the dictionary and the COPY library.

[0026]

An embodiment of the present invention will be described in detail below with reference to the drawings.

FIG. 1 is a block diagram of the first embodiment of the present invention.
It is a block diagram which shows the Example which applied this invention to the data dictionary managed by the host computer.

In FIG. 1, 10 and 20 are subsystems, and the entity of the dictionary is divided for each subsystem,
Further, the information before pre-compilation (hereinafter referred to as “dictionary source”) and the information after pre-compilation (hereinafter referred to as “dictionary load”) are held separately. Each subsystem 1
The dictionary source storage file 111 and the dictionary load storage file 112 of 0 and 20 show this. In the present embodiment, both the dictionary load storage file 111 and the dictionary load storage file 112 are divided data sets, and are composed of directories 111a and 112a and bodies 111b and 112b, respectively. In addition to the real name (for example, "sales slip"), an abbreviation within 8 digits of Chinese characters (for example, "URADPY") is given to the dictionary search name, and this abbreviation is used as a member name of the divided data set. The search key (real name of the dictionary search name) is held in the user area of the directory of the relevant segment data set, and when searching, the entire directory is read once and sorted in the order of the search key. In this way, a simple partitioned data set is used instead of having a complicated structure for searching the dictionary data, and the structure can be simplified, so that the overhead of dictionary reference can be reduced. It should be noted that a library management system that provides the same function as the partitioned data set access may be used instead of the normal partitioned data set.

Unlike the conventional dictionary, the dictionary source of the dictionary source storage file 111 has a plurality of lines. Therefore, in this embodiment, the registration / correction of the dictionary source is performed by each subsystem 1 instead of preparing a dedicated registration / correction program.
The function of the full screen editor 113 is utilized for each 0 and 20 so that it can be easily performed. As a result, the definition contents in the dictionary source have a free format to some extent, and therefore it is necessary to perform a syntax analysis in order to use the definition contents. This is done by the preliminary compilation processing program 114 of each subsystem 10, 20. Do.

The preliminary compilation processing program 114 is
When importing the dictionary information of another subsystem at the time of pre-compilation, refer to the dictionary load of the dictionary load storage file 112 (information after pre-compilation) instead of the dictionary source of the dictionary source storage file 111 (information before pre-compilation). To do. Further, if there is a reference to another data item from the information of another dictionary that has been imported, that information is also imported from the dictionary load. When the dictionary compiling processing program 114 loads a dictionary of another subsystem, it accesses it by using another subsystem name (as a so-called dd name). Therefore, the subsystem name should be 8 alphanumeric characters or less. Other programs that use dictionaries 115
Refers only to the dictionary load of the dictionary load storage file 112. After all, the dictionary source of the dictionary source storage file 111 is referred to only when the relevant member is precompiled.

FIG. 2 shows the data dictionary of this embodiment.
An example of contents registered as a dictionary source in the dictionary source storage file 111 is shown. This is a feature of the present invention,
It includes examples of references to other data items registered in the dictionary, references to data items registered in other dictionaries, and registration of aggregated data items.

In FIG. 2, for example, the sales subsystem dictionary has a sales slip (abbreviation URADPY), sales details (abbreviation URAMSI), product quantity (abbreviation SHOSU).
R), order number (abbreviated name JCHNO), customer code (abbreviated name KKYCD), and product name code (abbreviated name HINCD), six data items are registered, of which the previous three (201 in the figure) are aggregates. Other data items (202 in the figure) are data items that are not aggregates. The data items of the sales slip include JUCHU-NO, NOUNYUU
-SAKI, JUCHUBI, NOKI, TANTO,
There are six MEISAI components, arranged in this order. The attributes of these constituent elements are not directly specified in this data item, but are specified so as to refer to other data items registered in the dictionary. That is, "% order number", "% customer code", and "% sales details" should refer to the data items "order number", "customer code", and "sales details" in the dictionary of the subsystem. "KYOTSU% date" and "JINJI% employee code" indicate that the data items "date" in the common dictionary and "employee code" in the personnel subsystem dictionary should be referred to, respectively. The meaning content is specified by enclosing it in [], and the check condition is specified by enclosing it in CK []. These are given to each component and also to the entire data item that is an aggregate. The same applies to the registered contents of the dictionary of the personnel subsystem and the common dictionary. In this example, the program access name is expressed in Roman letters and the dictionary search name is expressed in Japanese.

FIG. 3 shows an example of a dictionary load record format in the dictionary load storage file 112. This dictionary load is generated by the pre-compilation processing program parsing the dictionary source in the format of FIG. 2 and fetching information from another dictionary (dictionary load of another subsystem) as needed. It The dictionary load is a separate dataset from the dictionary source, and the member name is the same as the member name of the dictionary source.

In the present embodiment, the dictionary load consists of 6 types of records in the format of FIG. 3, and each record is distinguished by the leading identifier. In FIG. 3, 301 is a header record, 302 is a fetch record, 303 is a data item record, 304 is a component record, 305 is a check condition record, and 306 is a member end record. These records have a variable length, and in fact, a field indicating the length at the beginning of the record (so-called record description word)
However, it is omitted here. Header record 30
Only one is generated at the beginning of the dictionary load, and 1 has information about the original dictionary source member that was the subject of preliminary compilation. The fetch record 302 is generated each time information is fetched from the dictionary of another subsystem, and has information on the fetched members. Data item record 30
3 is generated for each data item (that is, one member of the dictionary source), and has attributes (direct designation or reference to other items), meaning content, etc. of the data item. However, when the data item is an aggregate, since an attribute is given to each component, a value such as '% @' is set as a dummy in the attribute field.
The component record 304 is generated for each component of the aggregate. The check condition record 305 is generated corresponding to each check condition regarding the relation between the data items in the collection. The member end record 306 is
It corresponds to the data item record and indicates the end of one member of the dictionary source.

FIG. 4 shows the contents of the dictionary load generated when the data item "sales slip" in the dictionary source registration example shown in FIG. 2 is precompiled. Where 3
The meanings of 01 to 306 are as described in FIG. According to FIG. 2, six component records 304 are generated corresponding to the six components in the “sales slip”.
Furthermore, among these components, "JUCHUB
For “I”, “NOKI”, and “TANTO”, the attribute specifies a reference to the data items of the common dictionary and the personnel subsystem dictionary, so that the common dictionary (KYOTSU) follows the corresponding component record 304. ), And the content 402 from the human resources subsystem dictionary (JINJI) are inserted respectively. For "TANTO" in this, data items (human resources subsystem of another designated dictionary are included. A data item of another dictionary (“numerical code” of the common dictionary) is referred to by nesting from the “employee code” of the dictionary.
This way, nested data items are different (third)
If it belongs to the dictionary, the "employee code" of the personnel subsystem dictionary load already contains the "numerical code" of the common dictionary, so this is also taken into the dictionary load of the "sales slip" as it is. If the nested data item belongs to the HR subsystem dictionary, the nested data item is tracked in the HR subsystem dictionary and is also taken into the "sales slip" dictionary load. If the nested data item belongs to its own dictionary, that is, the sales subsystem dictionary, the information of the nested data item is not necessary for loading the "sales slip" dictionary, so it is deleted after the error check.

FIG. 5 shows an example of precompilation performed on the data declaration section of the program in order to use a program access name different from the dictionary search name in the procedure section of the program, and is shown in FIG. In the dictionary source registration example, the data item “sales slip” is used in the program. Here, "pre-compilation" refers to a pre-conversion operation of the program for incorporating the information of the dictionary into the program prior to compilation, and is different from "pre-compilation" which is an operation within the dictionary. In this example, in the data declaration part of the program before precompilation, the program access name "URIAGE1" is associated with the dictionary search name "sales slip". By precompilation, the information of the dictionary is taken in and expanded as a structure. Semantic contents and check conditions specified in the data items at each stage of tracking are displayed as comments. Note that this example is an example of the PL / I language.

The first embodiment makes use of existing tools such as a partitioned data set and a full-screen editor, so that the data dictionary can be constructed relatively simply. Further, since the target to be taken in at the time of preliminary compilation is not the dictionary source of another dictionary but the dictionary load, the following effects are obtained. (1) Preliminary compilation is efficient because only one member is subject to syntax analysis during preliminary compilation. Also, as the dictionary of other subsystems that should be prepared for pre-compilation, only the dictionary directly referenced by the target member needs to be prepared. Even if there is a reference to another dictionary due to nesting, it is not necessary to prepare it. . (2) The dictionary source is referenced only when the relevant member is precompiled. Since the pre-compilation is done by the dictionary person in charge of the subsystem, it prevents the contents from being mistakenly referred to while the dictionary is being modified.

FIG. 6 is a block diagram showing a second embodiment of the present invention, in which the present invention is applied to a distributed development environment centered on workstations.

The environment to which this embodiment is applied is a distributed development environment, and development is carried out at different development locations for each subsystem.
A local area network 601 is provided at each development site, and these are connected by a wide area network 602. The local area network 601 at each development location is provided with a workstation 603 for a person in charge of dictionary and a workstation 604 for a person in charge of program. Then, the dictionary source storage file 111, the dictionary load storage file 112, and the transfer information storage file 605 are placed for each subsystem under the control of the person in charge of the dictionary. The registered contents of the dictionary source storage file 111 and the dictionary load storage file 112 are the same as in the first embodiment. Information (dictionary load) transferred from another subsystem is temporarily stored in the transfer information storage file 605. The pre-compilation processing program is prepared on the workstation 603 for dictionary personnel.

The data item is registered in the dictionary of its own subsystem by the person in charge of the dictionary with respect to the dictionary source storage file 111, as shown at 610. When referencing a data item of a dictionary of another subsystem, ask the dictionary person in charge of the other subsystem to send the contents of the data item, and in response to this, take charge of the dictionary of the relevant subsystem. The person transfers the required dictionary load, as indicated at 611, and this is stored in the transfer information storage file 605.
The pre-compilation is performed using both the dictionary source storage file 111 of the own system and the transfer information storage file 605 from another subsystem, as indicated by 612. The program person uses the dictionary by using the dictionary load storage file 112, as shown at 620.

According to this embodiment, in a distributed development environment, even if the central host computer does not have a dictionary at once, it is possible to perform system development with consistency only by the dictionary for each development location.

[0042]

According to the invention of claim 1, the uniqueness of the name is relaxed, and the data having the same meaning can be referred to by different names.

According to the second aspect of the invention, data analysis can be performed in subsystem units, and the difficulty of data analysis unique to a large-scale system is reduced.

According to the third aspect of the invention, the independence of each subsystem can be enhanced, and the distributed development can be easily handled.

According to the invention of claim 4, a simple name can be freely given to the program access name, and a name using Japanese or the like can be given to the dictionary search name.
This makes the data names easier to understand. Furthermore, even if a plurality of areas having the same attributes and meanings are secured, the duplication of names can be avoided.

According to the fifth aspect of the present invention, it is possible to accurately express the aggregate, and to register the meaning content determined only after being incorporated into the aggregate and the check condition regarding the relation between the data items. In addition, the names of the data items and the aggregate name do not overlap. Furthermore, the data dictionary can also have the function of the COPY library.

[Brief description of drawings]

FIG. 1 is a configuration diagram showing a first embodiment (embodiment 1) in which the present invention is applied to a data dictionary managed by a host computer.

FIG. 2 is a diagram showing an example of contents registered in a dictionary source.

FIG. 3 is a layout diagram showing a record format of dictionary loading.

FIG. 4 is a diagram showing an example of contents of dictionary loading.

FIG. 5 is a coding image diagram showing an example of pre-compilation.

FIG. 6 is a configuration diagram showing a second embodiment in which the present invention is applied to a distributed development environment centered on workstations.

FIG. 7 is a diagram showing a representation example of a correspondence relationship between a data item and an aggregate.

[Explanation of symbols]

 10, 20 Subsystem 111 Dictionary source storage file 112 Dictionary load storage file 113 Full screen editor 114 Pre-compilation processing program 115 Program using dictionary 601 Local area network 602 Wide area network 603 Workstation of the person in charge of the dictionary 604 Work of the person in charge of the program Station 605 Transfer information storage file

Claims (5)

[Claims] 1. A method for constructing a data dictionary used in a search, wherein at least an attribute of a data item accessed by a program is registered in an information processing system. Specifies that the attribute of another data item registered in the dictionary should be referred to, and when the specification is made, it is necessary to trace the specified other data item when searching for that data item. A method for constructing a data dictionary characterized by searching various attributes. 2. Having a plurality of independently managed dictionaries,
2. The data dictionary construction method according to claim 1, wherein references to attributes of data items registered in other dictionaries are designated as references to attributes of other data items registered in the dictionary. 3. After registering a data item in a dictionary, the information of the data item is precompiled, and if there is a reference to a data item registered in another dictionary at that time, the information is registered from the other dictionary. The data dictionary construction method according to claim 2, wherein the data dictionary construction is performed. 4. When a data item is accessed by a program, a dictionary used as a dictionary search key and a name (hereinafter referred to as a program access name) used when accessing the data item in the program in the data declaration section of the program. The name registered in the above (hereinafter referred to as a dictionary search name) is also specified, and the procedure section of the program can access the data item using a program access name different from the dictionary search name. The data dictionary construction method according to claim 1, 2 or 3. 5. When a certain data item is an aggregate in which a plurality of other data items (hereinafter referred to as “components”) are arranged, the arrangement order of the data item and the data item are set. 5. The data dictionary configuring method according to claim 4, wherein the program access name and the attribute of each constituent element are registered.