When one wishes to convert the files of an application into a
database, then one only needs to choose the most interesting tables for this database. It
seems expedient to us to migrate the files (personnel file with the accumulation of the
payments, catalogue of the product references, tarifs, etc...). On the other hand, is
doesn't seem suitable to bring in temporary files (files of temps, of interim
calculations, etc...; those files can be generated at every new request).
The first phase will exist in establishing a list of files to be
converted into a database, as a result of which they will be able to be accessed for
query, for concluding synthesis, ratios, statistics, etc... in SQL, RDSQL or 4GL
note : the same COBOL-program can be
simultaneously accessible to classical indexed files, as well as to database tables, seen
as classical files : the files that belong to a database can be accessed by means of COBOL
ACCESS+, the other can be accessed in the usual way.
By means of this list, the utility, delivered with COBOL ACCESS+,
that exploit the file descriptors in the source-language COBOL (the 'FDs' or File
Descriptors) and a file will be automatically generated (DID). This file will enable to
create the database, adapting the structure to the datafiles.
In this phase you don't need to have any knowledge whatsoever of the language of the
chosen relational database. Everything goes automatically !
The user only needs to know the characteristics of the COBOL-files,
in order to describe the different access indexes and cases where the records overlap one
another (REDEFINES COBOL). If necessary, new indexes need to be added to improve the
SQL-access, without having to modify the structure of the SQL-programs.
The product can process every possible way in which data are being
presented (binary, decimal, COMP-1, COMP-2...PIC X, PIC 9, sign field or not, sign leading
separate or sign trailing separate, etc...) and convert these into an adapted description,
typical of the chosen RDBMS.
Based on the DID, generated in a previous phase and after adding
some declaratives (index, etc...), the product COBOL ACCESS+ will be able to create the
database, all by itself, naming every record. In this way, SQL can be used, but
screengenerators like FORMS or reportgenerators (ACE, REPORT, etc...) as well.
note : Whereas INFORMIX is concerned,
there are no preparations requiered in this phase. Only the directories that contain the
tables have to be created; no matter where on the UNIX-system.
However with ORACLE, one needs to create a TABLESPACE in advance, that can receive the
data (eg. manuel administrator ORACLE-base).
The structures are created automatically !
When the existing application needs to be retaken, then it is
indispensable that the basic data from these classical files, can be accessed in the newly
COBOL ACCESS+ does have the necessary utilities.
We guarantee that the old existing application can be retaken !
The product has, of course, every necessary utility to reorganise,
to take back-ups, recoveries, etc...
From the moment a COBOL-program will be executed, every file to be
used, will automatically be placed in the OPEN-mode.
COBOL ACCESS+ will intercept these OPEN-modes in order to trace back
whether the file involved belongs to the database (by comparing to a table in the memory,
without alterating the performance).
When the file doesn't belong to the list, the accessing of the file
will be done in the classical way, and no other access to this file will be intercepted by
COBOL ACCESS+ (performance for access to classical files).
However, when the file complies with a database-table, its accessing
will be done by COBOL ACCESS+, and the status belonging to the program are in accordance
with the standards of COBOL 85 (the file descriptions are made by COBOL ACCESS+, in order
to address every accessing in the future to COBOL ACCESS+ and to be able to read, write,
erase, modify, etc... to the CLOSE-mode).
In short, COBOL ACCESS+ analyses every OPEN-mode of the program. In
this way only the modes involved will be retained and executed. Only the accessorders,
typical of the database-files will continue.
When chosen for INFORMIX SE, the accessfiles will immediately be
made up in C-ISAM-files. The performance is almost the same as the one accessing classical
files; apart form the conversion of data into a format that can be assimilated by SQL
(decimal, signed, etc...).
With other RDBMS, such as ORACLE, COBOL ACCESS+ turns to the SQL-EXEC of the RDBM so that
the READ COBOL can be modified into a SELECT of the RDBM. So the performances are those of