BBS: Inland Empire Archive
Date: 05-18-92 (21:21)             Number: 112
From: MATT HART                    Refer#: NONE
  To: BLAINE CARMICHAEL             Recvd: NO  
Subj: Help!                          Conf: (2) Quik_Bas
 BC> I have it set up now, I have to edit the Data Type and recompile. I
 BC> would like to be able to have some sort of data file that would hold the
 BC> new Data Type and Old Data Type and make the changes to the file. This

     There are several ways to do this.  It depends on how you know you need
a new data type.  Is this a change made through the course
of the project (A "Woops! I need another field!") or a user
selected change (Joe, let's add a field in our customer
database program we bought from Blaine).

     In the first case, you simply need to convert existing
'test' or beta data.  This could be done through the use of
a sequential file which keeps information on the field of a
file.  I use a program written to document file record
types.  It has the structure:
Idx  TYPE (1 byte)   LENGTH (integer)  COMMENTS (20 bytes)
 1    A               ??               Alpha-numeric / string
 2    I                2               Integer
 3    L                4               Long integer
 4    S                4               Single precision
 5    D                8               Double precision
 6    C                8               Currency

The Idx numbers are the key to the conversion.  When
inserting a line, the program will automatically assign a
new and unique index to the Idx field.

With this program, I save these structures in a data file
that has a similiar name to the program data file I.E.
Employee file is "EMPL.???" and the information file is
"EMPL.FIL".  Here's a sample of it:
1    A         10        ID
2    A         15        First name
3    A          1        Middle Initial
4    A         15        Last Name
5    A          9        Social Security Number
6    A         25        Address
7    A         15        City
8    A          2        State
9    A         10        Zip Code
10   C          8        Salary
11   S          4        Standard Rate

If I need to change this, and have already entered dozens
of employees as test data, then I copy this EMPL.FIL to
EMPLFIL.OLD and run my program, insert lines and such where
needed.  Then run an auto conversion program:
     ftconvrt emplfil.old empl.fil data\EMPL.*
     ftconvrt {old data file} {new data file} {files to change}

Not only do you always have a record of your data structures (and it all
prints out with "Page x of y" in a nice laser form layout)
but conversion is quick, simple, automatic.

Here's a side note on fielding data for your programs : You
can field within a FOR/NEXT loop, reading the info from a
sequential file, needing nothing more than the length of
each field.


25        (Record length)
5         (number of fields)
10,5,5,4,1  (field lengths)

INPUT #1,Reclen
OPEN "R",2,"FILE.RND",Reclen
INPUT #1,NumFields
REDIM Dat$(1 TO NumFields)
Dum = 0
FOR i = 1 TO NumFields
     INPUT #1,FldLen
     FIELD 2,Dum AS Dummy$, FldLen AS Dat$(i)
     Dum = Dum + FldLen

Now buffer 2 is fielded into Dat$() with the proper lengths.

Okay - next item.  A resizable, field changeable, all-
purposeable file type.  This is best accomplished with a
file header.  The only catch here is that the header may
have some wasted space depending on the record size.  The
program described at the top can create such a header
pretty easily.

Header info:
     Integer - Record Size
     Integer - Number of Fields
4 bytes to start with.
FIELD 1,2 AS Num$
GET 1,1 : RecSize = CVI(Num$)
GET 1,2 : NumFields = CVI(Num$)
REDIM Dat$(1 TO NumFields),DatType$(1 TO NumFields),DatLen(1 TO NumFields)

Field info:
     Integer - Field Length
     2 byte Ascii - Field Type Identifier (A,I,S,L,D,C)

Rec = 3        ' starting location of FIELD info
FOR i = 1 TO NumFields
     GET 1,Rec : Rec = Rec + 1
     DatLen(i) = CVI(Num$)
     GET 1,Rec : Rec = Rec + 1
     DatType$ = RTRIM$(Num$)

You need to figure out where the actual records begin (after the header).

FirstDataByte = Rec * 2       ' First byte past the header info
IF FirstDataByte MOD RecSize = 0 THEN   ' Directly on a record boundary
     FirstRec = FirstDataByte \ RecSize
     FirstRec = FirstDataByte \ RecSize + 1

Dum = 0
OPEN "R",1,"FILE.DAT",RecSize
FOR i = 1 TO NumFields
     FIELD 1,Dum AS Dummy$,DatLen(i) AS Dat$(i)
     Dum = Dum + DatLen(i)

This type of record is great for dBase type applications,
because you can dynamically change the record size and
type, and do conversions on the fly.
 * Origin: Midnight Micro!  V.32/REL  (918)451-3306 (1:170/600)
Outer Court
Echo Basic Postings

Books at Amazon:

Back to BASIC: The History, Corruption, and Future of the Language

Hackers: Heroes of the Computer Revolution (including Tiny BASIC)

Go to: The Story of the Math Majors, Bridge Players, Engineers, Chess Wizards, Scientists and Iconoclasts who were the Hero Programmers of the Software Revolution

The Advent of the Algorithm: The Idea that Rules the World

Moths in the Machine: The Power and Perils of Programming

Mastering Visual Basic .NET