The different layouts must be uniquely identified IN THE PRIMARY KEY.

The padding character can be specified either by the word SPACE, a single letter, or a standard escape character:

If a single uppercase character is specified it is converted internally to lowercase. If you wish to use an uppercase character as a separator, use the ASCII character code.

Note: The UD8 driver only ever checks the PRIMARY key (key number 1) for uniqueness, not other ‘candidate’ keys.

Similar to the sql ":TABLENAME:create:c:\filename" which allows the actual ‘tablename’ to be redefined, when the path is closed then re-opened, it will be re-opened as the named driver.

To close the path without these effects issue the proc command sql "become UD","$UD8".

Do not send any ‘become’ command to the UD8 driver other than "UD" or the driver specified in the setting in the assignment file. ie: if in the assignment file you specified become SOL, then from proc you can only issue sql "become SOL so7222","$UD8" or sql "become UD none","$UD8". Do not add .DLL to the end of the DLL name.

Even though the entity has ‘become’ assigned to a new path (eg: SOL), you should still refer to it as $UD8 (eg: close "$UD8").

The ‘become’ setting greatly effects the general behavior of the UD8 driver. Most of the default data type mapping and many other features are effectively ‘overwritten’ by the ‘becoming’ driver. This can be catered for in your application model (eg: use Numerics and packing code C). You should always use translatechars binary while testing this option to ensure that it will not produce binary data (eg: ASCII zero).

March Hare strongly suggest exhaustive testing of this feature in the required environment before deployment. Limited testing has been performed on the following drivers:

DB3
SOL

The first time the driver open’s, it will log in as UD8. The second time as the ‘become’ driver, then it will revert back to UD8. You can force it to once again become using the sql "become SOL SOL7222","$UD8" proc command.

Table1 has two (2) fields rat and rathelsdorp. If the field name rat is placed in the file filename then both the field rat and rathelsdorp are converted.

The file must have 1 one only 1 line.

If IGNORELINES is also specified, then the lines containing the string are ignored, otherwise the first line is ignored.

If IGNORELINES is not specified, when records are written to a ‘new’ file, then a line is written at the beginning of the file as a dummy heading.

If HEADING and IGNORELINES are both specified, then no dummy heading is written to the beginning of a ‘new’ file.

C Character Strings
U TRX Format
R Raw
N Number as ASCII String
P Packed decimal
Q Packed decimal sign last
I Signed binary Integer
F Floating Point
G Floating Point (VMS only)
D Date
T Time
E Combined Date Time
B Boolean
M Money
J Number as ASCII String with no Decimal Point
O Zoned Numeric
Y User defined
Z User defined

BOOLFORCE

1:ASCII Boolean (0|1)
2:ASCII Boolean (F|T)
3:ASCII Boolean (N|Y)
4:Binary Boolean (0|1)
5:Binary Boolean (0|-1)

1:ASCII Date dd-mmm-yyyy
2:ASCII Date yyyymmdd
3:ASCII Date ddmmyyyy
4:ASCII Date yymmdd
5:ASCII Date ddmmyy
6:Binary Date yymd
7:Binary Date dmyy
8:Binary Date ymd
9:Binary Date dmy
10:Binary Date yymmdd
11:Binary Date ddmmyy
12:ASCII Date mm/dd/yyyy
13:ASCII Date yyyy-mm-dd

1:ASCII Time hh:nn:ss
2:ASCII Time hhnnss
4:Time hh:nn:ss
5:Time hh:nn:ss.s
6:Time hh:nn:ss.ss

1:SYBASE linear four-byte Date and four-byte Time
2:RMS linear Datetime
3:ASCII Date ddmmyyyy Time hh:nn:ss
4:ASCII Date ddmmyyyy Time hhnnss
5:Ingres Date dd-mmm-yyyy Time hh:nn:ss
6:ORACLE internal Datetime format
7:SYBASE ASCII Date mm/dd/yyyy hh:nn:ss.tt
8:ASCII Datetime yyyymmddhhnnss (like 2)
9:SYBASE binary Small Datetime
10:ASCII Datetime yyyy-mm-dd hh:nn:ss
11:Rdb date yyyy-mm-dd Time hh:nn:ss
13:Rdb date yyyy-mm-dd Time hh:nn:ss.ss

If a single uppercase character is specified it is converted internally to lowercase. If you wish to use an uppercase character as a separator, use the ASCII character code.

The default characters to translate are \n and \r. To use this option you must also set translate.

If BINARY is specified all non-ASCII characters are translated into \xhh and \ is translated to \\. If BINARY is not specified, then all Uniface BINARY types (Raw, and VC* for instance) are mapped to the TRX data type. Note: if the entity (not the field) has an interface definition that maps it to TRX then the BINARY parameter will not effect it (eg: FORMPIC.UFORM.DICT). Using TRANSLATECHARS BINARY uses more space in your CSV file and is slower to process. Use TRANSLATECHARS BINARY only if you need to process binary data outside of Uniface.

Using quoting without using translate can cause unpredictable results.

Do not disable this feature without first carefully checking your data for Special Strings (SS) that contain carriage returns.

If a line ends in the CSV file with a carriage return then an entire record may not have been entirely read if one of the fields has an embedded carriage return.

If you are reading the CSV file outside of the UD8 driver, you also need to take into account cases where the last field is NULL.