OV_INDEXED_SPACE
The Things that Make Up this Example | To Execute this Example | BIF Descriptions |
DLL Required : U_BIF989.DLL
------------
Description : Allows you to define and manipulate an indexed space.
-----------
Availability : Windows -> YES
Unix -> NO
OS/400 -> NO
ARGUMENTS
---------
Type Required/ Length Decimals
Number A/N Optional Description Min Max Min Max
------ --- -------- ----------- ------- -------
1 A Req Type of indexed space operation to 1 50
be performed.
Pass as one of :
CREATE : Create a new indexed space.
INSERT : Unconditionally insert a new entry into an indexed space.
PUT : Insert a new or update an existing entry in an indexed space.
GET : Get an entry from an indexed space.
FIRST : Get the first entry in an indexed space.
NEXT : Get the next entry in an indexed space.
DESTROY : Destroy an indexed space and free associated system resources.
Note that this builtin only validates and acts upon the first character
of the requested index space operation (ie: C,I,P,G,F,N,D) but to
maximise RDML function readability it is recommended that you use the
full words CREATE, INSERT, PUT, GET, FIRST, NEXT and DESTROY.
2 A Req Definition string or indexed space 1 256
identifier (or handle). An identifier
(or handle) is the value returned to you
in argument 2 when you create a new indexed
space that uniquely identifies the indexed
space.
When Arg 1 is Pass this argument as
------------- ---------------------
CREATE The indexed space definition string. See the following
information for details of definition strings.
any other The identifier (or handle) of the indexed space
that is to be used by the requested operation
(the identifier or handle is the value returned
in return value 2 when a new index space is
created. It uniquely identifies the indexed space
which you wish to use).
3 A Opt Definition string continuation. 1 256
Only valid for CREATE operations.
This value is ignored for other operations.
4 A Opt Definition string continuation. 1 256
Only valid for CREATE operations.
This value is ignored for other operations.
5 A Opt Definition string continuation. 1 256
Only valid for CREATE operations.
This value is ignored for other operations.
6 A Opt Definition string continuation. 1 256
Only valid for CREATE operations.
This value is ignored for other operations.
7 A Opt Definition string continuation. 1 256
Only valid for CREATE operations.
This value is ignored for other operations.
8 A Opt Definition string continuation. 1 256
Only valid for CREATE operations.
This value is ignored for other operations.
9 A Opt Definition string continuation. 1 256
Only valid for CREATE operations.
This value is ignored for other operations.
10 A Opt Definition string continuation. 1 256
Only valid for CREATE operations.
This value is ignored for other operations.
11 A Opt Definition string continuation. 1 256
Only valid for CREATE operations.
This value is ignored for other operations.
RETURN VALUES
-------------
Type Required/ Length Decimals
Number A/N Optional Description Min Max Min Max
------ --- -------- ----------- ------- -------
1 A Req Standard Return Code 2 2
Returned as "OK" (completed
normally), "NR" (no record found)
or "ER" (error occurred).
2 A Opt Returned indexed space identifier 10 10
or handle. This return value is
mandatory when argument 1 is passed
as CREATE because it returns the
identifier (or handle) of the indexed
space that was created.
INDEXED SPACES AND INDEXED SPACE DEFINITION STRINGS
---------------------------------------------------
When an index space is being created (ie: CREATE is used in argument 1) then arguments
2 through 11 must specify a definition string for the indexed space.
Arguments 2 through 11 are concatenated (trailing blanks in each separate
argument are ignored) to form a single definition string that must be formatted thus :
name keyword(value), name keyword(value), name keyword(value), ....... name keyword(value)
where :
"name" is the name of a valid RDML field that is defined in, or referenced by, the
RDML function that is creating the list.
"keyword" specifies the use of the field specified in "name". It may be one of KEY,
DATA, AVG, MAX, MIN, COUNT and SUM. If a keyword is not specified then DATA is
assumed as a default keyword.
"value" specifies the name of a valid RDML field that is defined in, or referenced by,
the RDML function defining the list. It is used to specify, for certain keyword values only,
the field upon which the keyword activity should take place. Thus ".., A SUM(B), .." defines
that indexed space field A is to contain the SUM (or total) of field B. Likewise, the
string ".., X AVG(Y), .." defines that indexed space field X is to contain the average
of field Y.
The "keywords" AVG, MAX, MIN and SUM require associated "values" and thus must be formated
as AVG(value), MAX(value), MIN(value) and SUM(value) where value is the name of a field defined
in the invoking RDML function.
The "keywords" KEY, DATA and COUNT do not require associated "values" and they should only be
specified as KEY(), DATA() and COUNT().
An indexed space is best visualised as a table or grid.
The definition string defines what the columns in the indexed space are to be and how they
should be used. The following examples illustrate this concept :
Definition String : deptment key(), deptdes
-----------------
Can be visualised as :
-------------------------------------
| Department | Department Description |
| (DEPTMENT) | (DEPTDESC) |
|------------|------------------------|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
-------------------------------------
where DEPTMENT is the single key to each table or grid entry. Note that "deptdesc"
adopts the default keyword "data()" in this example.
Definition String : deptment key(), section key(), secdesc
-----------------
Can be visualised as :
----------------------------------------------
| Department | Section | Section Description |
| (DEPTMENT) | (SECTION) | (SECDESC) |
|------------|---------------------------------|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
----------------------------------------------
where DEPTMENT and SECTION form an aggregate to each table or grid entry. Note that
"secdesc" adopts the default keyword "data()" in this example.
Definition String : deptment key(), empasal avg(salary), empxsal max(salary), empmsal min(salary)
-----------------
Can be visualised as :
---------------------------------------------------------------
| Department | Average Salary | Maximum Salary | Minimum Salary |
| (DEPTMENT) | (EMPASAL) | (EMPXSAL) | (EMPMSAL) |
|------------|----------------|----------------|----------------|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
---------------------------------------------------------------
where DEPTMENT is the single key to each table or grid entry.
NOTES / RULES / GUIDELINES FOR USAGE OF OV_INDEXED_SPACE
--------------------------------------------------------
* Indexed spaces have been primarily designed to support "batch" style functions that
process large volumes of information. They have been designed to provide :
-> An optimised way of accumulating aggregate multi-level summary information.
-> An optimised way of randomly and repeatedly accessing large lists of information
without the overhead of a database row access (ie: you can load the required
information into an indexed space at the start of your function and then
repeatedly (re)access the indexed space more efficiently than you can by
accessing the DBMS directly).
For examples of these types of usage please refer to the following examples.
* An indexed space must have at least one key field/column defined.
* An indexed space can have at most 20 key fields/columns defined.
* An indexed space must have at least one non-key field/column defined.
* An index space can have at most 100 non-key fields/columns defined.
* The aggregate byte length of all key fields/columns in an indexed space definition
cannot exceed 16K. Note that if you are coming even remotely close to this limit then
you should consult your product vendor about application design.
* The aggregate byte length of all non-key fields/columns in an indexed space definition
entry cannot exceed 16K. Note that if you are coming even remotely close to this limit
then you should consult your product vendor about application design.
* The significance of defined keys decreases with the order of their definition.
Thus the definition string "aaa key(), bbb key(), ccc key(), xxx data(), yyy data()"
defines "aaa" as the most significant key and "ccc" as the least significant key.
* Key definitions can take place at any point in the definition string, but it is customary
to place them at the beginning.
* There is no effective limit to how many entries can be place into an indexed space, but
you must REMEMBER AT ALL TIMES that indexed spaces use allocated system memory. The more
entries that exist in an index space the more overhead you are placing on the consumption
of a system resource. References in this section to "no effective limit" actually mean that
you are effectively constrained by how much memory your system can viably allocate and use.
* The operations FIRST and NEXT support sequential key order access, however they can only be
used with indexed spaces where there are less than 32MB / 6
entries in the indexed space. This is because the storage of the indexed
entry's key always takes 6 bytes (regardless of the actual aggregate length of the key fields).
If you try to use FIRST/NEXT sequential processing on an indexed space that is too large
to support it you will receive a specific error message and your application will be aborted.
GET, PUT and INSERT operations are not subjected to this limitation and can be used with
no effective limit on the number of entries in the indexed space.
* An indexed space can only be used by the function that creates it. Although you can easily
pass the identifier (or handle) of an indexed space to another function, any attempt
by the other function to access the index space may lead to application failure and or
unpredictable results. Do not attempt to do this. It will not work because when you CREATE
an indexed space, a unique access plan to data fields stored in the creating function is
formed. This access plan is unique to the creating function and cannot be effectively used
by any other RDML function.
* You should use the DESTROY operation in your functions. However, all indexed spaces created
by an RDML function are automatically destroyed when it terminates.
* If you are using multiple definition strings with a CREATE operation please remember that
trailing blanks are ignored. Therefore :
USE OV_INDEXED_SPACE (CREATE 'A KEY(), B KEY(), C ' 'KEY(), D DATA(), E DATA()')
will cause a run time syntax error because the strings will be concatenated to form the
definition string :
A KEY(), B KEY(), CKEY(), D DATA(), E DATA()
* Indexed space key fields are treated (and sorted) as pure binary data. The indexed space
has no sense of the definition of a field as alpha, packed, signed, DBCS, etc and will not
account for this in any operation.
* The PUT and INSERT operations are significantly different. A PUT operation checks whether
an entry in the index exists with the specified key already. If it does, it is aggregated and
updated as appropriate. If it does not exist, then a new entry is created. An INSERT operation
does not check. It unconditionally creates a new entry.
This means that if you wish to fill an indexed space from a DBMS table when you know you
are creating unique entries, then using INSERT is substantially more efficient than using PUT.
* You can put duplicate keys into an indexed space (ie: 2 or more entries that have exactly
the same key values) by using the INSERT operation. However, the effects of doing this and/or
the order in which the duplicates are processed is unspecified and may vary from platform
to platform.
* The aggregation operations AVG(), MAX(), MIN(), SUM() and COUNT() are all performed with
a maximum of 15 significant digits of precision.
* You must not use arrays or array indicies in indexed spaces.