Release notes for Fast/Unload V4.6: Difference between revisions
No edit summary |
|||
(16 intermediate revisions by 3 users not shown) | |||
Line 1: | Line 1: | ||
This document lists the enhancements and other changes contained in | This document lists the enhancements and other changes contained in | ||
the newest release of <var class="product">Fast/Unload</var>: version 4.6, | the newest release of <var class="product">[[Fast/Unload]]</var>: version 4.6, | ||
which was released in July, 2012. | which was released in July, 2012. | ||
The previous generally released | The previous generally released | ||
Line 733: | Line 733: | ||
the field is referenced: | the field is referenced: | ||
<ol> | <ol> | ||
<li>The <var>IS FIXED</var> and <var>IS FLOAT</var> tests for the missing first occurrence are true if | <li>The <var>IS FIXED</var> and <var>IS FLOAT</var> tests for the missing first occurrence are true if the <var>DEFAULT-VALUE</var> is convertible to a fixed or float representation, respectively. </li> | ||
the <var>DEFAULT-VALUE</var> is convertible to a fixed or float representation, respectively. | |||
<li>When the missing first occurrence is used in a <var>PUT</var> statement that <i>does not</i> | <li>When the missing first occurrence is used in a <var>PUT</var> statement that <i>does not</i> | ||
Line 775: | Line 774: | ||
<li>The <var>IF/ELSEIF</var> statement's <var>MISSING</var> test for the missing first occurrence is true. </li> | <li>The <var>IF/ELSEIF</var> statement's <var>MISSING</var> test for the missing first occurrence is true. </li> | ||
<li>The missing first occurrence may <b>not</b> be used with a <var>DELETE</var> | <li>The missing first occurrence may <b>not</b> be used with a <var>DELETE</var> or <var>UNLOAD</var> statement (<var>DELETEC</var>, and <var>UNLOADC</var>, respectively, <b>can</b> be used), nor may it be used on the left side of a <var>CHANGE</var> statement. | ||
</ul> | </ul> | ||
Line 782: | Line 781: | ||
<div id="nestbl"></div> | <div id="nestbl"></div> | ||
===Nested FOR FIELDGROUP blocks=== | ===Nested FOR FIELDGROUP blocks=== | ||
<!--Caution: <div> above--> | <!--Caution: <div> above--> | ||
Line 1,164: | Line 1,164: | ||
The <var>PAI</var> statement in FUEL supports <var>FILEORG</var> X'100' files, providing | The <var>PAI</var> statement in FUEL supports <var>FILEORG</var> X'100' files, providing | ||
the same information that <code>PAI CTOFIELDS</code> in <var class="product">SOUL</var> provides. | the same information that <code>[[PAI statement#PAI|PAI CTOFIELDS]]</code> in <var class="product">SOUL</var> provides. | ||
There are no operands of the FUEL <var>PAI</var> statement | There are no operands of the FUEL <var>PAI</var> statement to control display | ||
of automatic fields or <var>LOB</var> fields | of automatic fields or <var>LOB</var> fields as there are in the <var class="product">SOUL</var> <var>PAI</var> statement. | ||
Fieldgroups are displayed with a <code><i>fieldgroup name</i> = <i>ID</i></code> line at the start | Fieldgroups are displayed with a <code><i>fieldgroup name</i> = <i>ID</i></code> line at the start and end of the fieldgroup. For example, repeating the example <var>PAI</var> output | ||
and end of the fieldgroup. For example, repeating the example <var>PAI</var> output | |||
shown in [[#str100|Record structure in FILEORG X'100' files, context and other fieldgroup concepts]]: | shown in [[#str100|Record structure in FILEORG X'100' files, context and other fieldgroup concepts]]: | ||
<p class="code"><nowiki>OUTFLD = OUT01 | <p class="code"><nowiki>OUTFLD = OUT01 | ||
Line 1,252: | Line 1,251: | ||
For example: | For example: | ||
<p class="code">%S = 'Don | <p class="code">%S = 'Don''t stop' | ||
%L = #LEN(%S) | %L = #LEN(%S) | ||
PUT %L | PUT %L | ||
Line 1,258: | Line 1,257: | ||
PUT %S | PUT %S | ||
OUTPUT | OUTPUT | ||
%Q = #QUOTE(%S, | %Q = #QUOTE(%S, '''') | ||
%L = #LEN(%Q) | %L = #LEN(%Q) | ||
PUT %L | PUT %L | ||
Line 1,268: | Line 1,267: | ||
The output from the above FUEL fragment is: | The output from the above FUEL fragment is: | ||
<p class="output">10 Don't stop | <p class="output">10 Don't stop | ||
13 'Don | 13 'Don''t stop' | ||
</p> | </p> | ||
Line 1,406: | Line 1,405: | ||
</ul> | </ul> | ||
The <var>MISSING</var> and ERROR sections will be combined, as shown in the next section. | The <var>MISSING</var> and <var>ERROR</var> sections will be combined, as shown in the next section. | ||
<div id="miser"></div> | <div id="miser"></div> | ||
=====MISSING and ERROR clauses===== | =====MISSING and ERROR clauses===== | ||
<!--Caution: <div> above--> | <!--Caution: <div> above--> | ||
Line 1,428: | Line 1,428: | ||
<td>One of the following: | <td>One of the following: | ||
<ul> | <ul> | ||
<li>A constant value, which is placed in the output record if <var class="term">info</var> is | <li>A constant value, which is placed in the output record if <var class="term">info</var> is missing </li> | ||
<li>Either of the keywords <var>SKIP</var> or <var>CANCEL</var>, as described below. </li> | <li>Either of the keywords <var>SKIP</var> or <var>CANCEL</var>, as described below. </li> | ||
Line 1,799: | Line 1,799: | ||
<li>Since the <var>STORE-DEFAULT</var> attribute is only allowed for a field with the <var>DEFAULT-VALUE</var> attribute, it is always displayed for such a field. | <li>Since the <var>STORE-DEFAULT</var> attribute is only allowed for a field with the <var>DEFAULT-VALUE</var> attribute, it is always displayed for such a field. | ||
This may differ from the output of the <var>DISPLAY FIELD</var> command | This may differ from the output of the <var>DISPLAY FIELD</var> command. </li> | ||
< | |||
<li>A <var>CONCATENATED</var> field is simply shown with <code>CAT</code>. The fields that are | <li>A <var>CONCATENATED</var> field is simply shown with <code>CAT</code>. The fields that are | ||
concatenated as the value are not shown. </li> | concatenated as the value are not shown. </li> | ||
<li>A <var>COUNT-OCCURRENCES-OF</var> field is simply shown with <code>CTO</code>. The field that is | <li>A <var>COUNT-OCCURRENCES-OF</var> field is simply shown with <code>CTO</code>. The field that is counted as the value is not shown. </li> | ||
counted as the value is not shown. | |||
</ul> | </ul> | ||
Line 2,011: | Line 2,009: | ||
comparison. Previously, only numeric constants implied numeric comparisons. | comparison. Previously, only numeric constants implied numeric comparisons. | ||
<p> | <p> | ||
This change | This change corrects comparison results that previously | ||
were incorrect, but of course this may change the behavior of some FUEL programs | were incorrect, but of course this may change the behavior of some FUEL programs | ||
(from incorrect to correct). </p> | (from incorrect to correct). </p> | ||
Line 2,032: | Line 2,030: | ||
is no longer allowed.</p> | is no longer allowed.</p> | ||
<p> | <p> | ||
Hence, the following | Hence, the <var>IF</var> and <var>WHEN</var> statements in the following three fragments were all allowed in version 4.4 but are | ||
<i>not</i> allowed in version 4.6: </p> | <i>not</i> allowed in version 4.6: </p> | ||
< | <ol> | ||
WHEN '0' | <li>SELECT #RECIN | ||
FOR I = 1 TO 10 | <p> WHEN '0'</p></li> | ||
<li>FOR I = 1 TO 10 | |||
</ | <p> IF I LT '3'</p></li> | ||
<li>IF SOMEFIELD(#) GE '4'</li> | |||
</ol> | |||
<div id="cmprflf"></div> | <div id="cmprflf"></div> | ||
====Float comparison of FLOAT fields now uses rounded value==== | ====Float comparison of FLOAT fields now uses rounded value==== | ||
<!--Caution: <div> above--> | <!--Caution: <div> above--> | ||
Line 2,114: | Line 2,116: | ||
Previously, when a float constant (for example, <code>123456789.0</code>) or a %variable contains | Previously, when a float constant (for example, <code>123456789.0</code>) or a %variable contains | ||
more than 8 significant digits and is used in a <var>PUT</var> statement, the value is | more than 8 significant digits and is used in a <var>PUT</var> statement, the value is | ||
rounded to 8 significant digits, rather than 15 as it | rounded to 8 significant digits, rather than 15 as it correctly is in versions of <var class="product">Fast/Unload</var> 4.6 or greater. | ||
For example: | For example: | ||
Line 2,136: | Line 2,138: | ||
<div id="back45"></div> | <div id="back45"></div> | ||
===Backwards compatibility with Fast/Unload 4.5=== | ===Backwards compatibility with Fast/Unload 4.5=== | ||
<!--Caution: <div> above--> | <!--Caution: <div> above--> | ||
Line 2,255: | Line 2,258: | ||
<div id="fix44"></div> | <div id="fix44"></div> | ||
===Fixes in | ===Fixes in Fast/Unload 4.6 but not in 4.4=== | ||
<!--Caution: <div> above--> | <!--Caution: <div> above--> | ||
Line 2,363: | Line 2,366: | ||
<div id="coreq"></div> | <div id="coreq"></div> | ||
===Version co-requisites=== | ===Version co-requisites=== | ||
<!--Caution: <div> above--> | <!--Caution: <div> above--> | ||
Line 2,369: | Line 2,373: | ||
(including <var class="product">Fast/Unload</var> itself) that will be imposed by using version 4.6 | (including <var class="product">Fast/Unload</var> itself) that will be imposed by using version 4.6 | ||
of <var class="product">Fast/Unload</var>. | of <var class="product">Fast/Unload</var>. | ||
[[Category:Fast/Unload]] | |||
[[Category:Release notes|Fast/Unload V4.6]] |
Latest revision as of 23:15, 13 February 2017
This document lists the enhancements and other changes contained in the newest release of Fast/Unload: version 4.6, which was released in July, 2012. The previous generally released version of Fast/Unload, 4.4, was released in October, 2007, and an interim version., 4.5, was released in May, 2010.
Overview
Differences between versions 4.5 and 4.6
Much of the information in these Version 4.6 release notes also applies to changes that were made in the interim version 4.5 of Fast/Unload. Changes listed in these release notes that do not apply to version 4.5 are listed in this section.
Version 4.5 does not support the following statements:
In addition, although the FOR FIELDGROUP fgrpName[(occ)] statement is allowed in version 4.5, it has little practical use. In V4.5, it does not establish context for fieldgroup members; it only establishes context for the #FIELDGROUPID special variable.
Other differences between version 4.5 and version 4.6 are:
- Version 4.5 does not support the #FIELDGROUPOCCURRENCE special variable.
- In version 4.5, the UAI output for a record with any EXACTLY-ONE field that is not physically present will contain the default value (either the DEFAULT-VALUE or, if none, the null string). In 4.6, only physically present EXACTLY-ONE occurrences are output for UAI. This is primarily a performance and file size issue, but also, as described in UAI of EXACTLY-ONE fields, there are some UAI/LAI bugs fixed by this change.
- Asterisks in the syntax of the PUT statement (see Asterisk in PUT statement syntax) are not available in 4.5.
- Although the DEFAULT-VALUE of an AT-MOST-ONE field serves as the value for PUT of the first occurrence, if it is missing and there is no MISSING clause, the other handling of missing AT-MOST-ONE DEFAULT-VALUE fields (as specfied in Handling of missing AT-MOST-ONE fields) is not performed in version 4.5.
- The FIELDGROUP keyword is not used in a fieldgroup occurrence count
in version 4.5, in which the following example is the correct syntax for
a fieldgroup named
GRP
:FOR I FROM 1 TO GRP(#)
In version 4.6, the following is the correct syntax:
FOR I FROM 1 TO FIELDGROUP GRP(#)
- The FIELDGROUP keyword is not used in the #IF/#ELSEIF statements
in version 4.5, in which the following example is the correct syntax for
a fieldgroup named
GRP
:#IF GRP DEFINED
In version 4.6, the following is the correct syntax:
#IF FIELDGROUP GRP DEFINED
Finally, in addition to the above functional differences, bugs were fixed in version 4.6 that have not been fixed (that is, via zaps) in version 4.5. These are listed in Fixes in Fast/Unload 4.6 but not in 4.5.
z/OS or z/CMS required
Starting with version 4.6, Fast/Unload must be running under z/OS or z/CMS.
Maintenance and support
These enhancements to Fast/Unload do not affect its intrinsic functionality, but rather affect the way maintenance and support for the product are delivered.
Improved error messages
The following Fast/Unload error messages have been improved:
Improved I/O error message | Error message FUNL0042 now contains the ending page number, if the read is for multiple Table B pages (for example, a track of pages when SBBUF=1 ). In that case, the last page number in the group of pages being read is shown after the first page being read. This may be of assistance in diagnosing the problem in the file. |
Line number on PUT condition error message | Error message FUNL0052 now contains the line number being executed in the FUEL program when a PUT condition (for example, PUT ... ERROR CANCEL ) cancels Fast/Unload. |
Documentation
Contrasting SOUL and FUEL comparisons
In addition to the changes to comparisons in Fast/Unload, there are some differences between SOUL and Fast/Unload comparisons, whose documentation is changed as described here.
FUEL does not imply numeric comparison for FLOAT fields
Given the following setup of a FLOAT field occurrence:
IN SOMEFIL INITIALIZE IN SOMEFIL DEFINE FIELD FLT (FLOAT LEN 8) IN SOMEFIL begin store record FLT = 10 end store end
In SOUL, a comparison involving this field occurrence implicitly uses a numeric (float) comparison:
begin %s is string len 2 %s = 3 frn 0 if FLT < %s then print '10 < 3' else print '10 >= 3' end if end for end
The result of the above SOUL request is 10 >= 3
.
In FUEL, a field occurrence in a comparison does not imply the comparison type:
FOR EACH RECORD %S = 3 IF #RECIN EQ 1 IF FLT < %S PUT '10 < 3' ELSE PUT '10 >= 3' END IF OUTPUT END IF END FOR
Because the default comparison type is string, the result of the above FUEL program is 10 < 3
. To force a float comparison of a field occurrence, use the plus (+) coercion operator:
IF +FLT < %S
Comparisons involving approximately equal float values
In SOUL, two float values are considered equal if they differ by less than a very small amount (.28764219523228867 E-92
).
In Fast/Unload, however, comparison of float values is done using the value rounded to the nearest 15 significant digits decimal number.
As a consequence of
this difference in comparison rules, some values that SOUL considers different are considered equal by Fast/Unload, and some values that SOUL considers equal are
considered different by Fast/Unload.
For example, the following SOUL fragment produces 1/3 ne 1/3 + 0
as the result:
%x = 3 %x = 1/%x %y = %x + 0 if %x eq %y then print '1/3 eq 1/3 + 0' else print '1/3 ne 1/3 + 0' end if
Whereas this FUEL fragment produces 1/3 eq 1/3 + 0
as the result:
%X = 3 %X = 1/%X %Y = %X + 0 IF %X EQ %Y PUT '1/3 eq 1/3 + 0' ELSE PUT '1/3 ne 1/3 + 0' END IF OUTPUT
This example is based on the fact that (in both SOUL and Fast/Unload) adding 0 causes decimal rounding, and it exhibits one case in which:
Fast/Unload considers the values to be equal, while SOUL considers the values to be different.
There are a large number of such values. There are also a large number of values in the obverse case, that is:
Fast/Unload considers the values to be different, while SOUL considers the values to be equal.
Even though there are a large number of such values, they are probably less likely to occur in applications. Typically, the two most direct way to obtain such values are:
- Using Images in SOUL (if they are stored in FLOAT fields, the values can be accessed in Fast/Unload)
- Using FLOD or IFAM to store values into float fields
For example:
IN SOMEFIL INITIALIZE IN SOMEFIL DEFINE FIELD FLT (FLOAT LEN 8) begin image fltIm flt1 is float len 8 str1 is string len 8 at flt1 flt2 is float len 8 str2 is string len 8 at flt2 end image prepare image fltIm %fltIm:str1 = '402000000018C0A5':x %fltIm:str2 = '402000000018C0BC':x store record FLT = %fltIm:flt1 FLT = %fltIm:flt2 then continue printText {~= FLT(1) } printText {~= FLT(2) } if FLT(1) eq FLT(2) then print 'UL considers them equal' else print 'UL considers them different' end if end store end
The above produces the following result:
FLT(1) = 0.125000000022512 FLT(2) = 0.125000000022513 UL considers them equal
But with Fast/Unload:
OPEN SOMEFIL FOR EACH RECORD PUT 'FLT(1) = ' PUT FLT(1) OUTPUT PUT 'FLT(2) = ' PUT FLT(2) OUTPUT IF +FLT(1) EQ FLT(2) THEN PUT 'FUEL considers them equal' ELSE PUT 'FUEL considers them different' END IF OUTPUT END FOR
The above produces the following result:
0.125000000022512 0.125000000022513 FUEL considers them different
Notes:
- The float comparison behavior for %variables has always been as described here. However, prior to version 4.6 of Fast/Unload, float comparison of FLOAT fields used the exact (that is, non-rounded) value of the field (as stated in Float comparison of FLOAT fields now uses rounded value).
- The above FUEL code uses the
+
coercion operator inIF +FLT(1) EQ FLT(2)
to force comparison of the (float) numeric value of the fields. Otherwise, the comparison would be done using the string value of the fields. This is not necessary in SOUL, as explained in FUEL does not imply numeric comparison for FLOAT fields.- See IS FLOAT/FIX disallowed for operands of numeric types for another compatibility issue involving the IF statement.
PUT statement
The changes to the documentation of the PUT statement, described in Revised syntax for the PUT statement, apply to Fast/Unload prior to versions 4.6 and 4.5, except for those aspects that are new, namely:
- The "asterisk placeholder" (
STRING(*)
,MISSING *
, andERROR *
is also an "override" forERROR
). - The treatment of AT-MOST-ONE DEFAULT-VALUE fields.
New Features
All new features in version 4.6 of Fast/Unload are presented in the subsections of this section.
Support for FILEORG X'100' files - summary
This version of Fast/Unload supports files whose FILEORG setting includes the X'100' bit, which were introduced in version 7.2 of Model 204. The various aspects of this support are:
UAI | UAI FUEL programs may be used to unload FILEORG X'100' files, for reloading with Fast/Reload. | ||||||
---|---|---|---|---|---|---|---|
PAI | The PAI statement in FUEL supports FILEORG X'100' files; see PAI. | ||||||
FSTATS | The field statistics produced by the FSTATS option or statement include information about fields and fieldgroups in a FILEORG X'100' file, including the new field attributes. See FSTATS for FILEORG X'100' files. | ||||||
FOR FIELDGROUP blocks | New FUEL statements are provided to establish a fieldgroup context for processing members of the fieldgroup. An overview of these statements is provided in FOR FIELDGROUP blocks. | ||||||
Other references to fieldgroups | In addition to FOR FIELDGROUP blocks, you can refer to fieldgroups by name (always preceded by the FIELDGROUP keyword) in the following FUEL constructs:
| ||||||
Occurrences of EXACTLY-ONE and AT-MOST-ONE fieldgroup members | You may refer to occurrences of non-nested, non-FG *, non-REPEATABLE fields without an enclosing FOR FIELDGROUP block. This facility is explained in References to fieldgroup members not in fieldgroup context. | ||||||
EXACTLY-ONE fields | EXACTLY-ONE fields are new in V7.2. Their handling in FUEL is explained in EXACTLY-ONE fields. | ||||||
Handling of DEFAULT-VALUE fields | This is explained in Handling of DEFAULT-VALUE fields. Also, for PUT of an AT-MOST-ONE DEFAULT-VALUE field which is MISSING, see Handling of missing AT-MOST-ONE fields. | ||||||
Handling of missing AT-MOST-ONE fields | This is explained in Handling of missing AT-MOST-ONE fields. |
There are many other field attributes introduced in FILEORG X'100' files, and they do not have any impact on version 4.6 of Fast/Unload. For example, Fast/Unload does not enforce the DATETIME format restriction in the ADD or CHANGE statement.
Record structure in FILEORG X'100' files, context and other fieldgroup concepts
A record in a Model 204 file with the FILEORG X'100' bit consists of a sequence of field and/or fieldgroup occurrences. These occurrences are called outer occurrences. A fieldgroup occurrence consists of a seqence of field and/or fieldgroup occurrences. These occurrences are called member occurrences. Each fieldgroup occurrence has a numeric ID that is different from all other fieldgroup occurrence IDs in the same record.
A fieldgroup is defined in a file by using the DEFINE FIELDGROUP command.
Note: This is different from the FIELDGROUP attribute, which can be specified on either the DEFINE FIELD or DEFINE FIELDGROUP command. The FIELDGROUP attribute can be abbreviated as FG, as is typically the case in this document.
The DEFINE command for a field or fieldgroup can specify that it is a member of a specific fieldgroup by using the FG attribute with the name of the fieldgroup. When a field or fieldgroup is defined as a member of a specific fieldgroup, that field or fieldgroup can only occur as a member within some occurrence of the fieldgroup named in its FG attribute.
Fields and fieldgroups defined without the FG attribute cannot occur within a fieldgroup, so when they occur in a record, they occur as outer occurrences.
These features are powerful and enable definition of a straightforward approach to repeating field groups, as shown in this example:
DEFINE FIELD OUTFLD DEFINE FIELDGROUP GRP DEFINE FIELD EXOMEM WITH FG GRP DEFINE FIELD EXOMEM2 WITH FG GRP DEFINE FIELD REPMEM WITH REPEATABLE FG GRP
An example PAI of a record in a file with these definitions is:
OUTFLD = OUT01 OUTFLD = OUT02 \GRP = 20 EXOMEM = MEM01 EXOMEM2 = MEM02 /GRP = 20 OUTFLD = OUT03 \GRP = 5 EXOMEM = MEM03 EXOMEM2 = REPMEM = REPMEM01 /GRP = 5 OUTFLD = OUT07
Notes:
- The order of fieldgroup IDs, as in this example (20 and 5), need not correspond to the order of the fieldgroups in the record. The ID of a fieldgroup is assigned when it is added to the record, and is equal to one more than the highest fieldgroup ID which had been used in the record (even if the fieldgroup with that ID has been deleted).
- The default repeatability for (non-FG *) fieldgroup member fields
is EXACTLY-ONE.
In the second occurrence of
GRP
above, whether or notEXOMEM2
has been stored cannot be determined; it is treated in all respects as if it had been stored with a value of the null string.
In order for Fast/Unload to operate on an occurrence of a fieldgroup member, the fieldgroup containing the member must be identified. This is obtained in either of two ways:
- By referring to the member of the fieldgroup within a FOR FIELDGROUP block that identifies an occurrence of the fieldgroup — this fieldgroup occurrence is the fieldgroup context for the fieldgroup member.
- For an AT-MOST-ONE or EXACTLY-ONE non-FG * field member of a non-nested fieldgroup, a reference outside the context of the fieldgroup implicitly identifies the fieldgroup occurrence; see References to fieldgroup members not in fieldgroup context.
The following FUEL program explains both cases of fieldgroup member reference, assuming that the record shown in the above PAI output is the current record:
FOR EACH RECORD FOR FIELDGROUP GRP(1) PUT EXOMEM /* In context of GRP(1): MEM01 END FOR FOR FIELDGROUP GRP(2) PUT EXOMEM /* In context of GRP(2): MEM03 PUT REPMEM /* In context of GRP(2): REPMEM01 END FOR PUT EXOMEM(1) /* Out of context: MEM01 PUT EXOMEM(2) /* Out of context: MEM03 * REPMEM illegal here; it requires a fieldgroup context END FOR
In addition to fieldgroup context:
- Record context is always available for fields and fieldgroups defined without the FG attribute.
Nested fieldgroups and FG *
Two advanced aspects of the fieldgroup feature, not explored in Record structure in FILEORG X'100' files, context and other fieldgroup concepts, are:
- The DEFINE FIELDGROUP command allows the FG attribute, which indicates that an occurrence (or more, if AT-MOST-ONE is not specfied) of the fieldgroup being defined may be contained within an occurrence of the fieldgroup specified in the FG attribute. A fieldgroup occurrence contained within another fieldgroup is called a nested fieldgroup occurrence.
- The FG attribute of the DEFINE FIELD or DEFINE FIELDGROUP
command can specify
FG *
. This indicates that the field or fieldgroup can occur either or both as an outer occurrence or as a member of any occurrence of any fieldgroup in a record. A field or fieldgroup defined with the FG * attribute is called an FG * field or fieldgroup, respectively.
Here is a contrived example illustrating the possibilities using the FG attribute:
DEFINE FIELD OUTFLD DEFINE FIELD STARFLD WITH FG * DEFINE FIELDGROUP STARGRP WITH FG * DEFINE FIELD INSTAR WITH FG STARGRP DEFINE FIELDGROUP GRP DEFINE FIELD INGRP WITH FG GRP DEFINE FIELDGROUP NEST WITH FG GRP
An example PAI of a record in a file with these definitions is:
OUTFLD = OUT01 OUTFLD = OUT02 STARFLD = STAROUT01 STARFLD = STAROUT02 \STARGRP = 15 INSTAR = INSTAR01 STARFLD = STARMEM01 STARFLD = STARMEM02 \STARGRP = 30 INSTAR = STARFLD = STARMEM03 STARFLD = STARMEM04 /STARGRP = 30 /STARGRP = 15 STARFLD = STAROUT03 OUTFLD = OUT03 \GRP = 20 INGRP = MEM01 STARFLD = STARMEM05 \STARGRP = 35 INSTAR = STARFLD = STARMEM06 STARFLD = STARMEM07 /STARGRP = 35 /GRP = 20 \GRP = 5 INGRP = MEM02 /GRP = 5
Since FG * fields and fieldgroups can exist as both outer and member occurrences, a reference to fstar, where fstar is defined with FG *, is to either:
- The occurrence of fstar within the fieldgroup occurrence established by the closest containing FOR FIELDGROUP block.
- The outer occurrence of fstar, if there is no enclosing FOR FIELDGROUP block.
For example, if the current record is the one described by the above PAI output:
FOR EACH RECORD PUT STARFLD /* In record context: STAROUT01 FOR FIELDGROUP GRP /* In record context: ID=20 PUT STARFLD /* In fieldgroup context: STARMEM05 FOR FIELDGROUP STARGRP /* In fieldgroup context: ID=35 PUT STARFLD /* In fieldgroup context: STARMEM06 END FOR END FOR END FOR
See Nested fieldgroups for futher discussion of nested fieldgroups.
FOR FIELDGROUP blocks
Three new statements, all with syntax "FOR ... FIELDGROUP ...", create
FOR FIELDGROUP blocks, which are
terminated by END FOR
.
Within each block, a fieldgroup occurrence (one for each iteration, in
the case of FOR EACH FIELDGROUP) is used as the
context for references to its members.
These statements are described in the following sections:
In addition, within any of the above blocks there is a new statement for terminating execution of the block (LEAVE FIELDGROUP), a new special variable for obtaining the fieldgroup ID of a fieldgroup occurrence (#FIELDGROUPID), and a new special variable for obtaining the occurrence number of a fieldgroup occurrence (#FIELDGROUPOCCURRENCE).
In addition to this new syntax for referencing fieldgroup members, some fieldgroup members can also be referenced outside FOR FIELDGROUP blocks in certain circumstances, as described in References to fieldgroup members not in fieldgroup context.
Statements that cannot reference fieldgroup members
For a field in the context of an occurrence of its fieldgroup (established by a containing FOR FIELDGROUP block), or for a fieldgroup member that can be used outside its fieldgroup context (as described in References to fieldgroup members not in fieldgroup context), that field can be used in any FUEL statement, except in the following:
- The UNLOAD[C] field or NOUNLOAD field statement
- The DELETE[C] field statement
- The sort or hash specification of the SORT or UAI statement
References to fieldgroup members not in fieldgroup context
A fieldgroup member can always be referenced within the context of an occurrence of its fieldgroup established by a FOR FIELDGROUP block, as described in FOR FIELDGROUP blocks (although, as listed in Statements which cannot reference fieldgroup members, certain statements may not reference fieldgroup members).
In addition to the fieldgroup context established by FOR FIELDGROUP blocks, you may make references to certain fieldgroup members without a containing context. A fieldgroup member can only be referenced outside its fieldgroup context if it is EXACTLY-ONE or AT-MOST-ONE, and it is non-FG *, and its containing fieldgroup is not nested.
Stated another way, the following references are only allowed within fieldgroup context for fieldgroup members or, for non-fieldgroup members, within record context:
- REPEATABLE member of fieldgroup
- FG * field
- Field in nested fieldgroup
- Nested fieldgroup
For fieldgroup members that allow it, an "out of context" reference to an occurrence is actually a reference to occurrence number 1 of that field within the specified occurrence of its containing fieldgroup. For example, consider the following field definitions and FUEL program:
DEFINE FIELDGROUP GRP DEFINE FIELD FOO WITH FG GRP ... //FUNIN DD * OPEN ... FOR EACH RECORD PUT FOO(10) OUTPUT END FOR
In the above example, FOO(10)
is a reference to occurrence number 1 of FOO
within occurrence number 10 of fieldgroup GRP
.
Further, an "out of context" reference to the occurrence count of an EXACTLY-ONE (or AT-MOST-ONE, as well) member is actually a reference to the occurrence count of its containing fieldgroup. For example (assuming the same definitions as above), consider the following FUEL fragment:
FOR I FROM 1 TO FOO(#) ... END FOR
In the above fragment, the FOR loop is a executed as many times as
the occurrence count of fieldgroup GRP
.
First occurrence
In many discussions of EXACTLY-ONE and AT-MOST-ONE fields, the term first occurrence is used. This refers to any of the following:
- Occurrence number 1 of an outer EXACTLY-ONE or AT-MOST-ONE field.
- Occurrence number 1 of an EXACTLY-ONE or AT-MOST-ONE fieldgroup member, in the context of the containing fieldgroup.
- Any occurrence n of an EXACTLY-ONE or AT-MOST-ONE fieldgroup member, not in the context of the containing fieldgroup, if occurrence n of the fieldgroup exists.
Hence, given the following definitions within a file:
DEFINE FIELD OUTER WITH AT-MOST-ONE DEFINE FIELDGROUP GRP DEFINE FIELD INNER WITH AT-MOST-ONE FG GRP DEFINE FIELD BOTH WITH AT-MOST-ONE FG *
Then the following FUEL program contains comments illustrating which references
are to the first occurrence (all of the comments are also true if any
AT-MOST-ONE
above is changed to EXACTLY-ONE
):
FOR EACH RECORD PUT OUTER /* This is first occurrence PUT OUTER(3) /* This is NOT first occurrence PUT BOTH /* This is first occurrence PUT BOTH(3) /* This is NOT first occurrence PUT INNER /* This is first occurrence, if fieldgroup /* GRP has at least one occurrence PUT INNER(3) /* This is first occurrence, if fieldgroup /* GRP has at least three occurrences FOR FIELDGROUP GRP PUT BOTH /* This is first occurrence PUT BOTH(3) /* This is NOT first occurrence PUT INNER /* This is first occurrence PUT INNER(3) /* This is NOT first occurrence END FOR END FOR
EXACTLY-ONE fields
One of the new field attributes in Model 204 V7R2 was the EXACTLY-ONE "repeatability" attribute, which designates that there is one and only one occurrence of the field, either within its containing fieldgroup occurrence or within the record:
- An EXACTLY-ONE field that is a fieldgroup member occurs exactly once in every occurrence of that fieldgroup.
- An EXACTLY-ONE field that is not a fieldgroup member (that is, an outer field) occurs exactly once in every record in the file.
EXACTLY-ONE is the default repeatability attribute for a field defined as a member of a (specific) fieldgroup. EXACTLY-ONE is not allowed in combination with FG *, nor is it allowed on the DEFINE FIELDGROUP command. The other repeatability attributes are REPEATABLE, AT-MOST-ONE, and OCCURS. For fieldgroups, fields which are not fieldgroup members, and FG * fields, REPEATABLE is the default repeatability attribute.
Fast/Unload handling of EXACTLY-ONE fields is straightforward:
- The DELETE statement is not allowed with an EXACTLY-ONE field.
- The ADD statement is not allowed with an EXACTLY-ONE field.
The analogous constraint for AT-MOST-ONE fields, which is enforced in SOUL, is not enforced in FUEL. - The first occurrence (see First occurrence) of an EXACTLY-ONE field is never MISSING — that is, on the PUT statement, the SORT clause of the UAI statement, or on the IF and ELSEIF statements — and it always EXISTS on IF and ELSEIF.
- Within its context, the occurrence count of an EXACTLY-ONE field is always one.
- See References to fieldgroup members not in fieldgroup context for a discussion of an EXACTLY-ONE fieldgroup member outside its context.
In the above items, the context is either:
- The context of an occurrence of its fieldgroup, for fieldgroup members.
- Record context, for outer fields.
Most field constraints, all field derivations ignored by ADD, DELETE, CHANGE
As noted in EXACTLY-ONE fields, you cannot use the ADD nor DELETE statement for an EXACTLY-ONE field, just as you cannot in SOUL. That "EXACTLY-ONE fields" section also notes that ADD is freely allowed for an AT-MOST-ONE field, even though the corresponding statement in SOUL might result in a request cancellation.
Except for this constraint on EXACTLY-ONE fields, FUEL does not enforce field constraints (for example, DATETIME, LENGTH-GE, etc.) in the CHANGE and ADD statements, nor do ADD, DELETE, and CHANGE cause a change to a derived (CAT or CTO) field.
Handling of DEFAULT-VALUE fields
The DEFAULT-VALUE attribute, allowed in FILEORG X'100' files, specifies the value of the first occurrence (see First occurrence) of an EXACTLY-ONE or AT-MOST-ONE field if it has not been stored.
For an EXACTLY-ONE field, the result of this in FUEL is that any reference to the value of the first occurrence of the field, if it has not been stored, results in the DEFAULT-VALUE (and, note that there is no way in SOUL nor in FUEL to determine whether or not the field has been stored).
For an AT-MOST-ONE DEFAULT-VALUE field, any reference to the value of the first occurrence of the field, if it has not been stored, also generally results in the DEFAULT-VALUE but, since the field occurrence may also be MISSING, the PUT statement result may be something other than the DEFAULT-VALUE, as explained in Handling of missing AT-MOST-ONE fields.
Handling of missing AT-MOST-ONE fields
Missing field processing is done for:
- The MISSING/EXISTS clauses on the IF/ELSEIF statements
- The PUT statement
- The #ERROR special variable
- Retrieving a value
- The UAI SORT statement
- Some #function arguments, for example, the first argument of #N2DATE
In addition, all of the above processing is performed for a %variable to which a field occurrence has been assigned.
Prior to the introduction of FILEORG X'100' files, the handling of all missing field occurrences was the same. However, when the DEFAULT-VALUE attribute is used with an AT-MOST-ONE field, if the first occurrence (see First occurrence) of the field is missing, it still has a value (the value of the DEFAULT-VALUE attribute).
When an AT-MOST-ONE field has a DEFAULT-VALUE, and that field is missing, then the following take place when the first occurrence of the field is referenced:
- The IS FIXED and IS FLOAT tests for the missing first occurrence are true if the DEFAULT-VALUE is convertible to a fixed or float representation, respectively.
- When the missing first occurrence is used in a PUT statement that does not
contain the MISSING clause other than
MISSING *
:- If the DEFAULT-VALUE is convertible to the PUT AS format
(a DEFAULT-VALUE longer than the length of a
PUT AS STRING
statement is not convertible), then the DEFAULT-VALUE is output. - Otherwise, an ERROR condition occurs (in additon to the MISSING condition). See Revised syntax for the PUT statement for a description of the ERROR clause in the PUT statement.
- If the DEFAULT-VALUE is convertible to the PUT AS format
(a DEFAULT-VALUE longer than the length of a
- After a PUT of the missing first occurrence, #ERROR is 1 or, if the DEFAULT-VALUE is not convertible to the output format as just described, it is 3.
- If the missing first occurrence of the field is used in the SORT clause of the
UAI statement:
- If the MISSING clause is present, the value specified there is used in the sort key.
- Otherwise, the DEFAULT-VALUE is used in the sort key. If the DEFAULT-VALUE is not convertible to the type specified, then a missing first occurrence will terminate Fast/Unload.
- The value used for the missing first occurrence (for example, in a comparison, on the right hand side of an assignment, as a #function argument) is the DEFAULT-VALUE.
The above behavior will also take place for a %variable, if the %variable has been assigned from the missing first occurrence of an AT-MOST-ONE DEFAULT-VALUE field.
Other contexts in which the first occurrence of an AT-MOST-ONE field is missing is not affected by whether the field has the DEFAULT-VALUE attribute:
- The IF/ELSEIF statement's MISSING test for the missing first occurrence is true.
- The missing first occurrence may not be used with a DELETE or UNLOAD statement (DELETEC, and UNLOADC, respectively, can be used), nor may it be used on the left side of a CHANGE statement.
The DEFAULT-VALUE of a field (AT-MOST-ONE or EXACTLY-ONE) has no effect on references to occurrences other than the first.
Nested FOR FIELDGROUP blocks
A FOR FIELDGROUP block may contain other FOR FIELDGROUP blocks, and the context established by a containing block "remains active" within a contained block (unless the fieldgroup specified on the contained block is the same as that specified in a containing block).
For example, given the following definitions:
DEFINE FIELDGROUP PERSON DEFINE FIELD NAME WITH FG PERSON DEFINE FIELD PERSON.SOCSECNUM WITH FG PERSON DEFINE FIELDGROUP BANKACCT DEFINE FIELD BALANCE WITH FG BANKACCT DEFINE FIELD ACCT.SOCSECNUM WITH FG BANKACCT
The following FUEL program aggregates each person's bank balances:
FOR EACH RECORD FOR EACH FIELDGROUP PERSON %BAL = 0 FOR EACH FIELDGROUP BANKACCT IF ACCT.SOCSECNUM EQ PERSON.SOCSECNUM THEN %BAL = %BAL + BALANCE END IF END FOR PUT NAME PUT ' balance=' PUT %BAL OUTPUT END FOR /* EACH FIELDGROUP PERSON END FOR
Notice that IF ACCT.SOCSECNUM EQ PERSON.SOCSECNUM
references
field ACCT.SOCSECNUM
in the inner (BANKACCT
) block and references
field PERSON.SOCSECNUM
in the outer (PERSON
) block.
In the above example, and in many applications, nested FIELDGROUP blocks are not processing nested fieldgroups, but whenever processing nested fieldgroups, nested FOR FIELDGROUP blocks are required, as shown in Nested fieldgroups.
Nested fieldgroups
A nested fieldgroup is a fieldgroup which is itself defined to be a member of another fieldgroup. For example:
DEFINE FIELDGROUP PERSON DEFINE FIELD NAME WITH FG PERSON DEFINE FIELDGROUP MEDICALEXAM WITH FG PERSON DEFINE FIELD EXAM.DATE WITH FG MEDICALEXAM
In the above example, fieldgroup MEDICALEXAM
is nested within, that is, it
is a member of, fieldgroup PERSON
.
To access occurrences of nested fieldgroups, you must use nested fieldgroups. For example:
FOR EACH RECORD FOR EACH FIELDGROUP PERSON PUT NAME %NEED_EXAM = '(no exams)' FOR EACH FIELDGROUP MEDICALEXAM PUT EXAM.DATE AT 30 OUTPUT %NEED_EXAM = '' END FOR IF %NEED_EXAM NE '' THEN PUT %NEED_EXAM AT 30 OUTPUT END FOR END FOR /* EACH FIELDGROUP PERSON END FOR
Support for V7.4
A few changes in V7.4 of Model 204 required changes to Fast/Unload:
- Using the Fast/Unload SOUL Interface with Model 204 V7R4 requires Fast/Unload version 4.6.
- Processing a file with BLOB or CLOB fields which have the DEFAULT-VALUE attribute requires Fast/Unload version 4.6.
- Fast/Unload is not equipped to "fetch", ADD, or CHANGE UTF-8 fields, so such statements are not allowed in FUEL.
If
UTF8FLD
has the UTF8 attribute, only the following types of references to it are allowed:- UTF8FLD(#)
- DELETE UTF8FLD or DELETE UTF8FLD(occur)
- DELETEC UTF8FLD or DELETEC UTF8FLD(occur)
- UNLOAD UTF8FLD or UNLOAD UTF8FLD(occur)
- UNLOADC UTF8FLD or UNLOADC UTF8FLD(occur)
- NOUNLOAD UTF8FLD or NOUNLOAD UTF8FLD(occur)
Also, the PAI statement may not be used if any of the files which are input to Fast/Unload contains a field with the UTF8 attribute.
Version 4.5 does not enforce these UTF8 field restrictions, as noted in Preventing illegal access to UTF8 fields.
New or changed program parameters
The following new parameters may be passed to Fast/Unload:
[NO]ERRCAN | ERRCAN means that CANCEL REPORT is the default for the ERROR clause of the PUT statement. This allows you to make CANCEL REPORT the default just for the current Fast/Unload job, rather than as the default for all Fast/Unload jobs, which is provided by the customization zap for the PUT ERROR clause.
NOERRCAN can be used to override the customization zap for the PUT ERROR clause, if it has been applied, so the normal ERROR clause default is in place. |
---|---|
MISSZ | MISSN1 | MISSZ means that 0 is the default MISSING value for PUT statements with non-STRING formats. This allows you to make this the default just for the current Fast/Unload job, rather than as the default for all Fast/Unload jobs, which is provided by the customization zap for the MISSING default.
MISSN1 can be used to override the customization zap for the default MISSING value, if it has been applied, and the normal MISSING value default is in place. |
New or changed FUEL statements
This section lists new statements that may be used in Fast/Unload, or changes to existing statements.
These statements are all in support of fieldgroups in FILEORG X'100' files.
FOR FIELDGROUP fgrpName[(occ)]
The following block creates a context within which references to fieldgroup members and #FIELDGROUPID use the given fieldgroup occurrence:
FOR FIELDGROUP fgrpName[(occ)] . . . statements that use fgrpName[(occ)] as the . . . fieldgroup context END FOR
The fgrpName clause in the statement can be replaced by the name of any fieldgroup defined in the file.
The (occ) clause in the statement is optional and specifies the occurrence number of the fieldgroup being used as the context. Just as with the occurrence number of a regular field in FUEL, occ may be a positive integer count, a loop control variable, or a %variable. If it is a %variable, the value of the %variable must be a number greater than or equal to 1; any fractional part is dropped.
The given occurrence of the specified fieldgroup, within the current context, is the context for any members of that fieldgroup. If the (occ) clause is omitted, the first occurrence of the fieldgroup is used. If the occurrence of the fieldgroup does not exist within the current context, the entire block is skipped.
For example, assuming the following field definitions:
DEFINE FIELD FIELDA (FG GRP) DEFINE FIELD FIELDB (FG GRP)
The following statements put, on the output stream,
the values of the fields in the second occurrence of fieldgroup GRP
:
FOR FIELDGROUP GRP(2) PUT FIELDA PUT FIELDB OUTPUT END FOR
Note: If
GRP
is not nested, and each ofFIELDA
andFIELDB
is either EXACTLY-ONE, or is AT-MOST-ONE and not FG *, the following statements are equivalent to the above:PUT FIELDA(2) PUT FIELDB(2) OUTPUT
There is no significant difference in processing time between the above two approaches.
FOR FIELDGROUP fgrpName = id
The following block creates a context within which references to fieldgroup members and #FIELDGROUPID use the fieldgroup occurrence whose ID is specified:
FOR FIELDGROUP fgrpName = id . . . statements which use fgrpName with the . . . specified id as the fieldgroup context END FOR
The fgrpName clause in the statement can be replaced by the name of any fieldgroup defined in the file.
The id clause in the statement specifies the fieldgroup ID to be matched for the fieldgroup being used as the context. It may be a positive integer count, a loop control variable, a special variable, or a %variable. If it is a special variable or a %variable, its value must be a number greater than or equal to 1; any fractional part is dropped.
The occurrence of the specified fieldgroup, within the current context, whose fieldgroup ID is equal to the specified id, is the context for any members of that fieldgroup. If an occurrence of the fieldgroup does not exist with the specified id within the current context, the entire block is skipped.
For example, assuming the following field definitions:
DEFINE FIELD FIELDA (FG GRP) DEFINE FIELD FIELDB (FG GRP)
The following statements put on the output stream
the values of the fields in the occurrence of fieldgroup GRP
that has the ID 4
:
FOR FIELDGROUP GRP = 4 PUT FIELDA PUT FIELDB OUTPUT END FOR
Note that if both the fieldgroup ID and fieldgroup occurrence number are
available, it is faster to access a fieldgroup occurrence using FOR FIELDGROUP
fgrpName(occ)
than to access it using FOR FIELDGROUP fgrpName = id
.
FOR EACH FIELDGROUP fgrpName
The following block loops over all occurrences of a fieldgroup in the current context. Within the block, it creates a context within which references to fieldgroup members and #FIELDGROUPID use, in order, each of the occurrences of the fieldgroup within the current context:
FOR EACH FIELDGROUP fgrpName . . . statements that use consecutive occurrences of . . . fgrpName as the fieldgroup context END FOR
The fgrpName clause in the statement can be replaced by the name of any fieldgroup defined in the file.
FUEL's FOR EACH FIELDGROUP statement is identical to the SOUL FOR EACH OCCURRENCE OF FIELDGROUP statement.
Within the current context, each occurrence of the specified fieldgroup is the context for any members of that fieldgroup.
For example, assuming the following field definitions:
DEFINE FIELD FIELDA (FG GRP) DEFINE FIELD FIELDB (FG GRP)
The following statements put on the output stream
the values of the fields in all occurrences of fieldgroup GRP
:
FOR EACH FIELDGROUP GRP PUT FIELDA PUT FIELDB OUTPUT END FOR
Note: If
GRP
is not nested, and each ofFIELDA
andFIELDB
is either EXACTLY-ONE or is AT-MOST-ONE, and is not FG *, the following statements are equivalent to the above (except for the use of the loop control variableI
):FOR I FROM 1 TO FIELDA(#) PUT FIELDA(I) PUT FIELDB(I) OUTPUT END FOR
There is no significant difference in processing time between the above two approaches.
Also note that the following FUEL statements:
FOR EACH FIELDGROUP GRP ... body END FOR
are processed, for all intents and purposes (for example, with the same performance) the same as the following:
FOR I FROM 1 TO FIELDGROUP GRP(#) FOR FIELDGROUP GRP(I) ... body END FOR END FOR
The latter approach may be better in your FUEL program if you
need to access the occurrence number (I
, in this example)
of the fieldgroup, although that is also available in the FOR
EACH FIELDGROUP example above via the #FIELDGROUPOCCURRENCE
special variable (see #FIELDGROUPOCCURRENCE).
LEAVE FIELDGROUP
This statement will terminate execution of the current FOR FIELDGROUP block, causing execution to resume at the statement after the end of its matching END FOR.
[NO]UNLOAD[C] FIELDGROUP
The UNLOAD[C] FIELDGROUP statement allows the unloading of one outer occurrence, or all outer occurrences, of a fieldgroup.
The syntax is:
UNLOAD[C] FIELDGROUP [fldgrpName [qualifier]]
Where:
UNLOAD FIELDGROUP | This statement unloads the specified occurrence(s); if a single occurrence is specified or implied, the occurrence must be present; if it is not present, the Fast/Unload job is cancelled. It can also be used without specifying any fieldgroup name. | ||||||
---|---|---|---|---|---|---|---|
UNLOADC FIELDGROUP | This statement unloads the specified occurrence(s). If a single occurrence is specified or implied, and the occurrence is not present, no action is performed. It can also be used without specifying any fieldgroup name. | ||||||
fldgrpName | The name of the fieldgroup, whose occurrence(s) are to be unloaded. It must either be a non-nested fieldgroup, or it must be a FG * fieldgroup and be referenced outside any FOR FIELDGROUP block. The resulting outer occurrence(s) of the fieldgroup are unloaded.
If fldgrpName is omitted, the current occurrence of the fieldgroup specified on the containing FOR FIELDGROUP block is unloaded. The fieldgroup specified on that block must either be a non-nested fieldgroup, or it must be a FG * fieldgroup and be referenced outside any FOR FIELDGROUP block. | ||||||
qualifier | A specification of which occurence(s) of the fieldgroup named fldgrpName to be unloaded, of one of the following forms:
qualifier is optional. If it is omitted, the first occurrence of the named fieldgroup is unloaded. |
The NOUNLOAD FIELDGROUP statement allows one outer occurrence, or all outer occurrences, of a fieldgroup to be marked to not be unloaded as part of a subsequent blanket UNLOAD statement for the record. The syntax is:
NOUNLOAD FIELDGROUP [fldgrpName [qualifier]]
Where:
fldgrpName | The name of the fieldgroup, whose occurrence(s) are to be marked to not be unloaded. It must either be a non-nested fieldgroup, or it must be a FG * fieldgroup and be referenced outside any FOR FIELDGROUP block. The resulting outer occurrence(s) of the fieldgroup are marked to not be unloaded.
If fldgrpName is omitted, the current occurrence of the fieldgroup specified on the containing FOR FIELDGROUP block is marked to not be unloaded. The fieldgroup specified on that block must either be a non-nested fieldgroup, or it must be a FG * fieldgroup and be referenced outside any FOR FIELDGROUP block. | ||||||
---|---|---|---|---|---|---|---|
qualifier | A specification of which occurence(s) of the fieldgroup named fldgrpName are to be marked to not be unloaded, of one of the following forms:
qualifier is optional; if omitted, the first occurrence of the named fieldgroup is marked. |
PAI
The PAI statement in FUEL supports FILEORG X'100' files, providing
the same information that PAI CTOFIELDS
in SOUL provides.
There are no operands of the FUEL PAI statement to control display
of automatic fields or LOB fields as there are in the SOUL PAI statement.
Fieldgroups are displayed with a fieldgroup name = ID
line at the start and end of the fieldgroup. For example, repeating the example PAI output
shown in Record structure in FILEORG X'100' files, context and other fieldgroup concepts:
OUTFLD = OUT01 OUTFLD = OUT02 \GRP = 20 EXOMEM = MEM01 EXOMEM2 = MEM02 /GRP = 20 OUTFLD = OUT03 \GRP = 5 EXOMEM = MEM03 EXOMEM2 = REPMEM = REPMEM01 /GRP = 5 OUTFLD = OUT07
New or changed #functions and special variables
The following sections describe new or changed #functions or special variables in version 4.6 of Fast/Unload.
#FIELDGROUPID
This special variable returns the fieldgroup ID of the current occurrence of the fieldgroup specified on the containing FOR FIELDGROUP block.
#FIELDGROUPID is not usable in a SORT FIELDS shorthand. These shorthands are discussed in Using SORT FIELDS.
#FIELDGROUPOCCURRENCE
This special variable returns the occurrence number of the current occurrence of the fieldgroup specified on the containing FOR FIELDGROUP block.
#FIELDGROUPOCCURRENCE is not usable in a SORT FIELDS shorthand. These shorthands are discussed in Using SORT FIELDS.
#DELWORD, #WORD, and #WORDS now have delimiter argument
These #functions allow an additional argument to specify the delimiter character that separates words. If present, it must be a string of one character in length.
For example:
%X = 'Friends and Romans. Lend me your ears. I bury Brutus.' %Y = #DELWORD(%X, 2, 1, '.') PUT %Y OUTPUT %Y = #WORD(%X, 1, '.') PUT %Y OUTPUT %Y = #WORDS(%X, '.') PUT %Y OUTPUT
The result of the above fragment is:
Friends and Romans. I bury Brutus. Friends and Romans 3
Note: The delimiter used for #DEBLANK and #FIND is not changing.
#QUOTE: Wrap string in quote character, and double embedded quotes
This function returns the input string with a quote character added before and after the input, and with any quote characters within the input replaced by two copies of the quote character.
%quoted = #QUOTE(string, char)
Syntax Terms
%quoted | The input string wrapped in quotes with embedded quotes doubled. It may be a long string (that is, it may exceed 255 bytes in length). |
---|---|
string | The input string to be quoted. It may be a long string (that is, it may exceed 255 bytes in length). |
char | The quote character. It must be a string of length one. |
For example:
%S = 'Don''t stop' %L = #LEN(%S) PUT %L PUT ' ' PUT %S OUTPUT %Q = #QUOTE(%S, '''') %L = #LEN(%Q) PUT %L PUT ' ' PUT %Q OUTPUT
The output from the above FUEL fragment is:
10 Don't stop 13 'Don''t stop'
#SHADIGEST: SHA-1 digest ("hash") of string
This function returns the 20-byte (always) binary string that is the SHA-1 digest of the argument.
%hashval = #SHADIGEST(string)
Syntax Terms
%hashval | A %variable to receive the SHA-1 digest of the argument string. |
---|---|
string | The input string to be hashed. This may be a long string (that is, it may exceed 255 bytes in length). |
Usage Notes
- SHA (Secure Hash Algorithm) is a set of cryptographic hashing functions, and
#SHADIGEST provides
SHA-1
, the most commonly used. A complete explanation of SHA hashing can easily be found on the internet. - This FUEL #function operates the same as the SOUL SHAdigest method.
Examples
The 20-byte SHA-1 hash of a string is typically expressed as a 40-digit hex value. In the following example, the output string from #SHADIGEST is converted to hex using the #C2X function:
%HSH = #SHADIGEST('this is a simple test') %XHSH = #C2X(%HSH) PUT %XHSH OUTPUT
The result is:
BC38AA2D6769639946806616C14AF0C69477AABE
Other FUEL changes
Reference to the number of occurrences of a fieldgroup
Corresponding to references to the number of occurrences of a field as an entity in a FUEL program, you can also refer to the number of occurrences of a fieldgroup, by coding the FIELDGROUP keyword, followed by the fieldgroup name, followed by the number token (#) in parentheses:
FOR I FROM 1 TO FIELDGROUP PAYMENT.INFO(#)
#IF FIELDGROUP fieldgroup DEFINED
The #IF preprocessor statement has been extended to allow testing for the presence in the file being unloaded (or any of the files in a group being unloaded), of a fieldgroup.
For example:
#IF FIELDGROUP PAYMENT.INFO DEFINED
Asterisk in PUT statement syntax
The syntax for the PUT statement has been enhanced to support the use of the asterisk (*) in the following:
MISSING * or ERROR * | The MISSING and ERROR clauses now allow, in addition to a value or action keyword (immediately following MISSING or ERROR), an asterisk (*), which specifies that the default should be taken for the value or action and that an ERROR condition is exempted from the effect of the ERRCAN parameter so that CANCEL is not performed for the conversion error. It also allows you to specify REPORT or NOREPORT after the asterisk. See Asterisk in MISSING and ERROR clauses for examples. | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
STRING(*[,other format terms]) | Using an asterisk (*) as the length specification for the STRING output format causes the value to be placed in the output using as many bytes as there are in the value; that is, the output length is variable.
The use of STRING with a length of zero (its default) also specifies variable length output, but it has an undesirable effect on the MISSING clause, and so many forms of the STRING format are deprecated.
For a PUT statement without a MISSING constant, the output is the same whether you specify nothing, PUT MIDDLE.NAME AS STRING MISSING '(none)' If field PUT MIDDLE.NAME AS STRING(*) MISSING '(none)' The following formats are deprecated and cause a warning message to be issued. After the deprecated format, the preferred format is shown:
Even though any |
The syntax for the PUT statement is shown in the following two sections.
Revised syntax for the PUT statement
As described above, the asterisk (*) can be used in the MISSING and ERROR clauses. This section and the next include this as part of the syntax description. This revised syntax description replaced the PUT description in PUT.
The revised syntax for the PUT statement is:
[TO destination] - PUT info - [AT loc] - [AS format - [MISSING missActOrVal [repOrNot]] - [ERROR errActOrVal [repOrNot]] ]
The descriptions of the statement parameters was updated:
- A negative loc is allowed and causes "overlay" of previous bytes.
- * is added as a choice for STRING length.
- The default length for STRING is 0, for FLOAT it is 4, and for DECIMAL and ZONED it is 32. Formerly, these were not documented.
- Some language about the MISSING value was changed or removed from the various format options. For example, specifying STRING with length zero (explicitly or by default) to get variable output causes a constant specified in the MISSING clause to be ignored. Therefore * must be used with STRING if a MISSING constant is present.
- The AS format default format is
STRING(*)
. Formerly, this was not documented.
The MISSING and ERROR sections will be combined, as shown in the next section.
MISSING and ERROR clauses
The ERROR clause is allowed with any info except a constant.
The MISSING clause is allowed only if info may be missing, that is:
- A field occurrence
- Either of the special variables #FILENAME or #UPARM (even though #FILENAME cannot be missing)
- A %variable
The clauses may occur in either order. The terms in the MISSING and ERROR clauses are:
missActOrVal | One of the following:
If the MISSING clause is not specified, or if
| ||||||
---|---|---|---|---|---|---|---|
errActOrVal | One of the following:
| ||||||
repOrNot | For both the MISSING and ERROR clauses, you may add a trailing REPORT or NOREPORT keyword. NOREPORT indicates that the condition is not reported on the report data set. This is the default for the MISSING clause unless CANCEL or SKIP is specified as missActOrVal.
REPORT indicates that the condition is reported on the report data set. This is the default (starting with version 4.6) for the ERROR clause, and it is the default for the MISSING clause if CANCEL or SKIP is specified as missActOrVal. The ERROR repOrNot default is distinctly different prior to version 4.6; see Pre-4.5 defaults for the MISSING and ERROR clauses. |
The SKIP and CANCEL keywords are handled as follows:
SKIP | This means that the entire input record is discarded. Note that if output records had been created with an OUTPUT statement before a missing value causes a SKIP, the output records would remain in the output data set. A partial output record that has been created before the SKIP would not go to the output data set.
The SKIP keyword in the MISSING clause causes REPORT to be the default for MISSING. |
---|---|
CANCEL | This means the entire Fast/Unload job is terminated. Use this value if the MISSING or ERROR condition indicates a severe logic error in your data file structure.
The CANCEL keyword in the MISSING clause causes REPORT to be the default for MISSING. |
Starting with version 4.6, the defaults for missActOrVal and errActOrVal are as follows:
missActOrVal default | The default for missActOrVal is as described at the start of this section; see MISSING and ERROR clauses. |
errActOrVal default | The default for errActOrVal is as follows:
As can be seen, there is some asymmetry between the ERROR defaults for STRING versus numeric formats if the ERRCAN parameter is not used:
|
For Fast/Unload versions prior to 4.6, see "Pre-4.5 defaults for the MISSING and ERROR clauses," below, for a description of the defaults for the MISSING and ERROR clauses.
Pre-4.5 defaults for the MISSING and ERROR clauses
This section describes the defaults for the MISSING and ERROR clauses for Fast/Unload versions 4.4 and earlier (version 4.5 is the same as version 4.4, except that it uses the DEFAULT-VALUE, if any, for the missing first occurrence of an AT-MOST-ONE field).
missActOrVal | The default missActOrVal for a STRING format is to fill the output area with blanks. For a numeric format, the default for missActOrVal is to output either -1 or, if the customization zap for the MISSING default is applied, 0.
This is the same as described above for version 4.6, but the above description is simpler, since the DEFAULT-VALUE attribute is not supported prior to version 4.5. |
---|---|
MISSING repOrNot | NOREPORT is the default unless CANCEL or SKIP is specified as missActOrVal, in which case REPORT is the default.
This is exactly as described above for version 4.6. |
ERROR clause | The default for the ERROR clause is as follows:
|
Additional PUT examples
The following sections show some examples of the PUT statement.
Asterisk in MISSING and ERROR clauses
The asterisk (*) in the MISSING clause is used as a placeholder, so you can specify REPORT without overriding the default missing value. This can be particularly useful for fields that have the DEFAULT-VALUE attribute, which PUT uses as the MISSING value.
For example, if field COUNTRY
has AT-MOST-ONE DEFAULT-VALUE 'USA'
,
and the current record does not have an occurrence of COUNTRY
:
PUT COUNTRY AS STRING MISSING * REPORT OUTPUT
The result of the above fragment in this situation is USA
.
The asterisk in the ERROR clause allows you to override the effect of the ERRCAN parameter for a particular PUT statement. For example:
// EXEC PGM=FUNLOAD,PARM='ERRCAN' //FUNIN DD * ... PUT FLDA AS FLOAT(8) ERROR * REPORT PUT FLDB AS FLOAT(8) ...
If FLDA
contains Pizza
, the program is not cancelled by the first PUT
statement. But if FLDB
contains pie
, the program is cancelled.
The REPORT
keyword above is superfluous, since it is the default.
Comparisons in IF/ELSEIF statements
The processing for some kinds of IF/ELSEIF statements is now changed. In addition to these revised rules for comparisons, the Fast/Unload documentation is enhanced as described in Contrasting SOUL and FUEL comparisons.
A comparison is performed between two entities:
[coerc] ent cmp [coerc] ent
Each [coerc] ent phrase is a comparand:
coerc | Optionally, each entity (except a constant, and except where there is a conflict, as described below) can be prefixed by + , forcing a floating point comparison, or by $ , forcing a fixed comparison. If coerc is present in both comparands, both coerc operators must be the same. |
---|---|
ent | The comparison is between two entities. |
cmp | The comparison; for example, LT for the "less than" comparison. |
There are three types of comparison:
float | The "decimal rounded" (to 15 digits) value of the two comparands is compared, using a float comparison instruction. If either comparand cannot be converted, a zero is used in its place. |
---|---|
fixed | The truncated signed-integer value of the two comparands is compared, using a signed-integer comparison instruction. If either comparand cannot be converted, a zero is used in its place. |
string | The string value of the two comparands is compared, using a string comparison instruction. |
Fast/Unload first determines the type of each entity as follows:
- If ent is a constant, its type is the type of the constant, the coerc prefix is not allowed before the constant.
- If ent is the #FIELDGROUPID special variable, its type is
floating point, and neither comparand may have
$
(fixed) as its coerc prefix. - If ent is a loop control variable, a field occurrence count (fldNam(#)), a fieldgroup occurrence count (FIELDGROUP fldgrpNam(#)), or a special variable other than #FIELDGROUPID, #UPARM, or #FILENAME, its type is fixed.
- Otherwise (ent is a field occurrence, a %variable, #UPARM, or #FILENAME), its type is unknown.
Given the type of the entities being compared, the type of comparison is as follows:
- If either ent is prefixed by
+
, a floating-point comparison is performed. In this case, neither ent may be a string constant. - If either ent is prefixed by
$
, a fixed comparison is performed. In this case, neither ent may be a string constant, and the type of neither ent may be floating point. - Otherwise (no coerc present):
- If the type of either ent is floating point, a floating point comparison is performed, and neither ent may be a string constant.
- Otherwise, if the type of either ent is fixed, a fixed comparison is performed, and neither ent may be a string constant.
- Otherwise, a string comparison is performed.
Some of the above rules represent incompatibilities with earlier versions of Fast/Unload. See Only constant entities previously implied comparison type for an explanation of the modifications of the above rules to describe comparisons in earlier versions.
Some of the above rules change the behavior of comparisons; the following additional restrictions do not:
- If both comparands are constants, they must both be the same type.
- If one comparand is a constant, and coerc is specified (on the non-constant comparand), the type of constant must be the same as the type of comparison forced by coerc.
IF/ELSEIF examples
The following sections show some examples of comparisons in the IF statement (they apply equally to ELSEIF).
#RECIN exclude/UPARM
This example illustrates a technique for excluding records whose numbers are in a range that is specified in the Fast/Unload parameters.
// EXEC PGM=FUNLOAD,PARM='UPARM=4-11' ... %LOSKIP = #WORD(#UPARM, 1, '-') %HISKIP = #WORD(#UPARM, 2, '-') FOR EACH RECORD IF #RECIN LT %LOSKIP OR #RECIN GT %HISKIP PUT '#RECIN ' PUT #RECIN PUT ' not in excluded range ' PUT #UPARM OUTPUT END IF END FOR
If the input file has records numbered 0 through 15, the result of the above fragment is:
#RECIN 0 not in excluded range 4-11 #RECIN 1 not in excluded range 4-11 #RECIN 2 not in excluded range 4-11 #RECIN 3 not in excluded range 4-11 #RECIN 12 not in excluded range 4-11 #RECIN 13 not in excluded range 4-11 #RECIN 14 not in excluded range 4-11 #RECIN 15 not in excluded range 4-11
Note that a numeric comparison is performed in IF #RECIN LT %LOSKIP
.
Processing every other field occurrence
A limitation of FUEL is that the counted FOR loop does not have a BY clause. You can easily achieve that functionality, however, using a REPEAT loop and a %variable, and ensuring that the termination test uses a numeric comparison (which is implicit when using a field occurrence count):
%X = 1 REPEAT IF %X GT SOMEFIELD(#) LEAVE REPEAT END IF ... %X = %X + 2 END REPEAT
Loop control variable
A loop control variable is implicitly numeric in IF comparisons. For example:
%LOSKIP = 3 %HISKIP = 5 FOR I FROM 1 TO 10 IF I LT %LOSKIP OR I GT %HISKIP ... END IF END FOR
Statistics improvements
FSTATS for FILEORG X'100' files
Field and fieldgroup statistics are produced for FILEORG X'100' files as follows:
- Occurrence counts for fieldgroup members are shown per fieldgroup, as opposed to per record, for non-fieldgroup members.
- Occurrence counts for EXACTLY-ONE fields refer to physically stored occurrences.
- The length calculated for a fieldgroup is based on the actual length (including
the fieldgroup header) of physical fieldgroup items stored in table B or X.
The length of each fieldgroup item never exceeds 511.
For example, if the total length of the fields physically stored in a fieldgroup is 700, the fieldgroup occurrence will be split into multiple fieldgroup items, and the combined lengths of these items is used as the length of the fieldgroup occurrence in the fieldgroup's length statistics.
- The occurrence count calculated for a fieldgroup is based on "logical" fieldgroup occurrences. For example, if the total length of the fields physically stored in a fieldgroup is 700, even though this is physically stored as multiple fieldgroup items, this is treated as one fieldgroup occurrence.
- To help highlight fieldgroups distinctly from fields,
a fieldgroup will contain three asterisks (
***
) in the field sequence number column. - The new field attributes in FILEORG X'100' files are only displayed for FSTATS AVGTOT.
- Since the default for STORE-NULL differs for EXACTLY-ONE fields, FSTATS for it
may differ from the output of the DISPLAY FIELD command:
- If the STORE-NULL attribute is NONE or ALL, it is displayed (for any field).
- The STORE-NULL attribute is always displayed for an EXACTLY-ONE field.
- Otherwise, STORE-NULL is not displayed (and by inference, if STORE-NULL is allowed,
it is
LIT
).
- Since the STORE-DEFAULT attribute is only allowed for a field with the DEFAULT-VALUE attribute, it is always displayed for such a field. This may differ from the output of the DISPLAY FIELD command.
- A CONCATENATED field is simply shown with
CAT
. The fields that are concatenated as the value are not shown. - A COUNT-OCCURRENCES-OF field is simply shown with
CTO
. The field that is counted as the value is not shown.
Compatibility/Fixes
This section lists any compatibility issues with prior versions of Fast/Unload, in two sections — one covering compatibility with version 4.4, and one covering compatibility with version 4.5. There is also a section listing bugs fixed in this version of Fast/Unload but not, as of the date of this release, fixed in version 4.5 (all 4.6 fixes are also fixed in version 4.4).
In general, backward incompatibility means that an operation that was previously performed without any indication of error, now operates, given the same inputs and conditions, in a different manner. Possibly not listed as backwards incompatibilities are cases in which the previous behaviour, although not indicating an error, was "clearly and obviously" incorrect, and which are introduced as normal bug fixes (whether or not they were fixed with previous maintenance).
Backwards compatibility with Fast/Unload 4.4 and 4.5
This section lists any differences in processing that result from execution with Fast/Unload version 4.6, as compared with the same inputs to Fast/Unload versions 4.4 and 4.5 at current maintenance levels. In some cases zaps have been delivered to change the behavior to be the same as the version 4.6 behavior; these cases are explicitly listed.
See Backwards compatibility with Fast/Unload 4.5 for compatibility issues that exist only between versions 4.6 and 4.5 of Fast/Unload.
Considerations for compatibility issues
In general, a compatibility issue is introduced only if any risk is judged to well-justify the benefit, fixing behavior that is clearly wrong. One area of particular concern may be the changes involving IF/ELSEIF comparisons in Fast/Unload. The description of FUEL comparisons are found in the following sections:
There are also sections dealing with compatibility issues regarding comparisons:
- Only constant entities previously implied comparison type
- Float comparison of FLOAT fields now uses rounded value
- IS FLOAT/FIX disallowed for operands of numeric types
We believe that if this changes the operation of any existing FUEL programs, it is for the better. If you determine that the changes are risky, please contact Technical Support to discuss approaches to reduce the risk.
Detect PUT syntax error immediately after FIXED/DECIMAL/ZONED
Prior to version 4.6 of Fast/Unload, a single-character "garbage" character was allowed and ignored after the FIXED, DECIMAL, or ZONED formats in the PUT statement. For example:
PUT AMOUNT AS FIXED 2 MISSING -999
This invalid syntax is no longer allowed. Detecting this syntax error was part of the fix described in Allow MISSING or ERROR immediately after FIXED/DECIMAL/ZONED in PUT. Presumably, the statement intended in place of the above was:
PUT AMOUNT AS FIXED(2) MISSING -999
Disallow PUT constant>255 AS FIXED(1)
Previously, a FUEL statement was not treated as an error if it specified an integer constant whose value is greater than 255 with a format of FIXED(1). For example:
PUT 256 AS FIXED(1)
This should be reported as an error, because the value (256) does not fit in the length (1 byte) specified.
Any PUT statement that specifies an integer constant whose value is greater than 255, and with a format of FIXED(1), is now reported as an error.
This fix was also delivered as a maintenance zap to versions 4.4 and 4.5 of Fast/Unload.
Handle quotes in MISSING clause of UAI SORT
Previously, the contents of the MISSING clause in a UAI SORT statement were used exactly as is, without stripping quotes (and also with a couple of other problems). Consequently, the behavior prior to version 4.6 was as follows:
- MISSING values that should be reported as syntax
errors, namely, mismatched quotes and values longer than 255 bytes, were not reported in error.
Take this FUEL program, for example:
UAI SORT CHAD MISSING 'X
This should be reported as an invalid statement due to the missing trailing quote, but it is not. The output of this program is actually (because of another glitch) the same as if
MISSING 'X'
were specified. - The MISSING value stored in the sort key of UAI
records contains an incorrect value, so that such
records are not sorted in the proper sequence.
Take this FUEL program for example:
UAI SORT CHAD MISSING 'X'
If a record has a missing occurrence of
CHAD
, and another has an occurrence with the single characterX
, these will not be sorted together. Rather, if there is a record with the single character@
and another with the single character=
, the missing occurrence record will be sorted between them. As a little glitch, the final quote of the MISSING value is dropped if it is the last character in the FUEL statement.
The handing of the MISSING clause of UAI has been fixed, as reflected by the revised start of the description of that clause, as shown in "Revised description of UAI MISSING," next.
Revised description of UAI MISSING
As stated in Handle quotes in MISSING clause of UAI SORT, a fix has been made to the handling of the MISSING clause of the UAI statement. The rest of this section presents the new syntax of that clause, as a replacement for the start of its documentation.
- The MISSING keyword lets you provide a value for the sort
key when the field is missing from the database record or the %variable has
the MISSING value.
mvalue is the string of characters between the MISSING keyword and the following keyword (usually AND) or the end of the line, if there is no additional keyword on the UAI statement.
If there are no quotation marks (actually, apostrophe characters; the terms "apostrophe" and "quote" mean the same thing in this section) in this string of characters, then that string (which can be numeric) is the value to be used for a missing field or %variable (multiple consecutive blanks are collapsed to a single blank).
Quotes are necessary only if the value is to contain a UAI statement keyword which normally terminates the value, or if leading, trailing, or multiple consecutive blanks are required in the value. If there are quotes in the string, then quote normalization is repetitively performed, as follows:
Balanced There must be an even number of quotes. Start quote The first quote is discarded, and encloses a quoted region which begins after that quote and continues until a close quote. Combine doubled Within a quoted region, if a quote is immediately followed by another, then the two quotes are replaced with a single one. Close quote An "undoubled" quote terminates the quoted region, and is discarded. All characters within a quoted region (after undoubling of internal quotes) are appended to the preceding portion of the value.
mvalue must (after quote normalization as above) be less than 256 characters in length.
mvalue must be convertible to the sort key data type.
This fix was also delivered as a maintenance zap to versions 4.4 and 4.5 of Fast/Unload.
#IF errors which are now detected
As mentioned in Fixes to #IF, several problems in the #IF statement have been fixed. Some of these problems allowed erroneous #IF statements to compile without any indication of error. In version 4.6, these errors are now detected:
#IF DEFINED
(as a complete statement) was allowed, and it always operated as if checking for a field that is not found. This statement is no longer allowed.- #ELSE and #ELSEIF were allowed within an #IF block after an #ELSE. These are no longer allowed.
Only constant entities previously implied comparison type
As mentioned in Comparisons in IF/ELSEIF statements, several entities (field/group counts, loop control variables, and most special variables) now imply a numeric comparison. Previously, only numeric constants implied numeric comparisons.
This change corrects comparison results that previously were incorrect, but of course this may change the behavior of some FUEL programs (from incorrect to correct).
For example, assuming that the current record contains 10 occurrences of field FOO:
IF FOO(#) GT 3 PUT 'More than a few ELSE PUT 'Just a few' END IF OUTPUT
In version 4.4., the result of the above is Just a few
, but
in version 4.6., the result is More than a few
.
In addition, it was previously allowed to compare a string constant (with either IF/ELSEIF or WHEN) to an implicit numeric entity, but that is no longer allowed.
Hence, the IF and WHEN statements in the following three fragments were all allowed in version 4.4 but are not allowed in version 4.6:
- SELECT #RECIN
WHEN '0'
- FOR I = 1 TO 10
IF I LT '3'
- IF SOMEFIELD(#) GE '4'
Float comparison of FLOAT fields now uses rounded value
Prior to version 4.6 of Fast/Unload, a float comparison of a FLOAT field used the exact (that is, unrounded) value of the field. Version 4.6 now uses the value of the field rounded to the nearest 15-significant-digit decimal value.
So, for example, given the following single record file:
IN SOMEFIL INITIALIZE IN SOMEFIL DEFINE FIELD FLT (FLOAT LEN 8) IN SOMEFIL begin %x is float %x = 3 %x = 1/%x store record FLT = %x end store end
The following FUEL fragment:
%X = 1/3 + 0 IF +%X EQ FLT PUT '1/3 equal, %var and FLOAT field' ELSE PUT '1/3 different, %var and FLOAT field' END IF OUTPUT
Produces 1/3 equal, %var and FLOAT field
in version 4.6, but
1/3 different, %var and FLOAT field
in prior versions of
Fast/Unload.
Note: This change to rounded float value comparison also applies to the SELECT statement with a FLOAT field and a "contained" WHEN statement with a float constant.
IS FLOAT/FIX disallowed for operands of numeric types
The purpose of the IS FIXED and IS FLOAT tests is to check the format of the contents of a %variable or field occurrence. In version 4.4 these tests were allowed with implicitly numeric entities; however, their behaviour was unpredictable.
Version 4.6 no longer accepts this.
For example:
IF #RECIN IS FIXED IF 1.0 IS FLOAT
In version 4.4, these were allowed but the results were unpredicatble. In version 4.6, these are not allowed.
Note: There were some IS FLOAT/FIXED tests with numeric operands that worked correctly in version 4.4, but, for simplicity, all are now disallowed.
#IF/#ELSEIF now syntax error with nonsense field names
As described in #IF/#ELSEIF allowed nonsense field names, a statement such as the following was allowed in version 4.4 and produced results likely to be unintended:
#IF FOOBAR(*) DEFINED
In version 4.6, this statement results in a syntax error.
Round %var or float constant to 15 digits for ZONED format
Previously, when a float constant (for example, 123456789.0
) or a %variable contains
more than 8 significant digits and is used in a PUT statement, the value is
rounded to 8 significant digits, rather than 15 as it correctly is in versions of Fast/Unload 4.6 or greater.
For example:
PUT 1234567895.0 AS DECIMAL(10) /*Right: 1234567895 OUTPUT %X = 1234567895 PUT %X AS DECIMAL(10) /*Right: 1234567895 OUTPUT PUT 1234567895.0 AS ZONED(10) /*Wrong: 123456790{ OUTPUT /*Should be: 123456789E PUT %X AS ZONED(10) /*Wrong: 123456790{ OUTPUT /*Should be: 123456789E PUT 199999999234.0 AS DECIMAL(12) /*Right: 199999999234 OUTPUT PUT 199999999234.0 AS ZONED(12) /*Wrong: 20000000000{ OUTPUT /*Should be: 19999999923D
This fix was also delivered as a maintenance zap to versions 4.4 (ZAP4426) and 4.5 (ZAP4549) of Fast/Unload.
Backwards compatibility with Fast/Unload 4.5
This section lists any differences in processing that result from execution with Fast/Unload version 4.6, as compared with the same inputs to Fast/Unload version 4.5 at current maintenance levels.
In some cases, a version 4.5 zap was delivered to change the behavior to be the same as the version 4.6 behavior; these cases are explicitly listed.
See also Backwards compatibility with Fast/Unload 4.4 and 4.5 for additional compatibility issues between versions 4.6 and 4.5 of Fast/Unload.
Differences in PUT ONE DV fields with no MISSING constant
As described in Handling of missing AT-MOST-ONE fields, in version 4.6 the PUT statement of the first occurrence of a missing AT-MOST-ONE DEFAULT-VALUE field (or a %variable that has been assigned from the missing first occurrence of such a field), if the PUT statement does not have the MISSING keyword followed by a constant, has these effects:
- Results in the MISSING condition
- Outputs the DEFAULT-VALUE, if convertible
- Also results in the ERROR condition, if non-convertible
In version 4.5, a non-convertible DEFAULT-VALUE does not result in an ERROR condition.
Other compatibility issues with Fast/Unload 4.5
In general, the fixes described in Fixes in Fast/Unload 4.6 but not in 4.5 represent compatibility issues with version 4.5 of Fast/Unload.
Fixes in Fast/Unload 4.6 but not in 4.5
This section lists fixes to functionality existing in Fast/Unload version 4.5 but which, due to the absence of customer problems, have not, as of the date of the release, been fixed in that version.
None of the problems listed in this section pertain to Fast/Unload version 4.4. The fixes in Fast/Unload 4.6 that are not available in version 4.4 are listed in Fixes in Fast/Unload 4.6 but not in 4.4.
UAI of EXACTLY-ONE fields
In Fast/Unload version 4.5, the UAI output for a record with any EXACTLY-ONE field that is not physically present will contain the default value (the DEFAULT-VALUE or, if none, the null string). In 4.6, only physically present EXACTLY-ONE occurrences are output for UAI.
For the most part, the only consequences of this are:
- UAI can run more slowly in 4.5 than in 4.6.
- A subsequent LAI using the UAI output can increase table B/X usage if EXACTLY-ONE fields with DEFAULT-VALUE are STORE-DEFAULT LIT, or EXACTLY-ONE fields without DEFAULT-VALUE are STORE-NULL LIT.
However, it also will cause LAI to fail (with a field constraint violation) if the default value violates a constraint on the field.
PUT of ONE DV xx field AS non-STRING with no MISSING clause
Previously, a PUT statement of an AT-MOST-ONE field is not allowed if the PUT statement uses a non-STRING format, does not contain a MISSING clause, and the DEFAULT-VALUE of the field is not convertible to a number. This is now allowed. If a field occurrence is missing in the above situation, the default non-string MISSING value (that is, -1, or 0 with the MISSZ parameter) is placed in the output.
[NO]UNLOAD of outer EXACTLY-ONE field
In version 4.5 of Fast/Unload, the UNLOAD and NOUNLOAD statements are not performed correctly for outer EXACTLY-ONE fields (in both versions 4.5 and 4.6, they are not allowed for EXACTLY-ONE fieldgroup members). For version 4.5, UNLOAD or NOUNLOAD of an outer EXACTLY-ONE field produces a compilation error message (when maintenance ZAP4538 is applied).
Version 4.6 correctly performs UNLOAD or NOUNLOAD of outer EXACTLY-ONE fields.
References to AT-MOST-ONE fieldgroup members
In version 4.5, references to AT-MOST-ONE fieldgroup members is not supported. That is, version 4.5 does not provide FOR FIELDGROUP blocks for field references, and references to non-nested, non-FG * AT-MOST-ONE fieldgroup members without a FOR FIELDGROUP block (which are supported in 4.6 as described in References to fieldgroup members not in fieldgroup context), result (with ZAP4540) in a compilation error in 4.5.
DEFAULT-VALUE for AT-MOST-ONE field in UAI SORT
In version 4.5, when an AT-MOST-ONE field is used for the SORT key in a UAI SORT unload, the DEFAULT-VALUE of the field is not used as the default for the MISSING clause of the sort specification.
In version 4.6, as described in Handling of missing AT-MOST-ONE fields, the DEFAULT-VALUE is used.
Bug in PAI if fieldgroup present and outer field added
In version 4.5 of Fast/Unload, the following FUEL fragment is not handled correctly:
ADD REP = #RECIN PUT '*' OUTPUT PAI
For example, this may produce a "FUNL0056 Unknown field" error message and cancel Fast/Unload.
Preventing illegal access to UTF8 fields
Fast/Unload is not equipped to "fetch," ADD, or CHANGE UTF8 fields (doing so requires, at the least, conversion between EBCDIC and Unicode). Version 4.6 of Fast/Unload ensures that these unsupported accesses are not allowed by prohibiting the FUEL statements that would give rise to them.
Version 4.5 does not prohibit those statements, so it should not be used with files containing UTF8 fields.
Handling of fieldgroup IDs greater than 2**31 - 1
In version 4.5 of Fast/Unload, if a fieldgroup occurrence has an ID that is greater than 2,147,483,647 the following incorrect results occur:
- The PAI statement displays the ID as a negative number (in particular, value of ID - 4,294,967,296).
PUT #FIELDGROUPID
displays the ID as a negative number (in particular, value of ID - 4,294,967,296).
The above errors are corrected in version 4.6.
Fixes in Fast/Unload 4.6 but not in 4.4
This section lists fixes, delivered in Fast/Unload 4.6, to functionality existing in Fast/Unload version 4.4 but which, due to the absence of customer problems, have not, as of the date of the release, been fixed in that version.
Some of the problems in this section have also been fixed in version 4.5; such cases are indicated.
The Fast/Unload 4.6 fixes to problems introduced in version 4.5 that are not available in version 4.5 are listed in Fixes in Fast/Unload 4.6 but not in 4.5.
Allow MISSING or ERROR immediately after FIXED/DECIMAL/ZONED in PUT
The FIXED, DECIMAL, and ZONED formats of the PUT statement all have default lengths, which allows omission of the parenthesized format. For example, the following is allowed and puts a 4-byte binary integer to the output:
PUT AMOUNT AS FIXED
However, prior to version 4.6, you were not able to specify a MISSING or ERROR clause immediately after the format type. For example:
PUT AMOUNT AS FIXED MISSING -999
This is now allowed.
Note: When this bug was fixed, it also fixed another problem, as described in Detect PUT syntax error immediately after FIXED/DECIMAL/ZONED.
Long string access to FLOAT LEN 4 field
If a FLOAT LEN 4
field is passed as an argument to certain #functions that accept
string values longer than 255 bytes, an error may occur (specifically noted in version
4.4 has been an operation exception at FUNS + X'3870'
).
The following FUEL statement exhibits such a usage:
%X = #CONCAT('xyz', FLOAT4)
Separate extension record stats for each file in group
In version 4.4 of Fast/Unload, the Table B statistics for the files within a group have two incorrect lines:
Maximum extension chain length processed Non-adjacent extension records processed
In version 4.4, they are cumulative for all files in the group. In version 4.6 (and version 4.5 with ZAP4545 applied), they are separated for each of the files within the group.
Fixes to #IF
There are a number of fixes to the #IF statement that are present in version 4.6 of Fast/Unload but not in version 4.4 nor version 4.5:
- The preprocessor statements (
#...
) did not allow the "trailing comments" (/* ...
). - Sometimes the error messages were misleading.
- The parsing of #IF and #ELSEIF did not allow fieldnames that have
the word
DEFINED
orUNDEFINED
in them. There is a workaround for this: using quotes around part of the field name. - An erroneous
#IF DEFINED
(as a complete statement) was allowed, and it always operated as if checking for a field that is not found. - #ELSE and #ELSEIF were allowed within an #IF block after an #ELSE.
Items (4) and (5) above are both noted as incompatibilities; see #IF errors which are now detected.
Properly handle '=' in field name in ADD and CHANGE statements
Previously, a field name containing an equal sign (=) was not allowed
on the left hand side of an ADD or CHANGE statement.
For example, consider a field named WARNOT=PEACE
:
ADD WARNOT'='PEACE = 'Tolstoy'
The above statement is now allowed.
IS FLOAT/FIX unpredictable for operands of numeric types
As desribed in IS FLOAT/FIX disallowed for operands of numeric types, IS FLOAT/FIX is now disallowed for all implictily numeric operands. For example:
IF #RECIN IS FIXED IF 1.0 IS FLOAT
In version 4.4, these were allowed but the results were unpredicatble. In version 4.6, these are not allowed.
#IF/#ELSEIF allowed nonsense field names
In version 4.4, the following statement was allowed:
#IF FOOBAR(*) DEFINED
In version 4.6, such a statement is not allowed. The #IF and #ELSEIF statements only check to see whether a field (or fieldgroup) name is defined; an occurrence should not be specified.
In version 4.4, the above statement checks for a field named FOOBAR(*)
.
In the very bizarre case that you wanted to check for such a field, use #IF FOOBAR'(*)' DEFINED
.
Since such a statement was formerly accepted (although it quite possibly gave
an unintended result), this is mentioned in #IF/#ELSEIF now syntax error with nonsense field names.
Version co-requisites
This section lists any restrictions on usage of various products (including Fast/Unload itself) that will be imposed by using version 4.6 of Fast/Unload.