[DFDL-WG] Action Item 049: Built-in specification description and schemas
Steve Hanson
smh at uk.ibm.com
Tue Feb 2 03:59:07 CST 2010
Thanks for highlighting this Suman.
The reason for hiving off the properties for text numbers into a separate
named annotation was reuse. It was considered that a given data format
might have a large number of text number fields, but that they could be
described by a far lesser number of annotations, because a limited set of
'number patterns' were used. In Suman's example that's clearly not the
case, but it is an artificial one. We need to consider real world formats.
I've had a look through example COBOL copybooks, and while there is a
large variation in text number fields, reuse of 'number patterns' would be
a benefit. For example, a set of related values might be declared the
same:
15 ORIGINAL-PRICE PIC 9(013)V99.
15 DISCOUNTED-PRICE PIC 9(013)V99.
15 SALE-PRICE PIC 9(013)V99.
15 STAFF-PRICE PIC 9(013)V99.
15 TOTAL-PRICE PIC 9(013)V99.
The question then becomes what is the best way to achieve this reuse. If
you look at a dfdl:textNumberFormat annotation, it is the number pattern
that varies. Everything else would be defined once in a dfdl:format
annotation and scoped. So it does seem overkill to have a
dfdl:textNumberFormat for every number pattern, because the contained
properties can not be scoped and must be redeclared each time.
I suggest the best reuse mechanism for this scenario is the simple type.
In the above example I could declare a PRICE simple type and put the
number pattern on that.
I therefore agree with Suman. Remove dfdl:textNumberFormat and
dfdl:defineTextNumberFormat, add all the properties to dfdl:element and
dfdl:simpleType. In practice most will be set in a dfdl:format and scoped,
only the number pattern will vary per element or simple type.
We should also consider whether the same issue applies to
dfdl:calendarFormat and dfdl:escapeScheme. For both these the reuse
opportunity is high. There is likely to be just one escape scheme per data
format. There is likely to be a small number of calendar formats per data
format (eg, one for a date, one for a time, one for a timestamp). But in
the latter case, it is typically just the calendarPattern that would vary,
the rest of the properties would be set once.
I suggest that whatever we adopt for text numbers we also adopt for
calendars, for consistency.
Regards
Steve Hanson
Programming Model Architect, WebSphere Message Broker,
OGF DFDL WG Co-Chair,
Hursley, UK,
Internet: smh at uk.ibm.com,
Phone (+44)/(0) 1962-815848
From:
Suman Kalia/Toronto/IBM at IBMCA
To:
Alan Powell/UK/IBM at IBMGB, Steve Hanson/UK/IBM at IBMGB, Mike Beckerle
<mbeckerle.dfdl at gmail.com>
Cc:
dfdl-wg at ogf.org
Date:
02/02/2010 00:21
Subject:
Action Item 049: Built-in specification description and schemas
I am trying to create DFDL definition for COBOL copy book and have
experienced a usability issue with TextNumberFormat which have to be named
and referenced from dfdl:element and dfdl:simpleType annotations. Consider
a sample COBOL copy book, attached below, where I have 3 elements having
PIC 9999 display clause (a.k.a zoned decimal) and 2 external (standard)
decimal. They all have same length but the main difference between them is
number is sign which could leading or trailing. As per the V.38 spec, I
would have to create a named textNumberFormat for each of the picture
clause. The key difference in the named textNumberFormats for these
definitions would be numberPattern and rest of the attributes for standard
decimal and zoned decimal are going to be same for a particular platform
or data definition format. The generated DFDL schema will be containing
many occurrences of TextNumberFormat and in the worst case scenario one
for each element defined in the COBOL copy book. This is not very usable
and also user would have to carefully choose the name for these formats so
he can easily identify and distinguish if wants to resue them something
like TextNumberStandardLength5SignLeading etc..
01 CobolTypes.
* External decimal ( Zoned decimal)
05 elem9 PIC 9999 DISPLAY.
05 elem9Signed PIC S9999 DISPLAY.
05 elem9SignedLeading PIC S9999 DISPLAY
SIGN LEADING.
* in DFDL - modeled as standard decimal
05 elem9SignedLeadingSeparate PIC S9999 DISPLAY
SIGN LEADING SEPARATE.
05 elem9SignedTrailingSeparate PIC S9999 DISPLAY
SIGN TRAILING SEPARATE.
Number Format
When textNumberRepresentation is ‘zoned’ only the pattern for positive
numbers is used. Only the following pattern characters may be used: '+' to
indicate whether the leading or trailing digit carries the overpunched
sign, 'V' to indicate the location of an implied decimal point and '0' to
indicate the number of digits (including overpunched). The number is '0'
characters must match the number of digits in the representation otherwise
it is a schema definition error.
Better approach would be
Add numberPattern to dfdl:element and dfdl:simpleType annotation and rest
of the attributes from TextNumberFormat block to either a) dfdl:format
only or (b) both dfdl:format and dfdl:element and dfdl:simpleType.
Let's discuss this in the DFDL workgroup call tomorrow ..
Attached below is a schema coded with the assumption (a) listed above..
<xsd:complexType name="CobolTypes">
<xsd:sequence>
<!---------------- External Decimal
-------------------------------->
<xsd:element name="elem9" dfdl:ref=
"dfdlCobolFmt:CobolZonedDecimalFormat"
dfdl:length="4" dfdl:representation="text"
dfdl:numberPattern="0000">
<xsd:simpleType>
<xsd:restriction base="xsd:short">
<xsd:minInclusive value=
"0" />
<xsd:maxInclusive value=
"9999" />
</xsd:restriction>
</xsd:simpleType>
</xsd:element>
<xsd:element name="elem9Signed" dfdl:ref=
"dfdlCobolFmt:CobolZonedDecimalFormat"
dfdl:length="4" dfdl:representation="text"
dfdl:numberPattern="0000+" >
<xsd:simpleType>
<xsd:restriction base="xsd:short">
<xsd:minInclusive value=
"-9999" />
<xsd:maxInclusive value=
"9999" />
</xsd:restriction>
</xsd:simpleType>
</xsd:element>
<xsd:element name="elem9SignedLeading" dfdl:ref=
"dfdlCobolFmt:CobolZonedDecimalFormat"
dfdl:length="4" dfdl:representation="text"
dfdl:numberPattern="+0000">
<xsd:simpleType>
<xsd:restriction base="xsd:short">
<xsd:minInclusive value=
"-9999" />
<xsd:maxInclusive value=
"9999" />
</xsd:restriction>
</xsd:simpleType>
</xsd:element>
<xsd:element name="elem9SignedLeadingSeparate"
dfdl:ref="dfdlCobolFmt:CobolStandardDecimalFormat"
dfdl:length="5" dfdl:representation="text"
dfdl:numberPattern="+0000;-00000" >
<xsd:simpleType>
<xsd:restriction base="xsd:short">
<xsd:minInclusive value=
"-9999" />
<xsd:maxInclusive value=
"9999" />
</xsd:restriction>
</xsd:simpleType>
</xsd:element>
<xsd:element name="elem9SignedTrailingSeparate"
dfdl:ref="dfdlCobolFmt:CobolStandardDecimalFormat"
dfdl:length="5" dfdl:representation="text"
dfdl:numberPattern="0000+;00000-">
<xsd:simpleType>
<xsd:restriction base="xsd:short">
<xsd:minInclusive value=
"-9999" />
<xsd:maxInclusive value=
"9999" />
</xsd:restriction>
</xsd:simpleType>
</xsd:element>
----- Data format Definitions
<xsd:defineFormat name=
"CobolStandardDecimalFormat">
<xsd:format ref=
"tns:BaseTextNumberStandardDecimal" dfdl:lengthKind="explicit"
dfdl:lengthUnits="bytes"
dfdl:alignment="1" dfdl:alignmentUnits="bytes"
dfdl:leadingSkipBytes="0"
dfdl:trailingSkipBytes="0" />
</xsd:defineFormat>
<xsd:defineFormat name="CobolZonedDecimalFormat">
<xsd:format ref=
"tns:BaseTextNumberZonedDecimal" dfdl:lengthKind="explicit"
dfdl:lengthUnits="bytes"
dfdl:alignment="1" dfdl:alignmentUnits="bytes"
dfdl:leadingSkipBytes="0"
dfdl:trailingSkipBytes="0" />
</xsd:defineFormat>
-- Text number Formats ( added here for reference to identify applicable
attributes for standard and zoned decimal)
<xsd:defineTextNumberFormat name=
"ZonedDecimalNumberFormat">
<xsd:textNumberFormat numberCheckPolicy=
"lax" numberRoundingMode="roundUp"
numberZonedSignStyle=
"asciiStandard" />
</xsd:defineTextNumberFormat>
<xsd:defineTextNumberFormat name=
"StandardDecimalFormat">
<xsd:textNumberFormat
numberGroupingSeparator=","
numberDecimalSeparator="."
numberExponentCharacter="E" numberCheckPolicy="lax"
numberInfinityRep="\u221E"
numberNanRep="\uFFFD" numberRoundingMode="roundUp"
numberZeroRep="" "" />
</xsd:defineTextNumberFormat>
Suman Kalia
IBM Toronto Lab
WMB Toolkit Architect and Development Lead
WebSphere Business Integration Application Connectivity Tools
http://www.ibm.com/developerworks/websphere/zones/businessintegration/wmb.html
Tel : 905-413-3923 T/L 969-3923
Fax : 905-413-4850 T/L 969-4850
Internet ID : kalia at ca.ibm.com
Unless stated otherwise above:
IBM United Kingdom Limited - Registered in England and Wales with number
741598.
Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.ogf.org/pipermail/dfdl-wg/attachments/20100202/99490581/attachment-0001.html
More information about the dfdl-wg
mailing list