DR 16-0007 - SML: omissions and inconsistencies in the specification of attributes (Part 3)

Tue Aug 22 18:51:14 CEST 2017

Dear all

Here are proposed resolutions and actions for the final 9 elements in DR
16-0007. At least I’m not having to correct the references to MS-OI29500
this time

Francis

§18.10.1.86 serverFormat (@culture, @format) (complex type: CT_ServerFormat)

MS-OI29500, subsection 2.1.792 (page 352) includes the following note:

a. The standard allows for both culture and format attributes.

Office allows only one of culture and format attributes.

Clearly, this element must have either one or other of the two attributes,
since it is an empty element and has no meaning without one or other
attribute being specified. However, an implementation might choose to allow
both attributes, even though Excel does not.

Proposed resolution: Change the normative text above the attribute
description table by adding two sentences: 

Represents the numeric format specified by the OLAP server for a tuple. At
least one of the attributes culture and format shall be specified. If both
attributes are specified, the result is application-dependent.

[Nit note: I think we prefer “application-dependent” to
“implementation-dependent”. There is one instance of
“implementation-dependent in 29500-1, in the first paragraph of §17.18.24.]

§18.10.1.92 tpl (@fld, @hier) (complex type: CT_Tuple)

MS-OI29500, subsection 2.1.796 (page 361) includes the following note:

a. The standard does not require that @fld is mutually exclusive with @hier
in a SpreadsheetML document.

Office requires that @fld and @heir be mutually exclusive in SpreadsheetML
documents.

The usage of this element is not clear, as the only example only shows a
single tuple, but it would seem to be the case that a tuple may specify
either a pivotField or a pivotHierarchy, but not both. 

Proposed action: Ask the Excel team to confirm that a tuple (OLAP sheet data
entry member) may only specify one or other of these attributes. If
confirmed, change the normative text above the attribute table by adding two
sentences:

Represents an OLAP sheet data entry member. At least one of the attributes
fld and hier shall be specified. If both attributes are specified, the
result is application-dependent.

§18.13.1 connection (@name, @type) (complex type: CT_Connection)

MS-OI29500, subsection 2.1.826 (page 371) includes the following notes:

d. The standard states that the name attribute is optional.

Excel requires this attribute.

j. The standard states that the type attribute is optional.

Excel requires this attribute.

The description of @name in the standard says: “Each connection shall have a
unique name.” This is contradicted by the schema, in which @name is declared
to be optional.

The attribute @type might be omitted in a more limited implementation than
Excel, where only one connection type is supported. However, that would be
an argument for there to be a default value, possibly a value that means
“Custom data connection source”. Given that there is no obvious default
value, it probably makes most sense for this attribute to be required.

[Nit note: Some faux change-tracking seems to have been preserved on page
2000, where values 7 and 8 of @type are underlined, and the description of
value 8 ends DSP.]

Proposed resolution: Change the schema so that @name is required. 

§18.13.3 dbPr (@command) (complex type: CT_DbPr)

There is no note on this attribute in MS-OI29500.

Having reviewed the specification of this element, there is insufficient
information in the standard to determine whether there might be
circumstances in which this attribute may correctly be omitted. Might it be
omitted when the external data connection has been deleted? See notes under
§18.3.13 below.

Proposed action: Either take no action, or ask the Excel team to confirm
whether or not Excel requires @command in all circumstances, and
specifically when the external data connection has been deleted. 

§18.13.5 olapPr (@rowDrillCount) (complex type: CT_OlapPr)

MS-OI29500 contains a note about this attribute (limiting its value – see
subsection 2.1.828, page 372, note a), but says nothing about how to
interpret its omission.

This attribute specifies a maximum number of “drill-through rows” to return.
Its omission appears to imply that there is no explicit maximum number,
although in the case of Excel the maximum number must presumably be no
larger than the maximum value of this attribute. Other implementations might
wish to specify different limitations. The largest maximum number that can
be specified is the largest value of an unsigned integer according to the
W3C unsignedInt datatype, which is 4294967295.

Proposed resolution: No action.

§18.13.12 textPr (@characterSet) (complex type: CT_TextPr) – the attribute
@codePage was moved to Part 4, but is still mentioned here and in §18.2.24
webPublishing.

MS-OI29500, subsection 2.1.833 (page 374) includes the following note:

j. The standard states that characterSet is a valid attribute, and that when
it is present, codePage should be ignored.

Office ignores the characterSet attribute if it is present, and does not
ignore the codePage attribute if characterSet is present.

This note applies to the following products: Office 2010, Office 2010
Server, Office 2010 SP1.

The attribute @codePage is a legacy attribute specified in 29500-4, §15.7.1,
page 218. It is referred to normatively in the description of @characterSet
on page 2011 of 29500-1 – this should not be normative.

Proposed resolution: Change the description of @characterSet as follows: “If
this attribute is not present then the codepage attribute are used, the
behaviour is application-dependent.”

§18.13.13 webPr (@url) (complex type: CT_WebPr)

MS-OI29500, subsection 2.1.834 (page 374) includes the following notes:

b. The standard states that @url may be an empty string.

In Office @url cannot be an empty string, and must have a length less than
or equal to 230 characters.

c. The standard does not state the relationship between the url, xml, and
connection at deleted elements.

In Office, if @xml= true and the @deleted of the parent connection element
is false, a non-empty @url shall be specified. If @xml is false, @url shall
be specified if the @deleted of the parent connection element is false and
not specified otherwise.

These two notes imply that, for Excel at least, @url is required and must be
non-empty except when the external data connection has been deleted, in
which case the parent connection element will specify deleted= “true” and in
this case @url must be omitted.

This series of constraints might not apply to another implementation.

Proposed resolution: Either take no action, or add the following text to the
description of @url: “This attribute is required, unless the attribute
deleted of the parent connection element is specified to be true.”

§18.14.1 cell (@r) (complex type: CT_ExternalCell)

MS-OI29500, subsection 2.1.835 (page 374) includes the following note:

d. The standard implies that the r attribute is optional.

Excel requires this attribute.

The element is used to store cached values from external sources. It is
unclear how this element would be interpreted if @r is omitted. 

Proposed resolution: Change the schema to make this attribute required.

§18.14.5 definedName (@refersTo) (complex type: CT_ExternalDefinedName)

MS-OI29500 contains a note on this attribute (restricting its value – see
subsection 2.1.838, page 375, note b), but does not say how to interpret its
omission.

Proposed action: Ask the Excel team whether Excel requires this attribute.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.vse.cz/pipermail/sc34wg4/attachments/20170822/abf1ef25/attachment-0001.html>