ODF 1.1: 3.3 [Discard earlier submission]

[jose lorenzo <hozelda@yahoo.com>]

This link http://lists.oasis-open.org/archives/office-comment/200807/msg00083.html and this link http://lists.oasis-open.org/archives/office-comment/200807/msg00084.html SHOULD be disregarded.

> from:
> "As there is no semantics specified for such foreign
> content, applications need not process this information
> other than to preserve it when editing the document."
> 
> to:
> "As there is no semantics specified for such foreign
> content, applications thefreedictionary.com/need
> 
> need  (nd)

I apologize.. I was very tired last night as I was wrapping up. There was a mistake in the original post AND in the other nonsensical reply I made immediately after.

The end of the quoted text above is missing several paragraphs (before a last minute cut/paste snafu I didn't catch).

I can't remember the exact details of what I wrote last night, but the intention was as follows. In other words, what follows is the intended bug report.

****************
Propose changing the second sentence of ODF 1.1 section 3.3

from:
"As there is no semantics specified for such foreign content, applications need not process this information other than to preserve it when editing the document."

to:
"As there is no semantics specified for such foreign content, applications need not process such foreign content.

Rationale:

Section 3.3 reads in its entirety:
"In addition to the pre-defined metadata elements, applications should also preserve any additional content found inside the <office:meta> element. As there is no semantics specified for such foreign content, applications need not process this information other than to preserve it when editing the document."

1 -- The first sentence appears to state (see "Further discussion" below) that "applications" do not have to preserve unrecognized content within the <office:meta> element, though it is recommended that they do.

The second sentence contradicts this as it implies that "applications" MUST preserve all the content inside the element but need not do further processing to that content beyond preserving it. [Such is my understanding of the second sentence based on the use of "other than".]

2 -- However, it's also possible that while "should" in the first sentence does not require anything of the "application", the requirement expressed in the second sentence does. "Should" plus the equivalent of "must" used together could be considered consistent and imply a net "must". [You should do it; you must do it.] In this case there would be no error, technically speaking, but it might be clearer if the "should" were changed to "must" since "should" frequently is used to suggest doubt or possibility that something need not be fulfilled. [See def 2 of "should" below as well as the ISO/IEC "should" definition and sections 1.2. and 2.2.1.]

3 -- Finally, "should" can be used to describe a requirement, as shown by definition 1 below for should. In which case, there is no error in section 3.3, but this use of should, though technically allowed because it is NOT in *bold* as discussed in section 1.2, would be used in a different fashion than as prescribed by section 1.2, the ISO/IEC "Alternate keywords" definitions, and section 2.2.1. [see "Further discussion" below.] If "should" was intended to mean "must", then perhaps it would be best to replace it with "must" to avoid confusion (as just mentioned in point 2 above).

Of these three possibilities, which is correct? I am guessing 1 is correct, that the second sentence was worded badly and the intention was that "applications" need not preserve foreign content inside the meta tag, but arguments were just made that case 2 or case 3 could be what was intended instead.

Please clarify.

Further discussion..
(including text from sections 1.2 and 2.2.1 and from ISO/IEC):

http://www.thefreedictionary.com/need

need
n.
1. A condition or situation in which something is required or wanted

http://www.thefreedictionary.com/should

should
aux.v. Past tense of shall
1.  Used to express obligation or duty: You should send her a note.
2.  Used to express probability or expectation: They should arrive at noon.

ODF 1.1 section 1.2 in conjunction with the "Alternate keywords" section of http://docs.oasis-open.org/templates/TCHandbook/ConformanceGuidelines.html  (in lieu of the ISO/IEC directive I was not able to find) specify the meaning of a *bold faced* "should" to mean "to indicate that among several possibilities one is recommended as particularly suitable, without mentioning or excluding others."

But note that "should" is not bold faced in section 3.3 sentence 1, so we are left with a definition that itself either means something like "obligation" (def 1 above) or something like "expectation" (def 2 above). Leading one to wonder, if we are expected to do this, does this mean we are obligated as well?

HOWEVER, section 1.5 separately states:
"conforming applications should preserve meta information" ("should" is not bold faced),

and:
"elements contained within the <office:meta> element may have arbitrary element content and should be preserved (see section 2.2.1)" (should IS bold faced, meaning recommendation or expectation BUT NOT requirement or obligation).

Section 2.2.1 says:
"Custom metadata are arbitrary elements inside <office:meta>. Since their semantics is not defined in this specification, conforming applications in general cannot process or display this data. Applications should preserve this data when editing the document." Here "should" IS bold faced.

Based on 1.2, 1.5, and 2.2.1, my best guess is that the first sentence of 3.3 specifies recommendation not requirement (via dictionary definition 2), and so perhaps a recommendation would be to make the change proposed at the top of this email (or something similar).

Sorry, for the long discussion over a minor issue, but I wanted to pre-empt a back and forth email interchange and possibly even save the readers a bit of time hunting down some of the applicable passages. The issue is actually not that minor for implementers that would prefer to have both a conforming application and as reduced a feature set as possible, and the proposed fix is very small, almost at the level of a typo.

Also, I have not read the whole spec to know what other changes might need to be made.

[Aside: At this point, I have only done a very small amount of light reading, but from what I have seen so far, I can't be certain that the references above to "application" imply "conforming applications".]
****************