RE: [office-formula] Table:formula attribute and the Formula data type

["Andreas J. Guelzow" <aguelzow@math.concordia.ab.ca>]

On 5/13/2009, "Dennis E. Hamilton" <dennis.hamilton@acm.org> wrote:

>I think there may be some conflicts between other draft content and the ODF 1.2 Part 1 cd01 rev06 language of 1.4.2.4 [D4.4]. 
>
>(section 17.645 in the current draft is not helpful with simply the removal of typically.  Also, there is no definition of a default in the absence of a prefix, which is what I always thought the precursors to 17.645 were going on about, if anything.)
>
>The current (2009-05-08) draft of OpenFormula is not so clear about this in section 1.3, although the Rationale suggesting a preference for an attribute value beginning with "=" and having no leading namespace prefix as the default for OpenFormula.
>
>Section 5.1 of OpenFormula further recommends against using a prefix when there is an OpenFormula formula, making OpenFormula the default and using prefixes when it is not OpenFormula.  Also, the "=" is not optional, or if it is, forced recalculate should perhaps be signified by a "==" leader to avoid ambiguity. (That is, one "=" and no "=" would be the same, but "==" is never to be misunderstood.)
>
>Alternatively, for historical reasons, one might want to leave omission of a prefix to mean "implementation-defined" and have a prefix be mandatory for any OpenFormula table:formula expression in ODF 1.2.  Whether at least one "=" is always required (in the OpenFormula case) is something that needs to be nailed down.

I don't think taht "implementation defined" works in this circumstance
since 2 implementations are involved: Whoever wrote the file and whoever
reads it. With a prefix one can at least tell where the formula comes
from (even if one doesn't understand the formula). Without a prefix
(unless that case defaults to something well defined), the formula is
just a random string since no application can even assume that it is its
own.

Andreas




>I suspect that introduction of a formula data type is not helpful, since there is nothing that can be said about it other than it being a string, *except* when it is identified as being an OpenFormula expression.  In the case of other namespaces and implementation-defined cases, that is something we cannot speak to, it seems to me.
>
>I am not proposing anything in particular, just observing that there needs to be some discussion and tidying up in this area.
>