My comments on DR 10-0001

[Thorsten Behrens]

Jesper Lund Stocholm wrote:
> 1. There is no such thing as a serial date in Part 1. I know that some
> interprets the last paragraph in Part 1, section 18.17.4 as allowing
> serial dates, but this paragraph (in my reading) deals exclusively with
> guidance to strict consumers on how to interpret non-conforming
> spreadsheets with serial dates.
> 
> 2. Dates are persisted in strict documents (spreadsheets) as ISO-dates.
> Not as serial dates but as ISO-dates! This effectively means that the
> leap-year bug does not exist in S documents, since the date 1900-02-29
> cannot be persisted in ISO-8601 format.
> 
Hi Jesper, Chris,

not exactly sure if you guys are talking past each other here -

> > Because dates are
> > treated as numbers at runtime by spreadsheet applications, it's
> > impossible to tell whether a given cell truly contains a date.
> 
> No - in strict documents dates are not serial dates - they are persisted
> in ISO-8601 format. There is no issue wrt figuring out if the cell
> contains a date or not. The only time this would pose a difficulty would
> be if someone created an strict spreadsheet with a serial date in it and
> styling it using "the old transitional way". But that document would be
> non-conforming since it would be in violation with the constraint that
> 
> "Each unique instant in SpreadsheetML time is stored as an ISO
> 8601-formatted string"
> 
I think here lies the disconnect - from an implementer's
perspective, a cell content is only numbers. As soon as you leave
the serialized xml representation. So my interpretation of the
paragraph you quote is - "it's not a SML time anymore, at the very
least once you've used =SUM() on it". ;)

I thought we had discussed the idea of type-safe spreadsheet engines
in Stockholm, and relegated that to the (far) future?

FWIW, all relevant normative texts around runtime calculations are
using serial dates, I think.

Cheers,

-- Thorsten
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: not available
URL: <http://mailman.vse.cz/pipermail/sc34wg4/attachments/20100331/841f70ce/attachment.pgp>