Re: Please send us this file

[Nick Lamb <njl98r ecs soton ac uk>]

On Fri, Feb 01, 2002 at 01:41:45AM -0500, Jody Goldberg wrote:
> Damn MS to hell.
> ptgAttr type 0 is not documented.  Now we have example of 3
> additional flag values @#%! ^!
> I'll see if that will help crack this nut.

Nothing ever said that these were "flag" values AFAIK, just that the meaning
of the whole byte (word in Excel > 4) is dependent on the control byte that
precedes them, in this case zero. It could equally be a length, a pointer
into some Excel-specific table or anything. There is a very good chance that
it is completely unnecessary for Gnumeric to read it.

> Nice thought, but it does not seem to be it.  I tried your hunch
> and generated a simple sheet with implicit iteration, and none of
> the mystery flags appeared.  If you can simplify the results I'd
> appreciate it.  OpenCalc does not appear to have a clue either.

I have created an extremely simple example available from the same place
as the other Excel temporary files I've made, this has two such ptgAttr
type zero sub-records.

http://www.ecs.soton.ac.uk/~njl98r/excel/

Cells A1 to D1 contain the constant values { 6, 1, 8, 1 }
Cell A3 contains the expression =if(A1:D1>1,">1","not greater")
Cells A4 to D4 contain the array of =if(A1:D1>1,">1","not greater")

This is a simple sheet with implicit iteration. Gnumeric loads it 100%
correctly including the array, despite the unknown ptgAttr contents, but
the results are not correct once re-calculated due to the lack of
implicit iteration in Gnumeric :(

Nick.