Re: [office] Data Grid Size element proposal

[Warren Turkal <turkal@google.com>]

On Sat, Nov 29, 2008 at 10:15, Andreas J Guelzow
<aguelzow@math.concordia.ab.ca> wrote:
>
>
> On Sat, 2008-11-29 at 01:41 -0800, Warren Turkal wrote:
>> So...I asked a spreadsheet ninja coworker of mine if they had ever
>> used the named range wrapping functionality you showed me for anything
>> use. That person told me that they had never seen it used in the real
>> world. Considering that it only works on named ranges and not ranges
>> given in formulas, I am inclined to say that this feature (max row/col
>> attributes) is still not worth adding unless someone can come up with
>> a real world example of usage that would be enhanced with its
>> addition. I don't like the idea of having some magic value that alters
>> the way the spreadsheet calculates formulas, and I don't think that it
>> would be a valuable use of the TC's time to define all the corner
>> cases that result from that feature. I think it will make things
>> radically more complex for an edge case that rarely (if ever) is
>> encountered.
>
> I don't really care whether you and your coworkers have previously seen
> something. I have just used a default sized Gnumeric to create the
> following spreadsheet:
>
> A1 to E1 contain the integers 1 through 5
> A2 contains "=IV1" and this is copied into B2 to E2.
>
> As a consequence E2 shows the value 4.

I can't replicate this behavior in OO.o 3.0. Here's the what I did in
OO.o 3.0 since it has more columns.

A1..E1 contain 1..5.
AMJ is the last colume.
A2 is set to "=AMJ1" and copied to B2..E2.

A1 = 0 and B1..E1 say "#REF!" and are set to "=REF!".

I believe that OO.o will produce similar results using =IV1 in A1 in
OO.o 2.4.x, but I don't have a copy installed on the computer I am on.

So, I guess my next question is is the behavior defined by the spec.
If so, which (OO.o or Gnumeric or neither) is compliant to the spec?

> Now this spreadsheet exists in the real world (and has nothing to do
> with named ranges.)

So I just found that OO.o doesn't do that, so it's not a global issue.

I personally find OO.o's behavior in this situation much more helpful
than magically wrapping the references.

> How can I save this as an odf file so that a large-size gnumeric can at
> least warn the user that previously wrapped vales may change results.
> Without knowing the grid size of the creating application the opeining
> application can not even find out whether there was wrapping.

Who cares if it never been used for anything useful? I am finding it
hard to sympathize with the Gnumeric problem when I don't see the same
problem in OO.o and when I think the OO.o behavior is more useful.

wt