Eike Rathke wrote:
[2 e-mails joined together]
Hi Leonard,

...
Then, why does COPY => CTRL+A => PASTE copy/paste it *explicitly in every cell* (and crashes Calc by the way), when it could simply change that *default value*?

Because you copied cell content as well and while multiplying that
it exhausted your system's available memory.

Tried with OOo 2.2.1 and 1.3 GB physical RAM + additional swapfile, and got the same crash!

If you instead paste an empty cell or use Paste Special to only
paste formats you get away with much less memory consumption.

Did this with 'Paste special', selected only 'Strings'  => the same crash!

[Admittingly, the standard has NOT a 'default data value',
BUT then the previous explanation does NOT hold either:
 > Yes, if you want row heights / column widths / cell attributes to apply
 > to an entire sheet...

Hu? Press Ctrl+A and with the right mouse button on a row header select
Row Height from the context menu. Why should that be related to the
standard, assuming you're talking of ODF.

[I think there is some confusion here, see next paragraph.]
Basically, you cannot set a default *value* for the cell content. [see later for usefulness]

A. I still feel that NO-one needs exactly 65536 rows x 256 columns.
No, but a fast way to select the entire sheet, whatever size it has, and
apply some attribute or style.
BUT then, Calc could change only the default settings for the master cell (or the 256 cells). NO need to copy the value to every single cell, and 65536 x 256 becomes *irrelevant*. Calc could *fast scan* all cells, to see IF any attributes are explicitly set (IF that would circumvent the master attribute), and reset them. This would be much faster and surely more efficient (and elegant) then setting explicitly every attribute.

B. Why copy everything (attributes, possibly even content) thousand of times?

We don't copy attributes to each cell. Attributes are stored in sets of
items and identical sets in run lengths per column, so when pressing
Ctrl+A and applying some attribute, given that all resulting attribution
is identical, 256 copies of references to a set of items are stored in
memory.

This is different with cell content, which currently is stored
separately for each cell. Certainly an area that could be improved. And
just was for string cells with identical content.
...
A Master-Cell-Content doesn't make much sense, does it?

Yes, it does. It definitely does in multidimensional matrices.

This plain spreadsheet structure (2-dimensional table) is a construct of the '70s (or maybe even older). It should have been replaced a long time ago with new and modern designs. Actually, Sun Microsystems bought in the mid '90s a company designing such a spreadsheet model (IF I remember correctly), BUT somehow I never heard about it anymore.

This feature does constantly make it in my *"TOP 5"* of design flaws/missing features of current spreadsheets. I will write later a more detailed description, but in short: lets say we have a normal 2-dimensional matrix containing our data and we want to compute a derived value from this data.

Instead of writing a formula in *EVERY* cell in a new column, in this multidimensional concept, there is a third dimension, where one writes the formula *a single time* and it will be applied to *ALL cells* in the corresponding column. So, in some way this is the default value I spoke about!

This has the very neat feature that it separates *VALUE* from *LOGIC* (aka the formula from the data values, as they are in different dimensions), too.

I definitely recommend studying more advanced spreadsheet models, They do exist for at least 10 years, and newer designs are really powerful. When I will have some spare time, I will elaborate on my top 5 of spreadsheet issues (this multidimensionality and the data typing/transformation discussed on the OASIS list are just 2 of them).

Sincerely,

Leonard Mada

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to