Copy a cell that contains the factor, select the range to operate on and hit Edit, paste special. Check the features in the “Operation” frame.

- Select a Range
- Invoke a command “Apply over range”
- Enter a formula =a1*2

And have my range doubled.

“OK, numerical accuracy would be a good start.
64-bit integers
signed and unsigned types
double-double-floats …”

Why not arbitrary length integers and arbitrary precision floats? Sure, they’re slower than IEEE754, but we’ve been using IEEE754 (or a close approximation) since the days of the 8088/8087.

BTW, here are a couple links with other good ideas for Excel:

http://research.microsoft.com/%7Esimonpj/Papers/excel/index.htm

http://citeseer.ist.psu.edu/nunez02extended.html

http://citeseer.ist.psu.edu/dehoon95implementing.html

http://citeseer.ist.psu.edu/dehoon93designing.html

OK, numerical accuracy would be a good start.
64-bit integers
signed and unsigned types
double-double-floats …

Excel is LESS accurate than computers I was working on 25-y ago. Something to do with cross-platform stability.

- Does Excel make enough guarantees about execution order to make this work?

This expression:

=MY_FN(V(SOME_BIG_NASTY_EXPRESSION(A1, A2, A3)), V() + 3)

Depends on the second instance of V being called after the first. If Excel runs it the other way around, the evaluation fails.

- When does Excel/VBA clear the static? Is it cleared at tbe beginning of a recalculation? At the beginning of cell evaluation? Ever? Is this predictable behavior?

- Related to the last question: can I use V in multiple cells? What happens if I do? Does a V() in one cell pull from a V(…) in another cell? That would be a good way to introduce hidden dependancies.

- How does this interact with array formulas?

I like the idea, but it seems pretty limited in the end, and the loosely defined semantics concern me.

And now use in your worksheet:
=MY_FN(V(SOME_BIG_NASTY_EXPRESSION(A1, A2, A3)), V() + 3)

Frank

=MY_FN(SOME_BIG_NASTY_EXPRESSION(A1, A2, A3), SOME_BIG_NASTY_EXPRESSION(A1, A2, A3) + 3)

into this

=LET(i = SOME_BIG_NASTY_EXPRESSION(A1, A2, A3)) { MY_FN(i, i + 3) }

This has the potential for efficienty gains as well as readability gains…

This would be a _huge_ step forward. I’d also like to see:

- Pivot Tables that produce results that easily work with vlookup.

- Faster performance of VBA functions that access Excel data cells.

- A few more cell value types: generalized list and function values would both make Excel far more powerful than it is.

- More sophisticated control over cell formatting. (At the very least, I’d like to be able provide a VBA function to draw the contents of a cell myself.)

- More rows/cells.

- Continue cleaning up the behavior of the existing functions. More numerical reliability is always a good thing.

Generally, if Excel is roughly equivalent to the old HP41 calculator (powerful, extendable, but hitting architectural limits), I’d like to see the Excel analogue to the HP48 series: Take the core ideas of Excel and work to make them fit togther better (ie: vlookup/PivotTables) as well as generalize them to other domains (lists, first class functions, etc.)

Of course, this is expensive to develop, and I doubt there’s enough demand to warrant that kind of investment from Microsoft.

Typing in a single array formula in a column, referencing other columns, is fine. But decide you want to insert some rows you forgot about – at present I have to:
- copy the top cell out of the range,
- delete the whole array formula,

- insert the rows

- copy the cell back,
- delete the copy,
- highlight the new range for the array formula,
- Hit F2
- Press ctrl-shift-return.

Sigh: unless I’m missing something ;o)

Mpemba
PS: 64K columns would be nice too