[rvba]

It is funny how you talk tjat Excel is bad and the examples just dont work.

Also lack of accuracy in calculating some complicated statistics that "nobody" uses is not really a calculation bug.

Also most of those come from the fact that Excel has a precision of 15 digits.

[harry8]

My fav was when ms announced they'd paid an academic to fix =rand(), ignoring all the others.

Let's see it. Fill the visible part of a sheet with it, conditional format the cells, red for negative etc.

hit f9 to recalc and see a big bunches of cells turn red. Fixed random function to return a random number between 0 and 1.

I haven't worked on gnumeric for a long time - examples came from gnumeric.org

Excel was really bad for everything beyond basic arithmetic, way beyond the important issues with floating poiny listed here https://docs.oracle.com/cd/E19957-01/806-3568/ncg_goldberg.h...

Anyone seen anything about ms actually fixing the bugs? I haven't.

[rvba]

> My fav was when ms announced they'd paid an academic to fix =rand(), ignoring all the others.

I am not sure what that means.

Excel keeps all the bugs for backwards compatibility. So the sheets made years ago still provide same results. In few cases the depreciated some functions -> the old ones still work, but are relatively hidden and the users are encouraged to use the new ones.

They probably should do the same with the statistical functions that supposedly have problems due to rounding. But that cannot be fixed - precision is up to 15 digits.

Also if you wanted an article that talks on Excel precision, you can start with the wikipedia page:

https://en.wikipedia.org/wiki/Numeric_precision_in_Microsoft...

Excel has many problems, but you linking to few website that supposedly prove that "Excel bad" - but at the same time - those exiles simply dont work, is somehow very funny.

[harry8]

it's the web. links die. There's enough there to find the papers if you care. Which you clearly don't.

>Excel keeps all the bugs for backwards compatibility

They /fixed/ rand(). Announcements. Fanfare. The fixed version returned numbers less than zero.

yet again excel's calculating issues go way beyond double precision floating point truncation.

[pillefitz]

First rule of Excel development: Don't use sheet formulas, stick to VBA and apply normal dev practices

[harry8]

what proportion of excel use does this? 0.000000001% or not as much as that in your view?

[pillefitz]

Don't know, but the majority of sheets I've been in contact with (trading desks)