https://bugs.documentfoundation.org/show_bug.cgi?id=153598

--- Comment #15 from jollytall <[email protected]> ---
(In reply to Mark Pawelek from comment #11)
> Re: Jollytall's previous test using RAND(). That is not a sensible test. One
> is supposed to know the expected result of a formula. So any data created by
> RAND() will be pointless.

Sorry, I just now found this earlier comment.
I do not see why it was not sensible. Whatever rand numbers we add, it is
expected that the total we get is the same regardless if we add the numbers of
a table first in rows and then in columns or vice-versa. So the difference of
the two totals must be zero. That is a clear expected result.
If all cells have fixed numbers then doing recalculate many times should not
change the results, so to see the bug we need some changing cells, e.g. time()
or need manual update of cells to new values. Rand() seems the easiest way to
change many cells in one Recalculate cycle.

To the main issue:
Clearly some part of the recalculation was not done when the system had too
much to do. Unlike in Mark's case, it was not needed to have multiple sheets or
data ranges. It was enough to have too many cells to recalculate in one go.
Still, I have not seen this bug again since I upgraded to 7.5.4.2., so for me
the issue is solved.

-- 
You are receiving this mail because:
You are the assignee for the bug.

Reply via email to